This action might not be possible to undo. Are you sure you want to continue?
Lecture Notes on Optimization
Pravin Varaiya
ii
Contents
1 INTRODUCTION 1
2 OPTIMIZATION OVER AN OPEN SET 7
3 Optimization with equality constraints 15
4 Linear Programming 27
5 Nonlinear Programming 49
6 Discretetime optimal control 75
7 Continuoustime linear optimal control 83
8 Coninuoustime optimal control 95
9 Dynamic programing 121
iii
iv CONTENTS
PREFACE to this edition
Notes on Optimization was published in 1971 as part of the Van Nostrand Reinhold Notes on Sys
tem Sciences, edited by George L. Turin. Our aim was to publish short, accessible treatments of
graduatelevel material in inexpensive books (the price of a book in the series was about ﬁve dol
lars). The effort was successful for several years. Van Nostrand Reinhold was then purchased by a
conglomerate which cancelled Notes on System Sciences because it was not sufﬁciently proﬁtable.
Books have since become expensive. However, the World Wide Web has again made it possible to
publish cheaply.
Notes on Optimization has been out of print for 20 years. However, several people have been
using it as a text or as a reference in a course. They have urged me to republish it. The idea of
making it freely available over the Web was attractive because it reafﬁrmed the original aim. The
only obstacle was to retype the manuscript in LaTex. I thank Kate Klohe for doing just that.
I would appreciate knowing if you ﬁnd any mistakes in the book, or if you have suggestions for
(small) changes that would improve it.
Berkeley, California P.P. Varaiya
September, 1998
v
vi CONTENTS
PREFACE
These Notes were developed for a tenweek course I have taught for the past three years to ﬁrstyear
graduate students of the University of California at Berkeley. My objective has been to present,
in a compact and uniﬁed manner, the main concepts and techniques of mathematical programming
and optimal control to students having diverse technical backgrounds. A reasonable knowledge of
advanced calculus (up to the Implicit Function Theorem), linear algebra (linear independence, basis,
matrix inverse), and linear differential equations (transition matrix, adjoint solution) is sufﬁcient for
the reader to follow the Notes.
The treatment of the topics presented here is deep. Although the coverage is not encyclopedic,
an understanding of this material should enable the reader to follow much of the recent technical
literature on nonlinear programming, (deterministic) optimal control, and mathematical economics.
The examples and exercises given in the text form an integral part of the Notes and most readers will
need to attend to them before continuing further. To facilitate the use of these Notes as a textbook,
I have incurred the cost of some repetition in order to make almost all chapters selfcontained.
However, Chapter V must be read before Chapter VI, and Chapter VII before Chapter VIII.
The selection of topics, as well as their presentation, has been inﬂuenced by many of my students
and colleagues, who have read and criticized earlier drafts. I would especially like to acknowledge
the help of Professors M. Athans, A. Cohen, C.A. Desoer, JP. Jacob, E. Polak, and Mr. M. Ripper. I
also want to thank Mrs. Billie Vrtiak for her marvelous typing in spite of starting from a not terribly
legible handwritten manuscript. Finally, I want to thank Professor G.L. Turin for his encouraging
and patient editorship.
Berkeley, California P.P. Varaiya
November, 1971
vii
viii CONTENTS
Chapter 1
INTRODUCTION
In this chapter, we present our model of the optimal decisionmaking problem, illustrate decision
making situations by a few examples, and brieﬂy introduce two more general models which we
cannot discuss further in these Notes.
1.1 The Optimal Decision Problem
These Notes show how to arrive at an optimal decision assuming that complete information is given.
The phrase complete information is given means that the following requirements are met:
1. The set of all permissible decisions is known, and
2. The cost of each decision is known.
When these conditions are satisﬁed, the decisions can be ranked according to whether they incur
greater or lesser cost. An optimal decision is then any decision which incurs the least cost among
the set of permissible decisions.
In order to model a decisionmaking situation in mathematical terms, certain further requirements
must be satisﬁed, namely,
1. The set of all decisions can be adequately represented as a subset of a vector space with each
vector representing a decision, and
2. The cost corresponding to these decisions is given by a realvalued function.
Some illustrations will help.
Example 1: The Pot Company (Potco) manufacturers a smoking blend called Acapulco Gold.
The blend is made up of tobacco and maryjohn leaves. For legal reasons the fraction α of mary
john in the mixture must satisfy 0 < α <
1
2
. From extensive market research Potco has determined
their expected volume of sales as a function of α and the selling price p. Furthermore, tobacco can
be purchased at a ﬁxed price, whereas the cost of maryjohn is a function of the amount purchased.
If Potco wants to maximize its proﬁts, how much maryjohn and tobacco should it purchase, and
what price p should it set?
Example 2: Tough University provides “quality” education to undergraduate and graduate stu
dents. In an agreement signed with Tough’s undergraduates and graduates (TUGs), “quality” is
1
2 CHAPTER 1. INTRODUCTION
deﬁned as follows: every year, each u (undergraduate) must take eight courses, one of which is a
seminar and the rest of which are lecture courses, whereas each g (graduate) must take two seminars
and ﬁve lecture courses. A seminar cannot have more than 20 students and a lecture course cannot
have more than 40 students. The University has a faculty of 1000. The Weary Old Radicals (WORs)
have a contract with the University which stipulates that every junior faculty member (there are 750
of these) shall be required to teach six lecture courses and two seminars each year, whereas every
senior faculty member (there are 250 of these) shall teach three lecture courses and three seminars
each year. The Regents of Touch rate Tough’s President at α points per u and β points per g “pro
cessed” by the University. Subject to the agreements with the TUGs and WORs how many u’s and
g’s should the President admit to maximize his rating?
Example 3: (See Figure 1.1.) An engineer is asked to construct a road (broken line) connection
point a to point b. The current proﬁle of the ground is given by the solid line. The only requirement
is that the ﬁnal road should not have a slope exceeding 0.001. If it costs $c per cubic foot to excavate
or ﬁll the ground, how should he design the road to meet the speciﬁcations at minimum cost?
Example 4: Mr. Shell is the manager of an economy which produces one output, wine. There
are two factors of production, capital and labor. If K(t) and L(t) respectively are the capital stock
used and the labor employed at time t, then the rate of output of wine W(t) at time is given by the
production function
W(t) = F(K(t), L(t))
As Manager, Mr. Shell allocates some of the output rate W(t) to the consumption rate C(t), and
the remainder I(t) to investment in capital goods. (Obviously, W, C, I, and K are being measured
in a common currency.) Thus, W(t) = C(t) + I(t) = (1 − s(t))W(t) where s(t) = I(t)/W(t)
.
.
a
b
Figure 1.1: Admissable set of example.
∈ [0, 1] is the fraction of output which is saved and invested. Suppose that the capital stock decays
exponentially with time at a rate δ > 0, so that the net rate of growth of capital is given by the
following equation:
˙
K(t) =
d
dt
K(t) (1.1)
= −δK(t) +s(t)W(t)
= −δK(t) +s(t)F(K(t), L(t)).
The labor force is growing at a constant birth rate of β > 0. Hence,
1.1. THE OPTIMAL DECISION PROBLEM 3
˙
L(t) = βL(t).
(1.2)
Suppose that the production function F exhibits constant returns to scale, i.e., F(λK, λL) =
λF(K, L) for all λ > 0. If we deﬁne the relevant variable in terms of per capita of labor, w =
W/L, c = C/L, k = K/l, and if we let f(k) = F(k, l), then we see that F(K, L)−LF(K/L, 1) =
Lf(k), whence the consumption per capita of labor becomes c(t) = (l −s(t))f(k(t)). Using these
deﬁnitions and equations (1.1) and (1.2) it is easy to see that K(t) satisﬁes the differential equation
(1.3).
˙
k(t) = s(t)f(k(t)) −µk(t)
(1.3)
where µ = (δ +β). The ﬁrst term of the righthand side in (3) is the increase in the capitaltolabor
ratio due to investment whereas the second terms is the decrease due to depreciation and increase in
the labor force.
Suppose there is a planning horizon time T, and at time 0 Mr. Shell starts with capitaltolabor
ratio k
o
. If “welfare” over the planning period [0, T] is identiﬁed with total consumption
T
0
c(t)dt,
what should Mr. Shell’s savings policy s(t), 0 ≤ t ≤ T, be so as to maximize welfare? What
savings policy maximizes welfare subject to the additional restriction that the capitaltolabor ratio
at time T should be at least k
T
? If future consumption is discounted at rate α > 0 and if time horizon
is ∞, the welfare function becomes
∞
0
e
−
αt c(t)dt. What is the optimum policy corresponding to
this criterion?
These examples illustrate the kinds of decisionmaking problems which can be formulated math
ematically so as to be amenable to solutions by the theory presented in these Notes. We must always
remember that a mathematical formulation is inevitably an abstraction and the gain in precision may
have occurred at a great loss of realism. For instance, Example 2 is caricature (see also a faintly re
lated but more more elaborate formulation in Bruno [1970]), whereas Example 4 is lightyears away
from reality. In the latter case, the value of the mathematical exercise is greater the more insensitive
are the optimum savings policies with respect to the simplifying assumptions of the mathematical
model. (In connection with this example and related models see the critique by Koopmans [1967].)
In the examples above, the set of permissible decisions is represented by the set of all points
in some vector space which satisfy certain constraints. Thus, in the ﬁrst example, a permissible
decision is any twodimensional vector (α, p) satisfying the constraints 0 < α <
1
2
and 0 <
p. In the second example, any vector (u, g) with u ≥ 0, g ≥ 0, constrained by the number
of faculty and the agreements with the TUGs and WORs is a permissible decision. In the last
example, a permissible decision is any realvalued function s(t), 0 ≤ t ≤ T, constrained by
0 ≤ s(t) ≤ 1. (It is of mathematical but not conceptual interest to note that in this case a decision
is represented by a vector in a function space which is inﬁnitedimensional.) More concisely then,
these Notes are concerned with optimizing (i.e. maximizing or minimizing) a realvalued function
over a vector space subject to constraints. The constraints themselves are presented in terms of
functional inequalities or equalities.
4 CHAPTER 1. INTRODUCTION
At this point, it is important to realize that the distinction between the function which is to be
optimized and the functions which describe the constraints, although convenient for presenting the
mathematical theory, may be quite artiﬁcial in practice. For instance, suppose we have to choose
the durations of various trafﬁc lights in a section of a city so as to achieve optimum trafﬁc ﬂow.
Let us suppose that we know the transportation needs of all the people in this section. Before we
can begin to suggest a design, we need a criterion to determine what is meant by “optimum trafﬁc
ﬂow.” More abstractly, we need a criterion by which we can compare different decisions, which in
this case are different patterns of trafﬁclight durations. One way of doing this is to assign as cost to
each decision the total amount of time taken to make all the trips within this section. An alternative
and equally plausible goal may be to minimize the maximum waiting time (that is the total time
spent at stop lights) in each trip. Now it may happen that these two objective functions may be
inconsistent in the sense that they may give rise to different orderings of the permissible decisions.
Indeed, it may be the case that the optimum decision according to the ﬁrst criterion may be lead to
very long waiting times for a few trips, so that this decision is far from optimum according to the
second criterion. We can then redeﬁne the problem as minimizing the ﬁrst cost function (total time
for trips) subject to the constraint that the waiting time for any trip is less than some reasonable
bound (say one minute). In this way, the second goal (minimum waiting time) has been modiﬁed
and reintroduced as a constraint. This interchangeability of goal and constraints also appears at a
deeper level in much of the mathematical theory. We will see that in most of the results the objective
function and the functions describing the constraints are treated in the same manner.
1.2 Some Other Models of Decision Problems
Our model of a single decisionmaker with complete information can be generalized along two
very important directions. In the ﬁrst place, the hypothesis of complete information can be relaxed
by allowing that decisionmaking occurs in an uncertain environment. In the second place, we
can replace the single decisionmaker by a group of two or more agents whose collective decision
determines the outcome. Since we cannot study these more general models in these Notes, we
merely point out here some situations where such models arise naturally and give some references.
1.2.1 Optimization under uncertainty.
A person wants to invest $1,000 in the stock market. He wants to maximize his capital gains, and
at the same time minimize the risk of losing his money. The two objectives are incompatible, since
the stock which is likely to have higher gains is also likely to involve greater risk. The situation
is different from our previous examples in that the outcome (future stock prices) is uncertain. It is
customary to model this uncertainty stochastically. Thus, the investor may assign probability 0.5 to
the event that the price of shares in Glamor company increases by $100, probability 0.25 that the
price is unchanged, and probability 0.25 that it drops by $100. A similar model is made for all the
other stocks that the investor is willing to consider, and a decision problem can be formulated as
follows. How should $1,000 be invested so as to maximize the expected value of the capital gains
subject to the constraint that the probability of losing more than $100 is less than 0.1?
As another example, consider the design of a controller for a chemical process where the decision
variable are temperature, input rates of various chemicals, etc. Usually there are impurities in the
chemicals and disturbances in the heating process which may be regarded as additional inputs of a
1.2. SOME OTHER MODELS OF DECISION PROBLEMS 5
random nature and modeled as stochastic processes. After this, just as in the case of the portfolio
selection problem, we can formulate a decision problem in such a way as to take into account these
random disturbances.
If the uncertainties are modelled stochastically as in the example above, then in many cases
the techniques presented in these Notes can be usefully applied to the resulting optimal decision
problem. To do justice to these decisionmaking situations, however, it is necessary to give great
attention to the various ways in which the uncertainties can be modelled mathematically. We also
need to worry about ﬁnding equivalent but simpler formulations. For instance, it is of great signif
icance to know that, given appropriate conditions, an optimal decision problem under uncertainty
is equivalent to another optimal decision problem under complete information. (This result, known
as the CertaintyEquivalence principle in economics has been extended and baptized the Separation
Theorem in the control literature. See Wonham [1968].) Unfortunately, to be able to deal with
these models, we need a good background in Statistics and Probability Theory besides the material
presented in these Notes. We can only refer the reader to the extensive literature on Statistical De
cision Theory (Savage [1954], Blackwell and Girshick [1954]) and on Stochastic Optimal Control
(Meditch [1969], Kushner [1971]).
1.2.2 The case of more than one decisionmaker.
Agent Alpha is chasing agent Beta. The place is a large circular ﬁeld. Alpha is driving a fast, heavy
car which does not maneuver easily, whereas Beta is riding a motor scooter, slow but with good
maneuverability. What should Alpha do to get as close to Beta as possible? What should Beta
do to stay out of Alpha’s reach? This situation is fundamentally different from those discussed so
far. Here there are two decisionmakers with opposing objectives. Each agent does not know what
the other is planning to do, yet the effectiveness of his decision depends crucially upon the other’s
decision, so that optimality cannot be deﬁned as we did earlier. We need a new concept of rational
(optimal) decisionmaking. Situations such as these have been studied extensively and an elaborate
structure, known as the Theory of Games, exists which describes and prescribes behavior in these
situations. Although the practical impact of this theory is not great, it has proved to be among the
most fruitful sources of unifying analytical concepts in the social sciences, notably economics and
political science. The best single source for Game Theory is still Luce and Raiffa [1957], whereas
the mathematical content of the theory is concisely displayed in Owen [1968]. The control theorist
will probably be most interested in Isaacs [1965], and Blaquiere, et al., [1969].
The difﬁculty caused by the lack of knowledge of the actions of the other decisionmaking agents
arises even if all the agents have the same objective, since a particular decision taken by our agent
may be better or worse than another decision depending upon the (unknown) decisions taken by the
other agents. It is of crucial importance to invent schemes to coordinate the actions of the individual
decisionmakers in a consistent manner. Although problems involving many decisionmakers are
present in any system of large size, the number of results available is pitifully small. (See Mesarovic,
et al., [1970] and Marschak and Radner [1971].) In the author’s opinion, these problems represent
one of the most important and challenging areas of research in decision theory.
6 CHAPTER 1. INTRODUCTION
Chapter 2
OPTIMIZATION OVER AN OPEN
SET
In this chapter we study in detail the ﬁrst example of Chapter 1. We ﬁrst establish some notation
which will be in force throughout these Notes. Then we study our example. This will generalize
to a canonical problem, the properties of whose solution are stated as a theorem. Some additional
properties are mentioned in the last section.
2.1 Notation
2.1.1
All vectors are column vectors, with two consistent exceptions mentioned in 2.1.3 and 2.1.5 below
and some other minor and convenient exceptions in the text. Prime denotes transpose so that if
x ∈ R
n
then x
is the row vector x
= (x
1
, . . . , x
n
), and x = (x
1
, . . . , x
n
)
. Vectors are normally
denoted by lower case letters, the ith component of a vector x ∈ R
n
is denoted x
i
, and different
vectors denoted by the same symbol are distinguished by superscripts as in x
j
and x
k
. 0 denotes
both the zero vector and the real number zero, but no confusion will result.
Thus if x = (x
1
, . . . , x
n
)
and y = (y
1
, . . . , y
n
)
then x
y = x
1
y
1
+ . . . + x
n
y
n
as in ordinary
matrix multiplication. If x ∈ R
n
we deﬁne [x[ = +
√
x
x.
2.1.2
If x = (x
1
, . . . , x
n
)
and y = (y
1
, . . . , y
n
)
then x ≥ y means x
i
≥ y
i
, i = 1, . . . , n. In particular if
x ∈ R
n
, then x ≥ 0, if x
i
≥ 0, i = 1, . . . , n.
2.1.3
Matrices are normally denoted by capital letters. If A is an m n matrix, then A
j
denotes the jth
column of A, and A
i
denotes the ith row of A. Note that A
i
is a row vector. A
j
i
denotes the entry
of A in the ith row and jth column; this entry is sometimes also denoted by the lower case letter
a
ij
, and then we also write A = ¦a
ij
¦. I denotes the identity matrix; its size will be clear from the
context. If confusion is likely, we write I
n
to denote the n n identity matrix.
7
8 CHAPTER 2. OPTIMIZATION OVER AN OPEN SET
2.1.4
If f : R
n
→ R
m
is a function, its ith component is written f
i
, i = 1, . . . , m. Note that f
i
: R
n
→ R.
Sometimes we describe a function by specifying a rule to calculate f(x) for every x. In this case
we write f : x → f(x). For example, if A is an mn matrix, we can write F : x → Ax to denote
the function f : R
n
→ R
m
whose value at a point x ∈ R
n
is Ax.
2.1.5
If f : R
n
→ Ris a differentiable function, the derivative of f at ˆ x is the row vector ((∂f/∂x
1
)(ˆ x), . . . , (∂f/∂x
n
)(ˆ x)).
This derivative is denoted by (∂f/∂x)(ˆ x) or f
x
(ˆ x) or ∂f/∂x[
x=ˆ x
or f
x
[
x=ˆ x
, and if the argument ˆ x
is clear from the context it may be dropped. The column vector (f
x
(ˆ x))
is also denoted ∇
x
f(ˆ x),
and is called the gradient of f at ˆ x. If f : (x, y) → f(x, y) is a differentiable function from
R
n
R
m
into R, the partial derivative of f with respect to x at the point (ˆ x, ˆ y) is the ndimensional
row vector f
x
(ˆ x, ˆ y) = (∂f/∂x)(ˆ x, ˆ y) = ((∂f/∂x
1
)(ˆ x, ˆ y), . . . , (∂f/∂x
n
)(ˆ x, ˆ y)), and similarly
f
y
(ˆ x, ˆ y) = (∂f/∂y)(ˆ x, ˆ y) = ((∂f/∂y
1
)(ˆ x, ˆ y), . . . , (∂f/∂y
m
)(ˆ x, ˆ y)). Finally, if f : R
n
→ R
m
is
a differentiable function with components f
1
, . . . , f
m
, then its derivative at ˆ x is the mn matrix
∂f
∂x
(ˆ x) = f
x
ˆ x =
f
1x
(ˆ x)
.
.
.
f
mx
(ˆ x)
¸
¸
¸
=
∂f
1
∂x
1
(ˆ x)
.
.
.
∂fm
∂x
1
(ˆ x)
. . .
. . .
∂f
1
∂xn
(ˆ x)
.
.
.
∂fm
∂xn
(ˆ x)
¸
¸
¸
¸
2.1.6
If f : R
n
→ Ris twice differentiable, its second derivative at ˆ x is the nn matrix (∂
2
f/∂x∂x)(ˆ x) =
f
xx
(ˆ x) where (f
xx
(ˆ x))
j
i
= (∂
2
f/∂x
j
∂x
i
)(ˆ x). Thus, in terms of the notation in Section 2.1.5 above,
f
xx
(ˆ x) = (∂/∂x)(f
x
)
(ˆ x).
2.2 Example
We consider in detail the ﬁrst example of Chapter 1. Deﬁne the following variables and functions:
α = fraction of maryjohn in proposed mixture,
p = sale price per pound of mixture,
v = total amount of mixture produced,
f(α, p) = expected sales volume (as determined by market research) of mixture as a function of(α, p).
2.2. EXAMPLE 9
Since it is not proﬁtable to produce more than can be sold we must have:
v = f(α, p),
m = amount (in pounds) of maryjohn purchased, and
t = amount (in pounds) of tobacco purchased.
Evidently,
m = αv, and
t = (l −α)v.
Let
P
1
(m) = purchase price of m pounds of maryjohn, and
P
2
= purchase price per pound of tobacco.
Then the total cost as a function of α, p is
C(α, p) = P
1
(αf(α, p)) +P
2
(1 −α)f(α, p).
The revenue is
R(α, p) = pf(α, p),
so that the net proﬁt is
N(α, p) = R(α, p) −C(α, p).
The set of admissible decisions is Ω, where Ω = ¦(α, p)[0 < α <
1
2
, 0 < p < ∞¦. Formally, we
have the the following decision problem:
Maximize
subject to
N(α, p),
(α, p) ∈ Ω.
Suppose that (α
∗
, p∗) is an optimal decision, i.e.,
(α
∗
, p
∗
) ∈ Ω
N(α
∗
, p
∗
) ≥ N(α, p)
and
for all (α, p) ∈ Ω.
(2.1)
We are going to establish some properties of (a
∗
, p
∗
). First of all we note that Ω is an open subset
of R
2
. Hence there exits ε > 0 such that
(α, p) ∈ Ω whenever [(α, p) −(α
∗
, p
∗
)[ < ε (2.2)
In turn (2.2) implies that for every vector h = (h
1
, h
2
)
in R
2
there exists η > 0 (η of course
depends on h) such that
((α
∗
, p
∗
) +δ(h
1
, h
2
)) ∈ Ω for 0 ≤ δ ≤ η (2.3)
10 CHAPTER 2. OPTIMIZATION OVER AN OPEN SET

.
(α
∗
, p
∗
) +δ(h
1
, h
2
)
α
1
2
Ω
δh
h
p
(a
∗
, p
∗
)
Figure 2.1: Admissable set of example.
Combining (2.3) with (2.1) we obtain (2.4):
N(α
∗
, p
∗
) ≥ N(α
∗
+δh
1
, p
∗
+δh
2
) for 0 ≤ δ ≤ η (2.4)
Now we assume that the function N is differentiable so that by Taylor’s theorem
N(α
∗
+δh
1
, p
∗
+δh
2
) =
N(α
∗
, p
∗
)
+δ[
∂N
∂α
(δ
∗
, p
∗
)h
1
+
∂N
∂p
(α
∗
, p
∗
)h
2
]
+o(δ),
(2.5)
where
oδ
δ
→ 0 as δ → 0. (2.6)
Substitution of (2.5) into (2.4) yields
0 ≥ δ[
∂N
∂α
(α
∗
, p
∗
)h
1
+
∂N
∂p
(α
∗
, p
∗
)h
2
] +o(δ).
Dividing by δ > 0 gives
0 ≥ [
∂N
∂α
(α
∗
, p
∗
)h
1
+
∂N
∂p
(α
∗
, p
∗
)h
2
] +
o(δ)
δ
. (2.7)
Letting δ approach zero in (2.7), and using (2.6) we get
0 ≥ [
∂N
∂α
(α
∗
, p
∗
)h
1
+
∂N
∂p
(α
∗
, p
∗
)h
2
]. (2.8)
Thus, using the facts that N is differentiable, (α
∗
, p
∗
) is optimal, and δ is open, we have concluded
that the inequality (2.9) holds for every vector h ∈ R
2
. Clearly this is possible only if
∂N
∂α
(α
∗
, p
∗
) = 0,
∂N
∂p
(α
∗
, p
∗
) = 0. (2.9)
Before evaluating the usefulness of property (2.8), let us prove a direct generalization.
2.3. THE MAIN RESULT AND ITS CONSEQUENCES 11
2.3 The Main Result and its Consequences
2.3.1 Theorem
.
Let Ω be an open subset of R
n
. Let f: R
n
→ R be a differentiable function. Let x
∗
be an optimal
solution of the following decisionmaking problem:
Maximize
subject to
f(x)
x ∈ Ω.
(2.10)
Then
∂f
∂x
(x
∗
) = 0. (2.11)
Proof: Since x
∗
∈ Ω and Ω is open, there exists ε > 0 such that
x ∈ Ω whenever [x −x
∗
[ < ε. (2.12)
In turn, (2.12) implies that for every vector h ∈ R
n
there exits η > 0 (η depending on h) such that
(x
∗
+δh) ∈ Ω whenever 0 ≤ δ ≤ η. (2.13)
Since x
∗
is optimal, we must then have
f(x
∗
) ≥ f(x
∗
+δh) whenever 0 ≤ δ ≤ η. (2.14)
Since f is differentiable, by Taylor’s theorem we have
f(x
∗
+δh) = f(x
∗
) +
∂f
∂x
(x
∗
)δh +o(δ), (2.15)
where
o(δ)
δ
→ 0 as δ → 0 (2.16)
Substitution of (2.15) into (2.14) yields
0 ≥ δ
∂f
∂x
(x
∗
)h +o(δ)
and dividing by δ > 0 gives
0 ≥
∂f
∂x
(x
∗
)h +
o(δ)
δ
(2.17)
Letting δ approach zero in (2.17) and taking (2.16) into account, we see that
0 ≥
∂f
∂x
(x
∗
)h, (2.18)
Since the inequality (2.18) must hold for every h ∈ R
n
, we must have
0 =
∂f
∂x
(x
∗
),
and the theorem is proved. ♦
12 CHAPTER 2. OPTIMIZATION OVER AN OPEN SET
Table 2.1
Does there exist At how many points
an optimal deci in Ω is 2.2.2 Further
Case sion for 2.2.1? satisﬁed? Consequences
1 Yes Exactly one point, x
∗
is the
say x
∗
unique optimal
2 Yes More than one point
3 No None
4 No Exactly one point
5 No More than one point
2.3.2 Consequences.
Let us evaluate the usefulness of (2.11) and its special case (2.18). Equation (2.11) gives us n
equations which must be satisﬁed at any optimal decision x
∗
= (x
∗
1
, . . . , x
∗
n
)
.
These are
∂f
∂x
1
(x
∗
) = 0,
∂f
∂x
2
(x
∗
) = 0, . . . ,
∂f
∂xn
(x
∗
) = 0 (2.19)
Thus, every optimal decision must be a solution of these n simultaneous equations of n variables, so
that the search for an optimal decision from Ω is reduced to searching among the solutions of (2.19).
In practice this may be a very difﬁcult problem since these may be nonlinear equations and it may
be necessary to use a digital computer. However, in these Notes we shall not be overly concerned
with numerical solution techniques (but see 2.4.6 below).
The theorem may also have conceptual signiﬁcance. We return to the example and recall the
N = R − C. Suppose that R and C are differentiable, in which case (2.18) implies that at every
optimal decision (α
∗
, p
∗
)
∂R
∂α
(α
∗
, p
∗
) =
∂C
∂α
(α
∗
, p
∗
),
∂R
∂p
(α
∗
, p
∗
) =
∂C
∂p
(α
∗
, p
∗
),
or, in the language of economic analysis, marginal revenue = marginal cost. We have obtained an
important economic insight.
2.4 Remarks and Extensions
2.4.1 A warning.
Equation (2.11) is only a necessary condition for x
∗
to be optimal. There may exist decisions ˜ x ∈ Ω
such that f
x
(˜ x) = 0 but ˜ x is not optimal. More generally, any one of the ﬁve cases in Table 2.1 may
occur. The diagrams in Figure 2.1 illustrate these cases. In each case Ω = (−1, 1).
Note that in the last three ﬁgures there is no optimal decision since the limit points 1 and +1 are
not in the set of permissible decisions Ω = (−1, 1). In summary, the theorem does not give us any
clues concerning the existence of an optimal decision, and it does not give us sufﬁcient conditions
either.
2.4. REMARKS AND EXTENSIONS 13
Case 1 Case 2 Case 3
Case 5 Case 4
1
1 1 1
1 1
1 1 1 1
Figure 2.2: Illustration of 4.1.
2.4.2 Existence.
If the set of permissible decisions Ω is a closed and bounded subset of R
n
, and if f is continuous,
then it follows by the Weierstrass Theorem that there exists an optimal decision. But if Ω is closed
we cannot assert that the derivative of f vanishes at the optimum. Indeed, in the third ﬁgure above,
if Ω = [−1, 1], then +1 is the optimal decision but the derivative is positive at that point.
2.4.3 Local optimum.
We say that x
∗
∈ Ω is a locally optimal decision if there exists ε > 0 such that f(x
∗
) ≥ f(x)
whenever x ∈ Ω and [x
∗
− x[ ≤ ε. It is easy to see that the theorem holds (i.e., 2.11) for local
optima also.
2.4.4 Secondorder conditions.
Suppose f is twicedifferentiable and let x
∗
∈ Ω be optimal or even locally optimal. Then f
x
(x
∗
) =
0, and by Taylor’s theorem
f(x
∗
+δh) = f(x
∗
) +
1
2
δ
2
h
f
xx
(x
∗
)h +o(δ
2
), (2.20)
where
o(δ
2
)
δ
2
→ 0 as δ → 0. Now for δ > 0 sufﬁciently small f(x
∗
+δh) ≤ f(x
∗
), so that dividing
by δ
2
> 0 yields
0 ≥
1
2
h
f
xx
(x
∗
)h +
o(δ
2
)
δ
2
and letting δ approach zero we conclude that h
f
xx
(x
∗
)h ≤ 0 for all h ∈ R
n
. This means that
f
xx
(x
∗
) is a negative semideﬁnite matrix. Thus, if we have a twice differentiable objective function,
we get an additional necessary condition.
2.4.5 Sufﬁciency for local optimal.
Suppose at x
∗
∈ Ω, f
x
(x
∗
) = 0 and f
xx
is strictly negative deﬁnite. But then from the expansion
(2.20) we can conclude that x
∗
is a local optimum.
14 CHAPTER 2. OPTIMIZATION OVER AN OPEN SET
2.4.6 A numerical procedure.
At any point ˜ x ∈ Ω the gradient
x
f(˜ x) is a direction along which f(x) increases, i.e., f(˜ x+ε
x
f(˜ x)) > f(˜ x) for all ε > 0 sufﬁciently small. This observation suggests the following scheme for
ﬁnding a point x
∗
∈ Ω which satisﬁes 2.11. We can formalize the scheme as an algorithm.
Step 1. Pick x
0
∈ Ω. Set i = 0. Go to Step 2.
Step 2. Calculate
x
f(x
i
). If
x
f(x
i
) = 0, stop.
Otherwise let x
i+1
= x
i
+d
i
x
f(x
i
) and go
to Step 3.
Step 3. Set i = i + 1 and return to Step 2.
The step size d
i
can be selected in many ways. For instance, one choice is to take d
i
to be an
optimal decision for the following problem:
Max¦f(x
i
+d
x
f(x
i
))[d > 0, (x
i
+d
x
f(x
i
)) ∈ Ω¦.
This requires a onedimensional search. Another choice is to let d
i
= d
i−1
if f(x
i
+ d
i−1
x
f(x
i
)) > f(x
i
); otherwise let d
i
= 1/k d
i−1
where k is the smallest positive integer such that
f(x
i
+ 1/k d
i−1
x
f(x
i
)) > f(x
i
). To start the process we let d
−1
> 0 be arbitrary.
Exercise: Let f be continuous differentiable. Let ¦d
i
¦ be produced by either of these choices and
let
¦x
i
¦ be the resulting sequence. Then
1. f(x
i+1
) > f(x
i
) if x
i+1
= x
i
, i
2. if x
∗
∈ Ω is a limit point of the sequence ¦x
i
¦, f
x
(x
∗
) = 0.
For other numerical procedures the reader is referred to Zangwill [1969] or Polak [1971].
Chapter 3
OPTIMIZATION OVER SETS
DEFINED BY EQUALITY
CONSTRAINTS
We ﬁrst study a simple example and examine the properties of an optimal decision. This will
generalize to a canonical problem, and the properties of its optimal decisions are stated in the form
of a theorem. Additional properties are summarized in Section 3 and a numerical scheme is applied
to determine the optimal design of resistive networks.
3.1 Example
We want to ﬁnd the rectangle of maximum area inscribed in an ellipse deﬁned by
f
1
(x, y) =
x
2
a
2
+
y
2
b
2
= α.
(3.1)
The problem can be formalized as follows (see Figure 3.1):
Maximize
subject to
f
0
(x, y)
(x, y) ∈ Ω
= 4xy
= ¦(x, y)[f
1
(x, y) = α¦.
(3.2)
The main difference between problem (3.2) and the decisions studied in the last chapter is that
the set of permissible decisions Ω is not an open set. Hence, if (x
∗
, y
∗
) is an optimal decision we
cannot assert that f
0
(x
∗
, y
∗
) ≥ f
0
(x, y) for all (x, y) in an open set containing (x
∗
, y
∗
). Returning
to problem (3.2), suppose (x
∗
, y
∗
) is an optimal decision. Clearly then either x
∗
= 0 or y
∗
= 0. Let
us suppose y
∗
= 0. Then from ﬁgure 3.1 it is evident that there exist (i)ε > 0, (ii) an open set V
containing (x
∗
, y
∗
), and (iii) a differentiable function g : (x
∗
−ε, x
∗
+ε) → V such that
f
1
(x, y) = α and (x, y) ∈ V iff fy = g(x).
1
(3.3)
In particular this implies that y
∗
= g(x
∗
), and that f
1
(x, g(x)) = α whenever [x − x
∗
[ < ε. Since
1
Note that y
∗
= 0 implies f1y(x
∗
, Y
∗
) = 0, so that this assertion follows from the Implicit Function Theorem. The
assertion is false if y
∗
= 0. In the present case let 0 < ε ≤ a −x
∗
and g(x) = +b[α −(x/a)
2
]
1/2
.
15
16 CHAPTER 3. OPTIMIZATION WITH EQUALITY CONSTRAINTS
) ( 

y
∗
g(x)
Tangent plane to
Ω at (x
∗
, y
∗
)
(f
1x
, f
1y
)
V
x
∗
x
Ω
Figure 3.1: Illustration of example.
(x
∗
, y
∗
) = (x
∗
, g(x
∗
)) is optimum for (3.2), it follows that x
∗
is an optimal solution for (3.4):
Maximize
subject to
ˆ
f
0
(x) = f
0
(x, g(x))
[x −x
∗
[ < ε.
(3.4)
But the constraint set in (3.4) is an open set (in R
1
) and the objective function
ˆ
f
0
is differentiable,
so that by Theorem 2.3.1,
ˆ
f
0x
(x
∗
) = 0, which we can also express as
f
0x
(x
∗
, y
∗
) +f
0y
(x
∗
, y
∗
)g
x
(x
∗
) = 0 (3.5)
Using the fact that f
1
(x, g(x)) ≡ α for [x −x
∗
[ < ε, we see that
f
1x
(x
∗
, y
∗
) +f
1y
(x
∗
, y
∗
)g
x
(x
∗
) = 0,
and since f
1y
(x
∗
, y
∗
) = 0 we can evaluate g
x
(x
∗
),
g
x
(x
∗
) = −f
−1
1y
f
1x
(x
∗
, y
∗
),
and substitute in (3.5) to obtain the condition (3.6):
f
0x
−f
0y
f
−1
1y
f
1x
= 0 at (x
∗
, y
∗
). (3.6)
Thus an optimal decision (x
∗
, y
∗
) must satisfy the two equations f
1
(x
∗
, y
∗
) = α and (3.6). Solving
these yields
x
∗
=
+
−
(α/2)
1/2
a , y
∗
=
+
−
(α/2)
1/2
b.
3.2. GENERAL CASE 17
Evidently there are two optimal decisions, (x
∗
, y
∗
) =
+
−
(α/2)
1/2
(a, b), and the maximum area is
m(α) = 2αab. (3.7)
The condition (3.6) can be interpreted differently. Deﬁne
λ
∗
= f
0y
f
−1
1y
(x
∗
, y
∗
). (3.8)
Then (3.6) and (3.8) can be rewritten as (3.9):
(f
0x
, f
0y
) = λ
∗
(f
1x
, f
1y
) at (x
∗
, y
∗
) (3.9)
In terms of the gradients of f
0
, f
1
, (3.9) is equivalent to
f
0
(x
∗
, y
∗
) = [f
1
(x
∗
, y
∗
)]λ
∗
, (3.10)
which means that at an optimal decision the gradient of the objective function f
0
is normal to the
plane tangent to the constraint set Ω.
Finally we note that
λ
∗
=
∂m
∂α
. (3.11)
where m(α) = maximum area.
3.2 General Case
3.2.1 Theorem.
Let f
i
: R
n
→ R, i = 0, 1, . . . , m (m < n), be continuously differentiable functions and let x
∗
be
an optimal decision of problem (3.12):
Maximize
subject to
f
0
(x)
f
i
(x) = α
i
, i = 1, . . . , m.
(3.12)
Suppose that at x
∗
the derivatives f
ix
(x
∗
), i = 1, . . . , m, are linearly independent. Then there exists
a vector λ
∗
= (λ
∗
1
, . . . , λ
∗
m
)
such that
f
0x
(x
∗
) = λ
∗
1
f
1x
(x
∗
) +. . . +λ
∗
m
f
mx
(x
∗
) (3.13)
Furthermore, let m(α
1
, . . . , α
m
) be the maximum value of (3.12) as a function of α = (α
1
, . . . , α
m
)
.
Let x
∗
(α) be an optimal decision for (3.12). If x
∗
(α) is a differentiable function of α then m(α) is
a differentiable function of α, and
(λ
∗
)
=
∂m
∂α
(3.14)
Proof. Since f
ix
(x
∗
), i = 1, . . . , m, are linearly independent, then by relabeling the coordinates of
x if necessary, we can assume that the mmmatrix [(∂f
i
/∂x
j
)(x
∗
)], 1 ≤ i, j ≤ m, is nonsingular.
By the Implicit Function Theorem (see Fleming [1965]) it follows that there exist (i) ε > 0, (ii) an
18 CHAPTER 3. OPTIMIZATION WITH EQUALITY CONSTRAINTS
open set V in R
n
containing x
∗
, and (iii) a differentiable function g : U → R
m
, where U =
[(x
m+1
, . . . , x
n
)][ [x
m+
−x
∗
m+
[ < ε, = 1, . . . , n −m], such that
f
i
(x
1
, . . . , x
n
) = α
i
, 1 ≤ i ≤ m, and (x
1
, . . . , x
n
) ∈ V
iff
x
j
= g
j
(x
m+1
, . . . , x
n
), 1 ≤ j ≤ m, and (x
m+1
, . . . , x
n
) ∈ U (3.15)
(see Figure 3.2).
In particular this implies that x
∗
j
= g
j
(x
∗
m+1
, . . . , x
∗
n
), 1 ≤ j ≤ m, and
f
i
(g(x
m+1
, . . . , x
n
), x
m+1
, . . . , x
n
) = α
i
, i = 1, . . . , m. (3.16)
For convenience, let us deﬁne w = (x
1
, . . . , x
m
)
, u = (x
m+1
, . . . , x
n
)
and f = (f
1
, . . . , f
m
)
.
Then, since x
∗
= (w
∗
, u
∗
) = (g(u
∗
), u
∗
) is optimal for (3.12), it follows that u
∗
is an optimal
decision for (3.17):
Maximize
subject to
ˆ
f
0
(u) = f
0
(g(u), u)
u ∈ U.
(3.17)
But U is an open subset of R
n−m
and
ˆ
f
0
is a differentiable function on U (since f
0
and g are
differentiable), so that by Theorem 2.3.1 ,
ˆ
f
0u
(u
∗
) = 0, which we can also express using the chain
rule for derivatives as
ˆ
f
0u
(u
∗
) = f
0w
(x
∗
)g
u
(u
∗
) +f
0u
(x
∗
) = 0. (3.18)
Differentiating (3.16) with respect to u = (x
m+1
, . . . , x
n
)
, we see that
f
w
(x
∗
)g
u
(u
∗
) +f
u
(x
∗
) = 0,
and since the mm matrix f
w
(x
∗
) is nonsingular we can evaluate g
u
(u
∗
),
g
u
(u
∗
) = −[f
w
(x∗)]
−1
f
u
(x
∗
),
and substitute in (3.18) to obtain the condition
−f
0w
f
−1
w
f
u
+f
0u
= 0 at x
∗
= (w
∗
, u
∗
). (3.19)
Next, deﬁne the mdimensional column vector λ
∗
by
(λ
∗
)
= f
0w
f
−1
w
[x
∗
. (3.20)
Then (3.19) and (3.20) can be written as (3.21):
(f
0w
(x
∗
), f
0u
(x
∗
)) = (λ
∗
)
(f
w
(x
∗
), f
u
(x
∗
)). (3.21)
Since x = (w, u), this is the same as
f
0x
(x
∗
) = (λ
∗
)
f
x
(x
∗
) = λ
∗
1
f
1x
(x
∗
) +. . . +λ
∗
m
f
mx
(x
∗
),
3.2. GENERAL CASE 19
.
.
.
.
x
1
, . . . , x
m
x
∗
V
x
m+1
(x
m+1
, . . . , x
n
)
(x
∗
m+1
, . . . , x
∗
n
)
2
U
x
n
Ω =
¦x[f
i
(x) = α
i
¦
i = 1, . . . , m
(x
∗
1
, . . . , x
∗
m
)
g(x
m+1
, . . . , x
n
)
Figure 3.2: Illustration of theorem.
which is equation (3.13).
To prove (3.14), we vary α in a neighborhood of a ﬁxed value, say α. We deﬁne w
∗
(α) =
(x
∗
1
(α), . . . , x
∗
m
(α))
and u
∗
(α) = (x
∗
m+1
(α), . . . , x
∗
(
α))
. By hypothesis, f
w
is nonsingular at
x
∗
(α). Since f(x) and x
∗
(α) are continuously differentiable by hypothesis, it follows that f
w
is
nonsingular at x
∗
(α) in a neighborhood of α, say N. We have the equation
f(w
∗
(α), u
∗
(α)) = α, (3.22)
−f
0w
f
−1
w
f
u
+f
0u
= 0 at (w
∗
(α), u
∗
(α)), (3.23)
for α ∈ N. Also, m(α) = f
0
(x
∗
(α)), so that
m
α
= f
0w
w
∗
α
+f
0u
u
∗
α
(3.24)
Differentiating (3.22) with respect to α gives
f
w
w
∗
α
+f
u
u
∗
α
= I,
so that
w
∗
α
+f
−1
w
f
u
u
∗
α
= f
−1
w
,
20 CHAPTER 3. OPTIMIZATION WITH EQUALITY CONSTRAINTS
and multiplying on the left by f
0w
gives
f
0w
w
∗
α
+f
0w
f
−1
w
f
u
u
∗
α
= f
0w
f
−1
w
.
Using (3.23), this equation can be rewritten as
f
0w
w
∗
α
+f
0u
u
∗
α
= f
0w
f
−1
w
. (3.25)
In (3.25), if we substitute from (3.20) and (3.24), we obtain (3.14) and the theorem is proved. ♦
3.2.2 Geometric interpretation.
The equality constraints of the problem in 3.12 deﬁne a n −m dimensional surface
Ω = ¦x[f
i
(x) = α
i
, i = 1, . . . , m¦.
The hypothesis of linear independence of ¦f
ix
(x
∗
)[1 ≤ i ≤ m¦ guarantees that the tangent plane
through Ω at x
∗
is described by
¦h[f
ix
(x
∗
)h = 0 , i = 1, . . . , m¦, (3.26)
so that the set of (column vectors orthogonal to this tangent surface is
¦λ
1
x
f
1
(x
∗
) +. . . +λ
m
x
f
m
(x
∗
)[λ
i
∈ R, i = 1, . . . , m¦.
Condition (3.13) is therefore equivalent to saying that at an optimal decision x
∗
, the gradient of the
objective function
x
f
0
(x
∗
) is normal to the tangent surface (3.12).
3.2.3 Algebraic interpretation.
Let us again deﬁne w = (x
1
, . . . , x
m
)
and u = (x
m+1
, . . . , x
n
)
. Suppose that f
w
(˜ x) is nonsin
gular at some point ˜ x = ( ˜ w, ˜ u) in Ω which is not necessarily optimal. Then the Implicit Function
Theorem enables us to solve, in a neighborhood of ˜ x, the mequations f(w, u) = α. u can then vary
arbitrarily in a neighborhood of ˜ u. As u varies, w must change according to w = g(u) (in order to
maintain f(w, u) = α), and the objective function changes according to
ˆ
f
0
(u) = f
0
(g(u), u). The
derivative of
ˆ
f
0
at ˜ u is
ˆ
f
0u
(˜ u) = f
0w
g
u
+f
0u˜ x
= −
˜
λ
f
u
(˜ x) +f
0u
(˜ x),
where
˜
λ
= f
0w
f
−1
w˜ x
, (3.27)
Therefore, the direction of steepest increase of
ˆ
f
0
at ˜ u is
u
ˆ
f
0
(˜ u) = −f
u
(˜ x)
˜
λ +f
Ou
(˜ x) , (3.28)
and if ˜ u is optimal,
u
ˆ
f
0
(˜ u) = 0 which, together with (3.27) is equation (3.13). We shall use (3.27)
and (3.28) in the last section.
3.3. REMARKS AND EXTENSIONS 21
3.3 Remarks and Extensions
3.3.1 The condition of linear independence.
The necessary condition (3.13) need not hold if the derivatives f
ix
(x
∗
), 1 ≤ i ≤ m, are not linearly
independent. This can be checked in the following example
Minimize
subject to sin(x
2
1
+x
2
2
)
π
2
(x
2
1
+x
2
2
) = 1.
(3.29)
3.3.2 An alternative condition.
Keeping the notation of Theorem 3.2.1, deﬁne the Lagrangian function L : R
n+m
→ R by L :
(x, λ) → f
0
(x) −
¸
m
i=1
λ
i
f
i
(x). The following is a reformulation of 3.12, and its proof is left as
an exercise.
Let x
∗
be optimal for (3.12), and suppose that f
ix
(x
∗
), 1 ≤ i ≤ m, are linearly independent.
Then there exists λ
∗
∈ R
m
such that (x
∗
, λ
∗
) is a stationary point of L, i.e., L
x
(x
∗
, λ
∗
) = 0 and
L
λ
(x
∗
, λ
∗
) = 0.
3.3.3 Secondorder conditions.
Since we can convert the problem (3.12) into a problem of maximizing
ˆ
f
0
over an open set, all
the comments of Section 2.4 will apply to the function
ˆ
f
0
. However, it is useful to translate these
remarks in terms of the original function f
0
and f. This is possible because the function g is
uniquely speciﬁed by (3.16) in a neighborhood of x
∗
. Furthermore, if f is twice differentiable, so
is g (see Fleming [1965]). It follows that if the functions f
i
, 0 ≤ i ≤ m, are twice continuously
differentiable, then so is
ˆ
f
0
, and a necessary condition for x
∗
to be optimal for (3.12) and (3.13) and
the condition that the (n − m) (n − m) matrix
ˆ
f
0uu
(u
∗
) is negative semideﬁnite. Furthermore,
if this matrix is negative deﬁnite then x
∗
is a local optimum. the following exercise expresses
f
ˆ
f
0uu
(u
∗
) in terms of derivatives of the functions f
i
.
Exercise: Show that
ˆ
f
0uu
(u
∗
) = [g
u
.
.
.I]
¸
L
ww
L
uw
L
wu
L
uu
g
u
. . .
I
¸
¸
(w
∗
, u
∗
)
where
g
u
(u
∗
) = −[f
w
(x
∗
)]
−1
f
u
(x
∗
), L(x) = f
0
(x) −
m
¸
i=1
λ
∗
i
f
i
(x).
22 CHAPTER 3. OPTIMIZATION WITH EQUALITY CONSTRAINTS
3.3.4 A numerical procedure.
We assume that the derivatives f
ix
(x), 1 ≤ i ≤ m, are linearly independent for all x. Then the
following algorithm is a straightforward adaptation of the procedure in Section 2.4.6.
Step 1. Find x
0
arbitrary so that f
i
(x
0
) = α
i
, 1 ≤ i ≤ m. Set k = 0 and go to Step 2.
Step 2. Find a partition x = (w, u)
2
of the variables such that f
w
(x
k
) is nonsingular. Calculate λ
k
by (λ
k
)
= f
0w
f
−1
w(xk)
, and
ˆ
f
k
0
(u
k
) = −f
u
(x
k
)λ
k
+f
0u
(x
k
). If
ˆ
f
k
0
(u
k
) = 0, stop. Otherwise
go to Step 3.
Step 3. Set ˜ u
k
= u
k
+d
k
ˆ
f
k
0
(u
k
). Find ˜ w
k
such that f
i
( ˜ w
k
, ˜ u
k
) = 0, 1 ≤ i ≤ m. Set
x
k+1
= ( ˜ w
k
, ˜ u
k
), set k = k + 1, and return to Step 2.
Remarks. As before, the step sizes d
k
> 0 can be selected various ways. The practical applicability
of the algorithm depends upon two crucial factors: the ease with which we can ﬁnd a partition
x = (w, u) so that f
w
(x
k
) is nonsingular, thus enabling us to calculate λ
k
; and the ease with which
we can ﬁnd ˜ w
k
so that f( ˜ w
k
, ˜ u
k
) = α. In the next section we apply this algorithm to a practical
problem where these two steps can be carried out without too much difﬁculty.
3.3.5 Design of resistive networks.
Consider a network N with n + 1 nodes and b branches. We choose one of the nodes as datum
and denote by e = (e
1
, . . . , e
n
)
the vector of nodetodatum voltages. Orient the network graph
and let v = (v
1
, . . . , v
b
)
and j = (j
1
, . . . , j
b
)
respectively, denote the vectors of branch voltages
and branch currents. Let A be the n b reduced incidence matrix of the network graph. Then the
Kirchhoff current and voltage laws respectively yield the equations
Aj = 0 and A
e = v (3.30)
Next we suppose that each branch k contains a (possibly nonlinear)resistive element with the form
shown in Figure 3.3, so that
j
k
−j
sk
= g
k
(v
rk
) = g
k
(v
k
−v
sk
), 1 ≤ k ≤ b, (3.31)
where v
rk
is the voltage across the resistor. Here j
sk
, v
sk
are the source current and voltage in the
kth branch, and g
k
is the characteristic of the resistor. Using the obvious vector notation j
s
∈ R
b
,
v
s
∈ R
b
for the sources, v
r
∈ R
b
for the resistor voltages, and g = (g
1
, . . . , g
b
)
, we can rewrite
(3.30) as (3.31):
j −j
s
= g(v −v
s
) = g(v
r
). (3.32)
Although (3.30) implies that the current (j
k
−j
s
k) through the kth resistor depends only on the
voltage v
rk
= (v
k
−v
sk
) across itself, no essential simpliﬁcation is achieved. Hence, in (3.31) we
shall assume that g
k
is a function of v
r
. This allows us to include coupled resistors and voltage
controlled current sources. Furthermore, let us suppose that there are design parameters p =
(p
1
, . . . , p
)
which are under our control, so that (3.31) is replaced by (3.32):
j −j
x
= g(v
r
, p) = g(v−v
s
, p). (3.33)
2
This is just a notational convenience. The w variable may consist of any m components of x.
3.3. REMARKS AND EXTENSIONS 23
o

+ 
+ 
+
o
j
sk
v
rk
v
sk
j
k
−j
sk
j
k
v
k
Figure 3.3: The kth branch.
If we combine (3.29) and (3.32) we obtain (3.33):
Ag(A
e −v
s
, p) = i
s
, (3.34)
where we have deﬁned i
s
= A
js
. The network design problem can then be stated as ﬁnding p, v
s
, i
s
so as to minimize some speciﬁed function f
0
(e, p, v
s
, i
s
). Formally, we have the optimization prob
lem (3.34):
Minimize
subject to
f
0
(e, p, v
s
, i
s
)
Ag(A
e −v
s
, p) −i
s
= 0.
(3.35)
We shall apply the algorithm 3.3.4 to this problem. To do this we make the following assumption.
Assumption: (a) f
0
is differentiable. (b) g is differentiable and the nn matrix A(∂g/∂v)(v, p)A
is nonsingular for all v ∈ R
b
, p ∈ R
. (c) The network N described by (3.33) is determinate i.e.,
for every value of (p, v
s
, i
s
) there is a unique e = E(p, v
s
, i
s
) satisfying (3.33).
In terms of the notation of 3.3.4, if we let x = (e, p, v
s
, i
s
), then assumption (b) allows us to
identify w = e, and u = (p, v
s
, i
s
). Also let f(x) = f(e, p, v
s
, i
s
) = Ag(A
e−v
s
, p) −i
s
. Now the
crucial part in the algorithm is to obtain λ
k
at some point x
k
. To this end let ˜ x = (˜ e, ˜ p, ˜ v
s
,
˜
i
s
) be a
ﬁxed point. Then the corresponding λ =
˜
λ is given by (see (3.27))
˜
λ
= f
0w
(˜ x)f
−1
w
(˜ x) = f
0e
(˜ x)f
−1
e
(˜ x). (3.36)
From the deﬁnition of f we have
f
e
(˜ x) = AG(˜ v
r
, ˜ p)A
,
where ˜ v
r
= A
˜ e − ˜ v
s
, and G(˜ v
r
, ˜ p) = (∂g/∂v
r
)(˜ v
r
, ˜ p). Therefore,
˜
λ is the solution (unique by
assumption (b)) of the following linear equation:
AG
(˜ v
r
, ˜ p)A
˜
λ = f
0e
(˜ x). (3.37)
Now (3.36) has the following extremely interesting physical interpretation. If we compare (3.33)
with (3.36) we see immediately that
˜
λ is the nodetodatum response voltages of a linear network
N(˜ v
r
, ˜ p) driven by the current sources f
0e
(˜ x). Furthermore, this network has the same graph as
the original network (since they have the same incidence matrix); moreover, its branch admittance
matrix, G
(˜ v
r
, ˜ p), is the transpose of the incremental branch admittance matrix (evaluated at (˜ v
r
, ˜ p))
of the original network N. For this reason, N(˜ v
r
, ˜ p) is called the adjoint network (of N) at (˜ v
r
, ˜ p).
24 CHAPTER 3. OPTIMIZATION WITH EQUALITY CONSTRAINTS
Once we have obtained
˜
λ we can obtain
u
ˆ
f
0
(˜ u) using (3.28). Elementary calculations yield
(3.37):
u
ˆ
f
0
(˜ u) =
ˆ
f
0p
(˜ u)
ˆ
f
0vs
(˜ u)
ˆ
f
0is
(˜ u)
¸
¸
¸ =
[
∂g
∂p
(˜ v
r
, ˜ p)]
A
G
(˜ v
r
, ˜ p)A
−I
¸
¸ ˜
λ +
f
0p
(˜ x)
f
0vs
(˜ x)
f
0is
(˜ x)
¸
¸
(3.38)
We can now state the algorithm.
Step 1. Select u
0
= (p
0
, v
0
s
, i
0
s
) arbitrary. Solve (3.33) to obtain e
0
= E(p
0
, v
0
s
, i
0
s
). Let k = 0 and
go to Step 2.
Step 2. Calculate v
k
r
= A
e
k
−v
k
s
. calculate f
0e
(x
k
). Calculate the nodetodatum response λ
k
of
the adjoint network N(v
k
r
, p
k
) driven by the current source f
0e
(x
k
). Calculate
u
ˆ
f
0
(u
k
) from
(3.37). If this gradient is zero, stop. Otherwise go to Step 3.
Step 3. Let u
k+1
= (p
k+1
, v
k+1
s
, i
k+1
s
) = u
k
−d
k
u
ˆ
f
0
(u
k
), where d
k
> 0 is a predetermined
step size.
3
Solve (3.33) to obtain e
k+1
= (Ep
k+1
, v
k+1
s
, i
k+1
s
). Set k = k +1 and return to Step 2.
Remark 1. Each iteration from u
k
to u
k+1
requires one linear network analysis step (the
computation of λ
k
in Step 2), and one nonlinear network analysis step (the computation of e
k+1
in
step 3). This latter step may be very complex.
Remark 2. In practice we can control only some of the components of v
s
and i
s
, the rest being
ﬁxed. The only change this requires in the algorithm is that in Step 3 we set
p
k+1
= p
k
−d
k
ˆ
f
0p
(u
k
) just as before, where as v
k+1
sj
= v
k
sj
−d
k
(∂
ˆ
f
0
/∂v
sj
)(u
k
) and
i
k+1
sm
= i
k
sm
−d
k
(∂
ˆ
f
0
/∂i
sm
)(u
k
) with j and m ranging only over the controllable components and
the rest of the components equal to their speciﬁed values.
Remark 3. The interpretation of λ as the response of the adjoint network has been exploited for
particular function f
0
in a series of papers (director and Rohrer [1969a], [1969b], [1969c]). Their
derivation of the adjoint network does not appear as transparent as the one given here. Although
we have used the incidence matrix A to obtain our network equation (3.33), one can use a more
general cutset matrix. Similarly, more general representations of the resistive elements may be
employed. In every case the “adjoint” network arises from a network interpretation of (3.27),
[f
w
(˜ x)]
˜
λ = f
0w
(˜ x),
with the transpose of the matrix giving rise to the adjective “adjoint.”
Exercise: [DC biasing of transistor circuits (see Dowell and Rohrer [1971]).] Let N be a transistor
circuit, and let (3.33) model the dc behavior of this circuit. Suppose that i
s
is ﬁxed, v
sj
for j ∈ J
are variable, and v
sj
for j / ∈ J are ﬁxed. For each choice of v
sj
, j ∈ J, we obtain the vector e and
hence the branch voltage vector v = A
e. Some of the components v
t
, t ∈ T, will correspond to
bias voltages for the transistors in the network, and we wish to choose v
sj
, j ∈ J, so that v
t
is as
close as possible to a desired bias voltage v
d
t
, t ∈ T. If we choose nonnegative numbers α
t
, with
relative magnitudes reﬂecting the importance of the different transistors then we can formulate the
criterion
3
Note the minus sign in the expression u
k
−d
k
u
ˆ
f0(u
k
). Remember we are minimizing f0, which is equivalent to
maximizing (−f0).
3.3. REMARKS AND EXTENSIONS 25
f
0
(e) =
¸
t∈T
α
t
[v
t
−v
d
t
[
2
.
(i) Specialize the algorithm above for this particular case.
(ii) How do the formulas change if the network equations are written using an arbitrary cutset matrix
instead of the incidence matrix?
26 CHAPTER 3. OPTIMIZATION WITH EQUALITY CONSTRAINTS
Chapter 4
OPTIMIZATION OVER SETS
DEFINED BY INEQUALITY
CONSTRAINTS: LINEAR
PROGRAMMING
In the ﬁrst section we study in detail Example 2 of Chapter I, and then we deﬁne the general linear
programming problem. In the second section we present the duality theory for linear program
ming and use it to obtain some sensitivity results. In Section 3 we present the Simplex algorithm
which is the main procedure used to solve linear programming problems. In section 4 we apply
the results of Sections 2 and 3 to study the linear programming theory of competitive economy.
Additional miscellaneous comments are collected in the last section. For a detailed and readily ac
cessible treatment of the material presented in this chapter see the companion volume in this Series
(Sakarovitch [1971]).
4.1 The Linear Programming Problem
4.1.1 Example.
Recall Example 2 of Chapter I. Let g and u respectively be the number of graduate and undergradu
ate students admitted. Then the number of seminars demanded per year is
2g+u
20
, and the number of
lecture courses demanded per year is
5g+7u
40
. On the supply side of our accounting, the faculty can
offer 2(750) + 3(250) = 2250 seminars and 6(750) + 3(250) = 5250 lecture courses. Because of
his contractual agreements, the President must satisfy
2g+u
20
≤ 2250 or 2g +u ≤ 45, 000
and
5g+7u
40
≤ 5250 or 5g + 7u ≤ 210, 000 .
27
28 CHAPTER 4. LINEAR PROGRAMMING
Since negative g or u is meaningless, there are also the constraints g ≥ 0, u ≥ 0. Formally then the
President faces the following decision problem:
Maximize αg +βu
subject to 2g +u ≤ 45, 000
5g + 7u ≤ 210, 000
g ≥ 0, u ≥ 0 .
(4.1)
It is convenient to use a more general notation. So let x = (g, u)
, c = (α, β)
, b = (45000, 210000, 0, 0)
and let A be the 42 matrix
A =
2
5
−1
0
1
7
0
−1
¸
¸
¸
¸
.
Then (4.1) can be rewritten as (4.2)
1
Maximize c
x
subject to Ax ≤ b .
(4.2)
Let A
i
, 1 ≤ i ≤ 4, denote the rows of A. Then the set Ω of all vectors x which satisfy the constraints
in (4.2) is given by Ω = ¦x[A
i
x ≤ b
i
, 1 ≤ i ≤ 4¦ and is the polygon OPQR in Figure 4.1.
For each choice x, the President receives the payoff c
x. Therefore, the surface of constant payoff
k say, is the hyperplane π(k) = ¦x[c
x = k¦. These hyperplanes for different values of k are
parallel to one another since they have the same normal c. Furthermore, as k increases π(k) moves
in the direction c. (Obviously we are assuming in this discussion that c = 0.) Evidently an optimal
decision is any point x
∗
∈ Ω which lies on a hyperplane π(k) which is farthest along the direction
c. We can rephrase this by saying that x∗ ∈ Ω is an optimal decision if and only if the plane π
∗
through x
∗
does not intersect the interior of Ω, and futhermore at x
∗
the direction c points away
from Ω. From this condition we can immediately draw two very important conclusions: (i) at least
one of the vertices of Ω is an optimal decision, and (ii) x
∗
yields a higher payoff than all points
in the cone K
∗
consisting of all rays starting at x
∗
and passing through Ω, since K
∗
lies “below”
π
∗
. The ﬁrst conclusion is the foundation of the powerful Simplex algorithm which we present in
Section 3. Here we pursue consequences of the second conclusion. For the situation depicted in
Figure 4.1 we can see that x
∗
= Q is an optimal decision and the cone K
∗
is shown in Figure 4.2.
Now x
∗
satisﬁes A
x
x
∗
= b
1
, A
2
x
∗
= b
2
, and A
3
x
∗
< b
3
, A
4
x
∗
< b
4
, so that K
∗
is given by
K
∗
= ¦x
∗
+h[A
1
h ≤ 0 , A
2
h ≤ 0¦ .
Since c
x
∗
≥ c
y for all y ∈ K
∗
we conclude that
c
h ≤ 0 for all h such that A
1
h ≤ 0, A
2
h ≤ 0 . (4.3)
We pause to formulate the generalization of (4.3) as an exercise.
1
Recall the notation introduced in 1.1.2, so that x ≤ y means xi ≤ yi for all i.
4.1. THE LINEAR PROGRAMMING PROBLEM 29
,










x
2
π(k) = ¦x[c
x = k¦
π
∗
Q = x
∗
direction of
increasing
payoff k
¦x[A
2
x = b
2
¦
x
1
¦x[A
1
x = b
1
¦
R
A
4
O A
3
A
1
⊥ QR
c ⊥ π
∗
A
2
⊥ PQ
P
Figure 4.1: Ω = OPQR.
Exercise 1: Let A
i
, 1 ≤ i ≤ k, be ndimensional row vectors. Let c ∈ R
n
, and let b
i
, 1 ≤ i ≤ k,
be real numbers. Consider the problem
Maximize c
x
subject to A
i
x ≤ b
i
, 1 ≤ i ≤ k .
For any x satisfying the constraints, let I(x) ⊂ ¦1, . . . , n¦ be such that A
i
(x) = b
i
, i ∈ I(x), A
i
x <
b
i
, i / ∈ I(x). Suppose x
∗
satisﬁes the constraints. Show that x
∗
is optimal if an only if
c
h ≤ 0 for all h such that A
i
h ≤ 0 , i ∈ I(x
∗
).
Returning to our problem, it is clear that (4.3) is satisﬁed as long as c lies between A
1
and A
2
.
Mathematically this means that (4.3) is satisﬁed if and only if there exist λ
∗
1
≥ 0, λ
∗
2
≥ 0 such that
2
c
= λ
∗
1
, A
1
+λ
∗
2
A
2
. (4.4)
As c varies, the optimal decision will change. We can see from our analysis that the situation is as
follows (see Figure 4.1):
2
Although this statement is intuitively obvious, its generalization to n dimensions is a deep theorem known as Farkas’
lemma (see Section 2).
30 CHAPTER 4. LINEAR PROGRAMMING
P
x
∗
= Q
K
∗
π
∗
R
A
4
O
A
3
A
2
c
A
1
Figure 4.2: K
∗
is the cone generated by Ω at x
∗
.
1. x
∗
= Qis optimal iff c lies between A
1
and A
2
iff c
= λ
∗
1
A
1
+λ
∗
2
A
2
for some λ
∗
1
≥ 0, λ
∗
2
≥
0,
2. x
∗
∈ QP is optimal iff c lies along A
2
iff c
= λ
∗
2
A
2
for some λ
∗
2
≥ 0,
3. x
∗
= P is optimal iff c lies between A
3
and A
2
iff c
= λ
∗
2
A
2
+λ
∗
3
A
3
for some λ
∗
2
≥ 0, λ
∗
3
≥
0, etc.
These statements can be made in a more elegant way as follows:
x
∗
∈ Ω is optimal iff there exists λ
∗
i
≥ 0 , 1 ≤ i ≤ 4, such that
(a) c
=
4
¸
i=1
λ
∗
i
a
i
, (b) if A
i
x
∗
< b
i
then λ
∗
i
= 0 . (4.5)
For purposes of application it is useful to separate those constraints which are of the form x
i
≥ 0,
from the rest, and to reformulate (4.5) accordingly We leave this as an exercise.
Exercise 2: Show that (4.5) is equivalent to (4.6), below. (Here A
i
= (a
i1
, a
i2
).) x
∗
∈ Ω is optimal
iff there exist λ
∗
1
≥ 0 , λ
∗
2
≥ 0 such that
(a) c
i
≤ λ
∗
1
a
1i
+λ
∗
2
a
2i
, i = 1, 2,
(b) if a
j1
x
∗
1
+a
j2
x
∗
2
< b
j
then x
∗
j
= 0, j = 1, 2.
(c) if c
i
< λ
∗
1i
+λ
∗
2
a
2i
then x
∗
i
= 0, i = 1, 2.
(4.6)
4.1. THE LINEAR PROGRAMMING PROBLEM 31
4.1.2 Problem formulation.
A linear programming problem (or LP in brief) is any decision problem of the form 4.7.
Maximize c
1
x
1
+c
2
x
2
+. . . +c
n
x
n
subject to
a
il
x
1
+a
i2
x
2
+. . . +a
in
x
n
≤ b
i
, l ≤ i ≤ k ,
a
il
x
1
+. . . . . . . . . +a
in
x
n
≥ b
i
, k + 1 ≤ i ≤ ,
a
il
x
1
+. . . . . . . . . +a
in
x
n
= b
i
, + 1 ≤ i ≤ m ,
and
x
j
≥ 0 , 1 ≤ j ≤ p ,
x
j
≥ 0 , p + 1 ≤ j ≤ q;
x
j
arbitary , q + 1 ≤ j ≤ n ,
(4.7)
where the c
j
, a
ij
, b
i
are ﬁxed real numbers.
There are two important special cases:
Case I: (4.7) is of the form (4.8):
Maximize
n
¸
j=1
c
j
x
j
subject to
n
¸
j=1
a
ij
x
j
≤ b
i
,
x
j
≥ 0 ,
1 ≤ i ≤ m ,
1 ≤ j ≤ n
(4.8)
Case II: (4.7) is of the form (4.9):
Maximize
n
¸
j=1
c
j
x
j
subject to
n
¸
j=1
a
ij
x
j
= b
i
,
x
j
≥ 0 ,
1 ≤ i ≤ m ,
1 ≤ j ≤ n .
(4.9)
Although (4.7) appears to be more general than (4.8) and (4.9), such is not the case.
Proposition: Every LP of the form (4.7) can be transformed into an equivalent LP of the form (4.8).
Proof.
Step 1: Replace each inequality constraint
¸
a
ij
x
j
≥ b
i
by
¸
(−a
ij
)x
j
≤ (−b
i
).
Step 2: Replace each equality constraint
¸
a
ij
x
j
= b
i
by two inequality constraints:
¸
a
ij
x
j
≤ b
i
,
¸
(−a
ij
)x
j
≤ (−b
i
).
Step 3: Replace each variable x
j
which is constrained x
j
≤ 0 by a variable y
j
= −x
j
constrained
y
j
≥ 0 and then replace a
ij
x
j
by (−a
ij
)y
j
for every i and c
j
x
j
by (−c
j
)y
j
.
32 CHAPTER 4. LINEAR PROGRAMMING
Step 4: Replace each variable x
j
which is not constrained in sign by a pair of variables
y
j
−z
j
= x
j
constrained y
j
≥ 0, z
j
≥ 0 and then replace a
ij
x
j
by a
ij
y
j
+ (−a
ij
)z
j
for every i and
c
j
x
j
by c
j
y
j
+ (−c
j
)z
j
. Evidently the resulting LP has the form (4.8) and is equivalent to the
original one. ♦
Proposition: Every LP of the form (4.7) can be transformed into an equivalent LP of the from (4.9)
Proof.
Step 1: Replace each inequality constraint
¸
a
ij
x
j
≤ b
i
by the equality constraint
¸
a
ij
x
j
+y
i
= b
i
where y
i
is an additional variable constrained y
i
≥ 0.
Step 2: Replace each inequality constraint
¸
a
ij
x
j
≥ b
i
by the equality constraint
¸
a
ij
x
j
−y
i
= b
i
where y
i
is an additional variable constrained by y
i
≥ 0. (The new variables
added in these steps are called slack variables.)
Step 3, Step 4: Repeat these steps from the previous proposition. Evidently the new LP has the
form (4.9) and is equivalent to the original one. ♦
4.2 Qualitative Theory of Linear Programming
4.2.1 Main results.
We begin by quoting a fundamental result. For a proof the reader is referred to (Mangasarian
[1969]).
Farkas’ Lemma. Let A
i
, 1 ≤ i ≤ k, be ndimensional row vectors. Let c ∈ R
n
be a column vector.
The following statements are equivalent:
(i) for all x ∈ R
n
, A
i
x ≤ 0 for 1 ≤ i ≤ k implies c
x ≤ 0,
(ii) there exists λ
1
≥ 0, . . . , λ
k
≥ 0 such that c
=
k
¸
i=1
λ
i
A
i
.
An algebraic version of this result is sometimes more convenient.
Farkas’ Lemma (algebraic version). Let Abe a kn matrix. Let c ∈ R
n
. The following statements
are equivalent.
(i) for all x ∈ R
n
, Ax ≤ 0 implies c
x ≤ 0,
(ii) there exists λ ≥ 0, λ ∈ R
k
, such that A
λ = c.
Using this result it is possible to derive the main results following the intuitive reasoning of (4.1).
We leave this development as two exercises and follow a more elegant but less intuitive approach.
Exercise 1: With the same hypothesis and notation of Exercise 1 in 4.1, use the ﬁrst version of
Farkas
lemma to show that there exist λ
∗
i
≥ 0 for i ∈ I(x
∗
) such that
¸
i∈I(x
∗
)
λ
∗
i
A
i
= c
.
Exercise 2: Let x
∗
satisfy the constraints for problem (4.17). Use the previous exercise to show
that x
∗
is optimal iff there exist λ
∗
1
≥ 0, . . . , λ
∗
m
≥ 0 such that
(a) c
j
≤
m
¸
i=1
λ
∗
i
a
ij
, 1 ≤ j ≤ n
(b) if
n
¸
j=1
a
ij
x
∗
j
< b
i
then λ
∗
i
= 0 , 1 ≤ i ≤ m (c) if
m
¸
i=1
λ
∗
i
a
ij
> c
j
then x
∗
j
= 0 , 1 ≤ j ≤ m.
In the remaining discussion, c ∈ R
n
, b ∈
n
are ﬁxed vectors, and A = ¦a
ij
¦ is a ﬁxed m n
matrix, whereas x ∈ R
n
and λ ∈ R
m
will be variable. Consider the pair of LPs (4.10) and (4.11)
4.2. QUALITATIVE THEORY OF LINEAR PROGRAMMING 33
below. (4.10) is called the primal problem and (4.11) is called the dual problem.
Maximize
subject to
c
1
x
1
+. . . +c
n
x
n
a
i1
x
1
+. . . +a
in
x
n
≤ b
i
,
x
j
≥ 0 ,
1 ≤ i ≤ m
1 ≤ j ≤ n .
(4.10)
Maximize
subject to
λ
1
b
1
+. . . +λ
m
b
m
λ
1
a
1j
+. . . +λ
m
a
mj
≥ c
j
,
λ
i
≥ 0 ,
1 ≤ j ≤ n
1 ≤ i ≤ m .
(4.11)
Deﬁnition: Let Ω
p
= ¦x ∈ R
n
[Ax ≤ b, x ≥ 0¦ be the set of all points satisfying the constraints
of the primal problem. Similarly let Ω
d
= ¦λ ∈ R
m
[λ
A ≥ c
, λ ≥ 0¦. A point x ∈ Ω
p
(λ ∈ Ω
d
) is
said to be a feasible solution or feasible decision for the primal (dual).
The next result is trivial.
Lemma 1: (Weak duality) Let x ∈ Ω
p
, λ ∈ Ω
d
. Then
c
x ≤ λ
Ax ≤ λ
b. (4.12)
Proof: x ≥ 0 and λ
A − c
≥ 0 implies (λ
A−c
)x ≥ 0 giving the ﬁrst inequality. b−Ax ≥ 0 and
λ
≥ 0 implies λ
(b−Ax) ≥ 0 giving the second inequality. ♦
Corollary 1: If x
∗
∈ Ω and λ
∗
∈ Ω
d
such that c
x
∗
= (λ
∗
)
b, then x
∗
is optimal for (4.10) and λ
∗
is
optimal for (4.11).
Theorem 1: (Strong duality) Suppose Ω
p
= φ and Ω
d
= φ. Then there exists x
∗
which is optimum
for (4.10) and λ
∗
which is optimum for (4.11). Furthermore, c
x
∗
= (λ
∗
)
b.
Proof: Because of the Corollary 1 it is enough to prove the last statement, i.e., we must show that
there exist x ≥ 0, λ ≥ 0, such that Ax ≤ b, A
λ ≥ c and b
λ−c
x ≤ 0. By introducing slack
variables y ∈ R
m
, µ ∈ R
m
, r ∈ R, this is equivalent to the existence of x ≥ 0, y ≥ 0, λ ≥ 0, µ ≤
0, r ≤ 0 such that
A I
m
A
−I
n
−c
b
1
¸
¸
x
y
λ
µ
r
¸
¸
¸
¸
¸
¸
=
b
c
0
¸
¸
By the algebraic version of Farkas’ Lemma, this is possible only if
A
ξ −cθ ≤ 0 , ξ ≤ 0 ,
Aw = bθ ≤ 0 , −w ≤ 0 ,
θ ≤ 0
(4.13)
implies
b
ξ +c
w ≤ 0. (4.14)
34 CHAPTER 4. LINEAR PROGRAMMING
Case (i): Suppose (w, ξ, θ) satisﬁes (4.13) and θ < 0. Then (ξ/θ) ∈ Ω
d
, (w/−θ) ∈ Ω
p
, so that by
Lemma 1 c
w/(−θ) ≤ b
ξ/θ, which is equivalent to (4.14) since θ < 0.
Case (ii): Suppose (w, ξ, θ) satisﬁes (4.13) and θ = 0, so that −A
ξ ≥ 0, −ξ ≥ 0, Aw ≤ 0, w ≥ 0.
By hypothesis, there exist x ∈ Ω
p
, λ ∈ Ω
d
. Hence, −b
ξ = b
(−ξ) ≥ (Ax)
(−ξ) = x
(−A
ξ) ≥ 0,
and c
w ≤ (A
λ)
w = λ
(Aw) ≤ 0. So that b
ξ +c
w ≤ 0. ♦
The existence part of the above result can be strengthened.
Theorem 2: (i) Suppose Ω
p
= φ. Then there exists an optimum decision for the primal LP iff
Ω
d
= φ.
(ii) Suppose Ω
d
= φ. Then there exists an optimum decision for the dual LP iff Ω
p
= φ.
Proof Because of the symmetry of the primal and dual it is enough to prove only (i). The
sufﬁciency part of (i) follows from Theorem 1, so that only the necessity remains. Suppose, in
contradiction, that Ω
d
= φ. We will show that sup ¦c
x[x ∈ Ω
p
¦ = +∞. Now, Ω
d
= φ means
there does not exist λ ≥ 0 such that A
λ ≥ c. Equivalently, there does not exist λ ≥ 0, µ ≤ 0 such
that
¸
A
[
[
−I
n
λ
−−−
µ
¸
¸
=
c
By Farkas’ Lemma there exists w ∈ R
n
such that Aw ≤ 0, −w ≤ 0, and c
w > 0. By hypothesis,
Ω
p
= φ, so there exists x ≥ 0 such that Ax ≤ b. but then for any θ > 0, A(x + θw) ≤ b,
(x + θw) ≥ 0, so that (x + θw) ∈ Ω
p
. Also, c
(x + θw) = c
x + θc
w. Evidently then, sup
¦c
x[x ∈ Ω
p
¦ = +∞so that there is no optimal decision for the primal. ♦
Remark: In Theorem 2(i), the hypothesis that Ω
p
= φ is essential. Consider the following exercise.
Exercise 3: Exhibit a pair of primal and dual problems such that neither has a feasible solution.
Theorem 3: (Optimality condition) x
∗
∈ Ω
p
is optimal if and only if there exists λ
∗
∈ Ω
d
such that
m
¸
j=1
a
ij
x
∗
j
< b
i
implies λ
∗
i
= 0 ,
and
m
¸
i=1
λ
∗
i
a
ij
< c
j
implies x
∗
j
= 0 .
(4.15)
((4.15) is known as the condition of complementary slackness.)
Proof: First of all we note that for x
∗
∈ Ω
p
, λ
∗
∈ Ω
d
, (4.15) is equivalent to (4.16):
(λ
∗
)
(Ax
∗
−b) = 0, and (A
λ
∗
−c)
x
∗
= 0 . (4.16)
Necessity. Suppose x
∗
∈ Ω
p
is optimal. Then from Theorem 2, Ω
d
= φ, so that by Theorem 1
there exists λ
∗
∈ Ω
d
such that c
x
∗
= (λ
∗
)
b. By Lemma 1 we always have
c
x
∗
≤ (λ
∗
)
Ax
∗
≤ (λ
∗
)
b so that we must have c
x
∗
= (λ
∗
)
Ax
∗
= (λ
∗
)
b. But (4.16) is just an
equivalent rearrangement of these two equalities.
Sufﬁciency. Suppose (4.16) holds for some x
∗
∈ Ω
p
, λ
∗
∈ Ω
d
. The ﬁrst equality in (4.16) yields
(λ
∗
)
b = (λ
∗
)
Ax
∗
= (A
λ
∗
)
x
∗
, while the second yields (A
λ
∗
)
x
∗
= c
x
∗
, so that c
x
∗
= (λ
∗
)
b.
By Corollary 1, x
∗
is optimal. ♦
4.2. QUALITATIVE THEORY OF LINEAR PROGRAMMING 35
The conditions x
∗
∈ Ω
p
, x
∗
∈ Ω
d
in Theorem 3 can be replaced by the weaker x
∗
≥ 0, λ
∗
≥ 0
provided we strengthen (4.15) as in the following result, whose proof is left as an exercise.
Theorem 4: (Saddle point) x
∗
≥ 0 is optimal for the primal if and only if there exists λ
∗
≥ 0 such
that
L(x, λ
∗
) ≤ L(x
∗
, λ
∗
) ≤ L(x
∗
, λ) for all x ≥ 0, and allλ ≥ 0, (4.17)
where L: R
n
xR
m
→ R is deﬁned by
L(x, λ) = c
x −λ
(Ax −b) (4.18)
Exercise 4: Prove Theorem 4.
Remark. The function L is called the Lagrangian. A pair (x
∗
, λ
∗
) satisfying (4.17) is said to form
a saddlepoint of L over the set ¦x[x ∈ R
n
, x ≥ 0¦ ¦λ[λ ∈ R
m
, λ ≥ 0¦.
4.2.2 Results for problem (4.9).
It is possible to derive analogous results for LPs of the form (4.9). We state these results as exercises,
indicating how to use the results already obtained. We begin with a pair of LPs:
Maximize
subject to
c
1
x
1
+. . . +c
n
x
n
a
il
x
1
+. . . +a
in
x
n
= b
i
,
x
j
≥ 0 ,
1 ≤ i ≤ m ,
1 ≤ j ≤ n .
(4.19)
Minimize
subject to
λ
1
b
1
+. . . +λ
m
b
m
λ
1
a
1j
+. . . +λ
m
a
mj
≥ c
j
, 1 ≤ j ≤ n .
(4.20)
Note that in (4.20) the λ
i
are unrestricted in sign. Again (4.19) is called the primal and (4.20) the
dual. We let Ω
p
, Ω
d
denote the set of all x, λ satisfying the constraints of (4.19), (4.20) respectively.
Exercise 5: Prove Theorems 1 and 2 with Ω
p
and Ω
d
interpreted as above. (Hint. Replace (4.19)
by the equivalent LP: maximize c
x, subject to Ax ≤ b, (−A)x ≤ (−b), x ≥ 0. This is now of the
form (4.10). Apply Theorems 1 and 2.)
Exercise 6: Show that x
∗
∈ Ω
p
is optimal iff there exists λ
∗
∈ Ω
d
such that
x
∗
j
> 0 implies
m
¸
i=1
λ
∗
i
a
ij
= c
j
.
Exercise 7: x
∗
≥ 0 is optimal iff there exists λ
∗
∈ R
m
such that
L(x, λ
∗
) ≤ L(x
∗
, λ
∗
) ≤ L(x
∗
, λ) for all x ≥ 0, λ ∈ R
m
.
where L is deﬁned in (4.18). (Note that, unlike (4.17), λ is not restricted in sign.)
Exercise 8: Formulate a dual for (4.7), and obtain the result analogous to Exercise 5.
36 CHAPTER 4. LINEAR PROGRAMMING
4.2.3 Sensitivity analysis.
We investigate how the maximum value of (4.10) or (4.19) changes as the vectors b and c change.
The matrix Awill remain ﬁxed. Let Ω
p
and Ω
d
be the sets of feasible solutions for the pair (4.10) and
(4.11) or for the pair (4.19) and (4.20). We write Ω
p
(b) and Ω
d
(c) to denote the explicit dependence
on b and c respectively. Let B = ¦b ∈ R
m
[Ω
p
(b) = φ¦ and C = ¦c ∈ R
n
[Ω
d
(c) = φ¦, and for
(b, c) ∈ B C deﬁne
M(b, c) = max ¦c
x[x ∈ Ω
p
(b)¦ = min ¦λ
b[λ ∈ Ω
d
(c)¦ . (4.21)
For 1 ≤ i ≤ m, ε ∈ R, b ∈ R
m
denote
b(i, ε) = (b
1
, b
2
, . . . , b
i−1
, b
i
+ε, b
i+1
, . . . , b
m
)
,
and for 1 ≤ j ≤ n, ε ∈ R, c ∈ R
n
denote
c(j, ε) = (c
1
, c
2
, . . . , c
j−1
, c
j
+ε, c
j+1
, . . . , c
n
)
.
We deﬁne in the usual way the right and left hand partial derivatives of M at a point (
ˆ
b, ˆ c) ∈ BC
as follows:
∂M
+
∂b
i
(
ˆ
b, ˆ c) = lim
ε → 0
ε > 0
1
ε
¦M(
ˆ
b(i, ε), ˆ c) −M(
ˆ
b, ˆ c)¦ ,
∂M
−
∂b
i
(
ˆ
b, ˆ c) = lim
ε → 0
ε > 0
1
ε
¦M(
ˆ
b, ˆ c) −M(
ˆ
b(i, −ε), ˆ c)¦ ,
∂M
+
∂c
j
(
ˆ
b, ˆ c) = lim
ε → 0
ε > 0
1
ε
¦M(
ˆ
b, ˆ c(j, ε)) −M(
ˆ
b, ˆ c¦ ,
∂M
−
∂c
j
(
ˆ
b, ˆ c) = lim
ε → 0
ε > 0
1
ε
¦M(
ˆ
b, ˆ c −M(
ˆ
b, ˆ c(j, −ε))¦ ,
Let
◦
B,
◦
C denote the interiors of B, C respectively.
Theorem 5: At each (
ˆ
b, ˆ c) ∈
◦
B
◦
C, the partial derivatives given above exist. Furthermore, if
ˆ x ∈ Ω
p
(
ˆ
b),
ˆ
λ ∈ Ω
d
(ˆ c) are optimal, then
∂M
+
∂b
i
(
ˆ
b, ˆ c) ≤
ˆ
λ
i
≤
∂M
−
∂b
i
(
ˆ
b, ˆ c) , 1 ≤ i ≤ m , (4.22)
4.3. THE SIMPLEX ALGORITHM 37
∂M
+
∂c
j
(
ˆ
b, ˆ c) ≥ ˆ x
j
≥
∂M
−
∂c
j
(
ˆ
b, ˆ c) , 1 ≤ j ≤ n , (4.23)
Proof: We ﬁrst show (4.22), (4.23) assuming that the partial derivatives exist. By strong duality
M(
ˆ
b, ˆ c) =
ˆ
λ
ˆ
b, and by weak duality M(
ˆ
b(i, ε), ˆ c) ≤
ˆ
λ
ˆ
b(i, ε), so that
1
ε
¦M(
ˆ
b(i, ε), ˆ c) −M(
ˆ
b, ˆ c)¦ ≤
1
ε
ˆ
λ
¦
ˆ
b(i, ε) −
ˆ
b¦
ˆ
λ
i
, for ε > 0,
1
ε
¦M(
ˆ
b, ˆ c) −M(
ˆ
b(i, −ε), ˆ c)¦ ≥
1
ε
ˆ
λ
¦
ˆ
b −
ˆ
b(i, −ε)¦ =
ˆ
λ
i
, for ε > 0.
Taking limits as ε → 0, ε > 0, gives (4.22).
On the other hand, M(
ˆ
b, ˆ c) = ˆ c
ˆ x, and M(
ˆ
b, ˆ c(j, ε)) ≥ (ˆ c(j, ε))
ˆ x, so that
1
ε
¦M(
ˆ
b, ˆ c(j, ε)) −M(
ˆ
b, ˆ c)¦ ≥
1
ε
¦ˆ c(j, ε)
− ˆ c¦
ˆ x = ˆ x
j
, for ε > 0,
1
ε
¦M(
ˆ
b, ˆ c) −M(
ˆ
b, ˆ c(j, −ε))¦ ≤
1
ε
¦ˆ c − ˆ c(j, −ε)¦
ˆ x = ˆ x
j
, for ε > 0,
which give (4.23) as ε → 0, ε > 0.
Finally, the existence of the right and left partial derivatives follows from Exercises 8, 9 below. ♦
We recall some fundamental deﬁnitions from convex analysis.
Deﬁnition: X ⊂ R
n
is said to be convex if x, y ∈ X and 0 ≤ θ ≤ 1 implies (θx+(1−θ)y) ∈ X.
Deﬁnition: Let X ⊂ R
n
and f : X → R. (i) f is said to be convex if X is convex, and x, y ∈ X,
0 ≤ θ ≤ 1 implies f(θx + (1 − θ)y) ≤ θf(x) + (1 − θ)f(y). (ii) f is said to be concave if −f is
convex, i.e., x, y ∈ X, 0 ≤ θ ≤ 1 implies f(θx + (1 −θ)y) ≥ θf(x) + (1 −θ)f(y).
Exercise 8: (a) Show that Ω
p
, Ω
d
, and the sets B ⊂ R
m
, C ⊂ R
n
deﬁned above are convex sets.
(b) Show that for ﬁxed c ∈ C, M(, c) : B → R is concave and for ﬁxed b ∈ B, M(b, ) : C → R
is convex.
Exercise 9: Let X ⊂ R
n
, and f : X → R be convex. Show that at each point ˆ x in the interior of
X, the left and right hand partial derivatives of f exist. (Hint: First show that for
ε
2
> ε
1
> 0 > δ
1
> δ
2
,(1/ε
2
)¦f(ˆ x(i, ε
2
)) −f(ˆ x)¦ ≥ (1/ε
1
)¦f(ˆ x(i, ε
1
)) −f(ˆ x))¦ ≥
(1/δ
1
)¦f(ˆ x(i, δ
1
)) −f(ˆ x))¦ ≥ (1/δ
2
)¦f(ˆ x(i, δ
2
)) −f(ˆ x)¦. Then the result follows immediately.)
Remark 1: Clearly if (∂M/∂b
i
)(
ˆ
b) exists, then we have equality in (4.22), and then this result
compares with 3.14).
Remark 2: We can also show without difﬁculty that M(, c) and M(b, ) are piecewise linear (more
accurately, linear plus constant) functions on B and C respectively. This is useful in some
computational problems.
Remark 3: The variables of the dual problem are called Lagrange variables or dual variables or
shadowprices. The reason behind the last name will be clear in Section 4.
4.3 The Simplex Algorithm
4.3.1 Preliminaries
We now present the celebrated Simplex algorithm for ﬁnding an optimum solution to any LP of the
form (4.24):
Maximize
subject to
c
1
x
1
+. . . +c
n
x
n
a
il
x
1
+. . . +a
in
x
n
= b
i
,
x
j
≥ 0 ,
1 ≤ i ≤ m
1 ≤ j ≤ n .
(4.24)
38 CHAPTER 4. LINEAR PROGRAMMING
As mentioned in 4.1 the algorithm rests upon the observations that if an optimal exists, then at least
one vertex of the feasible set Ω
p
is an optimal solution. Since Ω
p
has only ﬁnitely many vertices (see
Corollary 1 below), we only have to investigate a ﬁnite set. The practicability of this investigation
depends on the ease with which we can characterize the vertices of Ω
p
. This is done in Lemma 1.
In the following we let A
j
denote the jth column of A, i.e., A
j
= (a
1j
, . . . , a
mj
)
. We begin with
a precise deﬁnition of a vertex.
Deﬁnition: x ∈ Ω
p
is said to be a vertex of Ω
p
if x = λy + (1 −λ)z, with y, z in Ω
p
and
0 < λ < 1, implies x = y = z.
Deﬁnition: For x ∈ Ω
p
, let I(x) = ¦j[x
j
> 0¦.
Lemma 1: Let x ∈ Ω
p
. Then x is a vertex of Ω
p
iff ¦A
j
[j ∈ I(x)¦ is a linearly independent set.
Exercise 1: Prove Lemma 1.
Corollary 1: Ω
p
has at most
m
¸
j=1
n!
(n −j)!
vertices.
Lemma 2: Let x
∗
be an optimal decision of (4.24). Then there is a vertex z
∗
of Ω
p
which is optimal.
Proof: If ¦A
j
[j ∈ I(x
∗
)¦ is linearly independent, let z
∗
= x
∗
and we are done. Hence suppose
¦A
j
[j ∈ I(x
∗
)¦ is linearly dependent so that there exist γ
j
, not all zero, such that
¸
j∈I(x
∗
)
γ
j
A
j
= 0 .
For θ ∈ R deﬁne z(θ) ∈ R
n
by
z
j
(θ) =
x
∗
j
= θγ
j
,
x
∗
j
= 0 ,
j ∈ I(x
∗
)
j ∈ I(x
∗
) .
Az(θ) =
¸
j∈I(x
∗
)
z
j
(θ)A
j
=
¸
j∈I(x
∗
)
x
∗
j
A
j
+θ
¸
j∈I(x
∗
)
γ
j
A
j
= b +θ 0 = b .
Since x
∗
j
> 0 for j ∈ I(x
∗
), it follows that z(θ) ≥ 0 when
[θ[ ≤ min
x
∗
j
γ
j

j ∈ I(x
∗
)
¸
= θ
∗
say .
Hence z(θ) ∈ Ω
p
whenever [θ[ ≤ θ
∗
. Since x
∗
is optimal we must have
c
x
∗
≥ c
z(θ) = c
x
∗
+θ
¸
j∈I(x
∗
)
c
j
y
j
for −
∗
θ ≤ θ ≤ θ
∗
.
Since θ can take on positive and negative values, the inequality above can hold on if
¸
J∈I(x
∗
)
c
j
γ
j
=
0, and then c
x
∗
= c
z(θ), so that z(θ) is also an optimal solution for [θ[ ≤ θ
∗
. But from the
deﬁnition of z(θ) it is easy to see that we can pick θ
0
with [θ
0
[ = θ
∗
such that z
j
(θ
0
) = x
∗
j
+θ
0
γ
j
= 0
for at least one j = j
0
in I(x
∗
). Then,
I(z(θ
0
)) ⊂ I(x
∗
) −¦j
0
¦ .
4.3. THE SIMPLEX ALGORITHM 39
Again, if ¦A
j
[j ∈ I(z(θ
0
))¦ is linearly independent, then we let z
∗
= z(θ
0
) and we are done.
Otherwise we repeat the procedure above with z(θ
0
). Clearly, in a ﬁnite number of steps we will
ﬁnd an optimal decision z
∗
which is also vertex. ♦
At this point we abandon the geometric term “vertex” and how to established LP terminology.
Deﬁnition: (i) z is said to be a basic feasible solution if z ∈ Ω
p
, and ¦A
j
[j ∈ I(z)¦ is linearly
independent. The set I(z) is then called the basis at z, and x
j
, j ∈ I(z), are called the basic
variables at z. x
j
, j ∈ I(z) are called the nonbasic variables at z.
Deﬁnition: A basic feasible solution z is said to be nondegenerate if I(z) has m elements.
Notation: Let z be a nondegenerate basic feasible solution, and let j
1
< j
2
< . . . < j
m
constitute I(z). Let D(z) denote the m m nonsingular matrix D(z) = [A
j
1
.
.
.A
j
2
.
.
. . . .
.
.
.A
jm
], let
c(z) denote the mdimensional column vector c(z) = (c
j
1
, . . . , c
jm
)
and deﬁne λ(z) by λ
(z) =
c
(z)[D(z)]
−1
. We call λ(z) the shadowprice vector at z.
Lemma 3: Let z be a nondegenerate basic feasible solution. Then z is optimal if and only if
λ
(z)A ≥ c
j
, for all , j ∈ I(z) . (4.25)
Proof: By Exercise 6 of Section 2.2, z is optimal iff there exists λ such that
λ
A
j
= c
j
, for , j ∈ I(z) , (4.26)
λ
A
j
≥ c
j
, for , j ∈ I(z) , (4.27)
But since z is nondegenerate, (4.26) holds iff λ = λ(z) and then (4.27) is the same as (4.25). ♦
4.3.2 The Simplex Algorithm.
The algorithm is divided into two parts: In Phase I we determine if Ω
p
is empty or not, and if not,
we obtain a basic feasible solution. Phase II starts with a basic feasible solution and determines if
it is optimal or not, and if not obtains another basic feasible solution with a higher value. Iterating
on this procedure, in a ﬁnite number of steps, either we obtain an optimum solution or we discover
that no optimum exists, i.e., sup ¦c
x[x ∈ Ω
p
¦ = +∞. We shall discuss Phase II ﬁrst.
We make the following simplifying assumption. We will comment on it later.
Assumption of nondegeneracy. Every basic feasible solution is nondegenerate.
Phase II:
Step 1. Let z
0
be a basic feasible solution obtained from Phase I or by any other means. Set k = 0
and go to Step 2.
Step 2. Calculate [D(z
k
)]
−1
,c(z
k
), and the shadowprice vector λ
(z
k
) = c
(z
k
)[D(z
k
)]
−1
. For
each j ∈ I(z
k
) calculate c
j
−λ
(z
k
)A
j
. If all these numbers are ≤ 0, stop, because z
k
is optimal
by Lemma 3. Otherwise pick any
ˆ
j ∈ I(z
k
) such that c
ˆ
j
−λ
(z
k
)A
ˆ
j
> 0 and go to Step 3.
Step 3. Let I(z
k
) consist of j
1
< j
2
< . . . < j
m
. Compute the vector
γ
k
= (γ
k
j
1
, . . . γ
k
jm
)
= [D(z
k
)]
−1
A
ˆ
j
. If γ
k
≤ 0, stop, because by Lemma 4 below, there is no
ﬁnite optimum. Otherwise go to Step 4.
Step 4. Compute θ = min ¦(z
k
j
γ
k
j
)[j ∈ i(z), γ
k
j
> 0¦. Evidently 0 < θ < ∞. Deﬁne z
k+1
by
40 CHAPTER 4. LINEAR PROGRAMMING
z
k+1
j
=
z
k
j
−θγ
k
j
θ
z
k
j
= 0
,
,
,
j ∈ I(z)
j =
ˆ
j
j =
ˆ
j and j ∈ I(z) .
(4.28)
By Lemma 5 below, z
k+1
is a basic feasible solution with c
z
k+1
> c
z
k
. Set k = k + 1 and return
to Step 2.
Lemma 4: If γ
k
≤ 0, sup ¦c
x[x ∈ Ω
p
¦ = ∞.
Proof: Deﬁne z(θ) by
z
j
(θ) =
z
j
−θγ
k
j
θ
z
j
= 0
,
,
,
j ∈ I(z)
j =
ˆ
j
j ∈ I(z) and j =
ˆ
j .
(4.29)
First of all, since γ
k
≤ 0 it follows that z(θ) ≥ 0 for θ ≥ 0. Next, Az(θ) = Az − θ
¸
j∈I(z)
γ
k
j
A
j
+
θA
ˆ
j
= Az by deﬁnition of γ
k
. Hence, z(θ) ∈ Ω
p
for θ ≥ 0. Finally,
c
z(θ)
= c
z −θc
(z
k
)γ
k
+θc
ˆ
j
= c
z +θ¦c
ˆ
j
−c
(z
k
)[D(z
k
)]
−1
A
ˆ
j
¦
= c
z +θ¦c
ˆ
j
−λ
(z
k
)A
ˆ
j
¦i .
(4.30)
But from step 2 ¦c
ˆ
j −λ
(z
k
)A
ˆ
j
¦ > 0, so that c
z(θ) → ∞as θ → ∞. ♦
Lemma 5: z
k+1
is a basic feasible solution and c
z
k+1
> c
z
k
.
Proof: Let
˜
j ∈ I(z
k
) be such that γ
k
˜
j
> 0 and z
k
˜
j
= θγ
k
˜
j
. Then from (4.28) we see that z
k+1
˜
j
= 0,
hence
I(z
k+1
) ⊂ (I(z) −¦
˜
j¦)
¸
¦
ˆ
j¦ , (4.31)
so that it is enough to prove that A
˜
j
is independent of ¦A
j
[j ∈ I(z), j =
˜
j¦. But if this is not the
case, we must have γ
k
˜
j
= 0, giving a contradiction. Finally if we compare (4.28) and (4.29), we see
from (4.30) that
c
z
k+1
−c
z
k
= θ¦c
ˆ
j
−γ
(z
k
)A
ˆ
j
¦ ,
which is positive from Step 2. ♦
Corollary 2: In a ﬁnite number of steps Phase II will obtain an optimal solution or will determine
that sup¦c
x[x ∈ Ω
p
¦ = ∞.
Corollary 3: Suppose Phase II terminates at an optimal basic feasible solution z
∗
. Then γ(z
∗
) is an
optimal solution of the dual of (4.24).
Exercise 2: Prove Corollaries 2 and 3.
Remark 1: By the nondegeneracy assumption, I(z
k+1
) has m elements, so that in (4.31) we must
have equality. We see then that D(z
k+1
) is obtained from D(z
k
) by replacing the column A
j
by
4.3. THE SIMPLEX ALGORITHM 41
the column A
ˆ
j
. More precisely if D(z
k
) = [A
j
1
.
.
. . . .
.
.
.A
j
i−1
.
.
.A
˜
j
.
.
.A
j
i+1
.
.
. . . .
.
.
.A
jm
] and if
j
k
<
ˆ
j < j
k+1
then D(z
k+1
) = [A
j
1
.
.
. . . .
.
.
.A
j
i−1
.
.
.A
j
i+1
.
.
. . . .
.
.
.A
j
k
.
.
.A
ˆ
j
.
.
.A
j
k+1
.
.
. . . .
.
.
.A
jm
]. Let E be the
matrix E = [A
j
1
.
.
. . . .
.
.
.A
j
i−1
.
.
.A
ˆ
j
.
.
.A
j
i+1
.
.
. . . .
.
.
.A
jm
]. Then [D(z
k+1
)]
−1
= P E
−1
where the matrix P
permutes the columns of D(z
k+1
) such that E = D(z
k+1
)P. Next, if A
ˆ
j
=
m
¸
=1
γ
j
A
j
, it is easy
to check that E
−1
= M[D(z
k
)]
−1
where
M =
1
1
.
.
.
1
−γ
j
1
γ
˜
j
1
γ
˜
j
−γ
jm
γ
˜
j
1
.
.
.
1
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
↑
ith column
Then [D(z
k+1
)]
−1
= PM[D(z
k
)]
−1
, so that these inverses can be easily computed.
Remark 2: The similarity between Step 2 of Phase II and Step 2 of the algorithm in 3.3.4 is
striking. The basic variables at z
k
correspond to the variables w
k
and nonbasic variables
correspond to u
k
. For each j ∈ I(z
k
) we can interpret the number c
j
−λ
(z
k
)A
j
to be the net
increase in the objective value per unit increase in the jth component of z
k
. This net increase is due
to the direct increase c
j
minus the indirect decrease λ
(z
k
)A
j
due to the compensating changes in
the basic variables necessary to maintain feasibility. The analogous quantity in 3.3.4 is
(∂f
0
/∂u
j
)(x
k
) −(λ
k
)
(∂f/∂u
j
)(x
k
).
Remark 3: By eliminating any dependent equations in (4.24) we can guarantee that the matrix A
has rank n. Hence at any degenerate basic feasible solution z
k
we can always ﬁnd
¯
I(z
k
) ⊃ I(z
k
)
such that
¯
I(z
k
) has m elements and ¦A
j
[j ∈
¯
I(z
k
)¦ is a linearly independent set. We can apply
Phase II using
¯
I(z
k
) instead of I(z
k
). But then in Step 4 it may turn out that θ = 0 so that
z
k+1
= z
k
. The reason for this is that
¯
I(z
k
) is not unique, so that we have to try various
alternatives for
¯
I(z
k
) until we ﬁnd one for which θ > 0. In this way the nondegeneracy
assumption can be eliminated. For details see (Canon, et al., [1970]).
We now describe how to obtain an initial basic feasible solution.
Phase I:
Step I. by multiplying some of the equality constraints in (4.24) by −1 if necessary, we can assume
that b ≥ 0. Replace the LP (4.24) by the LP (4.32) involving the variables x and y:
Maximize −
m
¸
i=1
y
i
subject to a
il
x
1
+. . . +a
in
x
n
+y
i
= b
i
, 1 ≤ i ≤ m ,
x
j
≥ 0 , y
i
≥ 0 , 1 ≤ j ≤ n , 1 ≤ i ≤ m .
(4.32)
42 CHAPTER 4. LINEAR PROGRAMMING
Go to step 2.
Step 2. Note that (x
0
, y
0
) = (0, b) is a basic feasible solution of (4.32). Apply phase II to (4.32)
starting with this solution. Phase II must terminate in an optimum based feasible solution (x
∗
, y
∗
)
since the value of the objective function in (4.32) lies between −
m
¸
i=1
b
i
and 0. Go to Step 3.
Step 3. If y
∗
= 0, x
∗
is a basic feasible solution for (4.24). If y
∗
= 0, by Exercise 3 below, (4.24)
has no feasible solution.
Exercise 3: Show that (4.24) has a feasible solution iff y
∗
= 0.
4.4 LP Theory of a Firm in a Competitive Economy
4.4.1 Activity analysis of the ﬁrm.
We think of a ﬁrm as a system which transforms input into outputs. There are m kinds of inputs
and k kinds of outputs. Inputs are usually classiﬁed into raw materials such as iron ore, crude oil,
or raw cotton; intermediate products such as steel, chemicals, or textiles; capital goods
3
such as
machines of various kinds, or factory buildings, ofﬁce equipment, or computers; ﬁnally various
kinds of labor services. The ﬁrm’s outputs themselves may be raw materials (if it is a mining
company) or intermediate products (if it is a steel mill) or capital goods (if it manufactures lathes)
or ﬁnished goods (if it makes shirts or bakes cookies) which go directly to the consumer. Labor is
not usually considered an output since slavery is not practiced; however, it may be considered an
output in a “closed,” dynamic Malthusian framework where the increase in labor is a function of the
output. (See the von Neumann model in (Nikaido [1968]), p. 141.)
Within the ﬁrm, this transformation can be conducted in different ways, i.e., different combina
tions of inputs can be used to produce the same combination of outputs, since human labor can
do the same job as some machines and machines can replace other kinds of machines, etc. This
substitutability among inputs is a fundamental concept in economics. We formalize it by specifying
which transformation possibilities are available to the ﬁrm.
By an input vector we mean any mdimensional vector r = (r
1
, . . . , r
m
)
with r ≥ 0, and by an
output vector we mean any kdimensional vector y = (y
1
, . . . , y
k
)
with y ≥ 0. We now make three
basic assumptions about the ﬁrm.
(i) The transformation of inputs into outputs is organized into a ﬁnite number, say n, of processes
or activities.
(ii) Each activity combines the k inputs in ﬁxed proportions into the m outputs in ﬁxed propor
tions. Furthermore, each activity can be conducted at any nonnegative intensity or level. Pre
cisely, the jth activity is characterized completely by two vectors A
j
= (a
1j
, a
2j
, . . . , a
mj
)
and
B
j
= (b
ij
, . . . , b
kj
)
so that if it is conducted at a level x
j
≥ 0, then it combines (transforms) the
input vector (a
1j
x
j
, . . . , a
mj
x
j
)
= x
j
A
j
into the output vector (b
1j
x
j
, . . . , b
kj
x
j
)
= x
j
B
j
. Let
A be the mn matrix [A
1
.
.
. . . .
.
.
.A
n
] and B be the k n matrix B = [B
1
.
.
. . . .
.
.
.B
n
].
3
It is more accurate to think of the services of capital goods rather than these goods themselves as inputs. It is these
services which are consumed in the transformation into outputs.
4.4. LP THEORY OF A FIRM IN A COMPETITIVE ECONOMY 43
(iii) If the ﬁrmconducts all the activities simultaneously with the jth activity at level x
j
≥ 0, 1 ≤ j ≤
n, then it transforms the input vector x
1
A
1
+. . . +x
n
A
n
into the output vector x
1
B
1
+. . . +x
n
B
n
.
With these assumptions we know all the transformations technically possible as soon as we spec
ify the matrices A and B. Which of these possible transformations will actually take place depends
upon their relative proﬁtability and availability of inputs. We study this next.
4.4.2 Shortterm behavior.
In the shortterm, the ﬁrm cannot change the amount available to it of some of the inputs such as
capital equipment, certain kinds of labor, and perhaps some raw materials. Let us suppose that these
inputs are 1, 2, . . . , and they are available in the amounts r
∗
1
, . . . , r
∗
, whereas the supply of the
remaining inputs can be varied. We assume that the ﬁrm is operating in a competitive economy
which means that the unit prices p = (p
1
, . . . , p
k
)
of the outputs, and q = (q
1
, . . . , q
m
)
of the
inputs is ﬁxed. Then the manager of the ﬁrm, if he is maximizing the ﬁrm’s proﬁts, faces the
following decision problem:
Maximize p
y −
m
¸
j=+1
q
j
r
j
subject to y = Bx,
a
i1
x
1
+. . . +a
in
x
n
≤ r
∗
i
, 1 ≤ i ≤ ,
a
i1
x
1
+. . . +a
in
x
n
≤ r
i
, + 1 ≤ i ≤ m ,
x
j
≥ 0, 1 ≤ j ≤ n; r
i
≥ 0 , + 1 ≤ i ≤ m .
(4.33)
The decision variables are the activity levels x
1
, . . . , x
n
, and the shortterm input supplies r
+1
, . . . , r
m
.
The coefﬁcients of B and Aare the ﬁxed technical coefﬁcients of the ﬁrm, the r
∗
i
are the ﬁxed short
term supplies, whereas the p
i
, q
j
are prices determined by the whole economy, which the ﬁrm ac
cepts as given. Under realistic conditions (4.33) has an optimal solution, say, x
∗
1
, . . . , x
∗
n
, r
∗
+1
, . . . , r
∗
m
.
4.4.3 Longterm equilibrium behavior.
In the long run the supplies of the ﬁrst inputs are also variable and the ﬁrm can change these
supplies from r
∗
1
, . . . , r
∗
by buying or selling these inputs at the market price q
1
, . . . , q
. Whether
the ﬁrm will actually change these inputs will depend upon whether it is proﬁtable to do so, and in
turn this depends upon the prices p, q. We say that the prices (p
∗
, q
∗
) and a set of input supplies
r
∗
= (r
∗
1
, . . . , r
∗
m
) are in (longterm) equilibrium if the ﬁrm has no proﬁt incentive to change r
∗
under the prices (p
∗
, q
∗
).
Theorem 1: p
∗
, q
∗
, r
∗
are in equilibrium if and only if q
∗
is an optimal solution of (4.34):
Minimize (r
∗
)
q
subject to A
q ≥ B
p
∗
q ≥ 0 .
(4.34)
Proof: Let c = B
p
∗
. By deﬁnition, p
∗
, q
∗
, r
∗
are in equilibrium iff for all ﬁxed ∆ ∈ R
m
,
M(∆) ≤ M(0) where M(∆) is the maximum value of the LP (4.35):
Maximize c
x −(q
∗
)
∆
subject to Ax ≤ r
∗
+ ∆ ,
x ≥ 0 .
(4.35)
44 CHAPTER 4. LINEAR PROGRAMMING
For ∆ = 0, (4.34) becomes the dual of (4.35) so that by the strong duality theorem, M(0) = (r
∗
)
q
∗
.
Hence p
∗
, q
∗
, r
∗
are in equilibrium iff
c
x −(q
∗
)
∆ ≤ M(0) = (r
∗
)
q
∗
, (4.36)
whenever x is feasible for (4.35). By weak duality if x is feasible for (4.35) and q is feasible for
(4.34),
c
x −(q
∗
)
∆ ≤ q
(r
∗
= ∆) −(q
∗
)
∆ , (4.37)
and, in particular, for q = q
∗
,
c
x −(q
∗
)
∆ ≤ (q
∗
)
(r
∗
+ ∆) −(q
∗
)
∆ = (q
∗
)
r
∗
♦
Remark 1: We have shown that (p
∗
, q
∗
, r
∗
are in longterm equilibrium iff q
∗
is an optimum
solution to the dual (namely (4.34)) of (4.38):
Maximize c
x
subject to Ax ≤ r
∗
x ≥ 0 .
(4.38)
This relation between p
∗
, q
∗
, r
∗
has a very nice economic interpretation. Recall that c = B
p
∗
, i.e.,
c
j
= p
∗
1
b
1j
+p
∗
2
b
2j
+. . . +p
∗
k
b
kj
. Now b
ij
is the amount of the ith output produced by operating the
jth activity at a unit level x
j
= 1. Hence, c
j
is the revenue per unit level operation of the jth activity
so that c
x is the revenue when the n activities are operated at levels x. On the other hand if the jth
activity is operated at level x
j
= 1, it uses an amount a
ij
of the ith input. If the ith input is valued at
a
∗
i
, then the input cost of operating at x
j
= 1, is
m
¸
i=1
q
i
a
ij
, so that the input cost of operating the n
activities at levels x is (A
q
∗
)
= (q
∗
)
Ax. Thus, if x
∗
is the optimum activity levels for (4.38) then
the output revenue is c
x
∗
and the input cost is (q
∗
)
Ax
∗
. But from (4.16), (q
∗
)
(Ax
∗
−r
∗
) = 0 so
that
c
x
∗
= (q
∗
)
r
∗
, (4.39)
i.e., at the optimum activity levels, in equilibrium, total revenues = total cost of input supplies. In
fact, we can say even more. From (4.15) we see that if x
=
ast
j
> 0 then
c
j
=
m
¸
i=1
q
∗
i
a
ij
,
i.e., at the optimum, the revenue of an activity operated at a positive level = input cost of that activity.
Also if
c
j
<
m
¸
i=1
q
∗
i
a
ij
,
then x
∗
j
= 0, i.e., if the revenue of an activity is less than its input cost, then at the optimum it is
operated at zero level. Finally, again from (4.15), if an equilibrium the optimum ith input supply r
∗
i
is greater than the optimum demand for the ith input,
4.4. LP THEORY OF A FIRM IN A COMPETITIVE ECONOMY 45
r
∗
i
>
n
¸
j=1
a
ij
x
∗
j
,
then q
∗
i
= 0, i.e., the equilibrium price of an input which is in excess supply must be zero, in other
words it must be a free good.
Remark 2: Returning to the shortterm decision problem (4.33), suppose that
(λ
∗
1
, . . . , λ
∗
, λ
∗
+1
, . . . , λ
∗
m
) is an optimum solution of the dual of (4.33). Suppose that the market
prices of inputs 1, . . . , are q
1
, . . . , q
. Let us denote by M(∆
1
, . . . , ∆
) the optimum value of
(4.33) when the amounts of the inputs in ﬁxed supply are r
∗
1
+ ∆
1
, . . . , r
∗
+ ∆
. Then if
(∂M/∂∆
i
)[
∆=0
exists, we can see from (4.22) that it is always proﬁtable to increase the ith input
by buying some additional amount at price q
i
if λ
∗
i
> q
i
, and conversely it is proﬁtable to sell some
of the ith input at price q
i
if λ
∗
i
< q
i
. Thus λ
∗
i
can be interpreted as the ﬁrm’s internal valuation of
the ith input or the ﬁrm’s imputed or shadow price of the ith input. This interpretation has wide
applicability, which we mention brieﬂy. Often engineering design problems can be formulated as
LPs of the form (4.10) or (4.19), where some of the coefﬁcients b
i
are design parameters. The
design procedure is to ﬁx these parameters at some nominal value b
∗
i
, and carry out the
optimization problem. Suppose the resulting optimal dual variables are λ
∗
i
. then we see (assuming
differentiability) that it is worth increasing b
∗
i
if the unit cost of increasing this parameter is less
than λ
∗
i
, and it is worth decreasing this parameter if the reduction in total cost per unit decrease is
greater than λ
∗
i
.
4.4.4 Longterm equilibrium of a competitive, capitalist economy.
The proﬁtmaximizing behavior of the ﬁrm presented above is one of the two fundamental building
blocks in the equilibrium theory of a competitive, capitalist economy. Unfortunately we cannot
present the details here. We shall limit ourselves to a rough sketch. We think of the economy as
a feedback process involving ﬁrms and consumers. Let us suppose that there are a total of h com
modities in the economy including raw materials, intermediate and capital goods, labor, and ﬁnished
products. By adding zero rows to the matrices (A, B) characterizing a ﬁrm we can suppose that all
the h commodities are possible inputs and all the h commodities are possible outputs. Of course,
for an individual ﬁrm most of the inputs and most of the outputs will be zero. the sole purpose for
making this change is that we no longer need to distinguish between prices of inputs and prices of
outputs. We observe the economy starting at time T. At this time there exists within the economy
an inventory of the various commodities which we can represent by a vector ω = (ω
1
, . . . , ω
h
) ≥ 0.
ω is that portion of the outputs produced prior to T which have not been consumed up to T. We are
assuming that this is a capitalist economy, which means that the ownership of ω is divided among
the various consumers j = 1, . . . , J. More precisely, the jth consumer owns the vector of commodi
ties ω(j) ≥ 0, and
J
¸
j=1
ω(j) = ω. We are including in ω(j) the amount of his labor services which
consumer j is willing to sell. Now suppose that at time T the prevailing prices of the h commodities
are λ = (λ
1
, . . . , λ
h
)
≥ 0. Next, suppose that the managers of the various ﬁrms assume that the
prices λ are not going to change for a long period of time. Then, from our previous analysis we
know that the manager of the ith ﬁrm will plan to buy input supplies r(i) ≥ 0, r(i) ∈ R
h
, such
46 CHAPTER 4. LINEAR PROGRAMMING
that (λ, r(i)) is in long term equilibrium, and he will plan to produce an optimum amount, say y(i).
Here i = 1, 2, . . . , I, where I is the total number of ﬁrms. We know that r(i) and y(i) depend on
λ, so that we explicitly write r(i, λ), y(i, λ). We also recall that (see (4.38))
λ
r(i, λ) = λ
y(i, λ) , 1 ≤ i ≤ I . (4.40)
Now the ith manager can buy r(i) from only two sources: outputs from other ﬁrms, and the con
sumers who collectively own ω. Similarly, the ith manager can sell his planned output y(i) either as
input supplies to other ﬁrms or to the consumers. Thus, the net supply offered for sale to consumers
is S(λ), where
S(λ) =
J
¸
j=1
ω(j) +
I
¸
i=1
y(i, λ) −
i
¸
i=1
r(i, λ) . (4.41)
We note two important facts. First of all, from (4.40), (4.41) we immediately conclude that
λ
S(λ) =
J
¸
j=1
λ
ω(j) , (4.42)
that is the value of the supply offered to consumers is equal to the value of the commodities (and
labor) which they own. The second point is that there is no reason to expect that S(λ) ≥ 0.
Now we come to the second building block of equilibrium theory. The value of the jth consumer’s
possessions is λ
ω(j). The theory assumes that he will plan to buy a set of commodities d(j) =
(d
1
(j), . . . , d
h
(j)) ≥ 0 so as to maximize his satisfaction subject to the constraint λ
d(j) = λ
ω(j).
Here also d(j) will depend on λ, so we write d(j, λ). If we add up the buying plans of all the
consumers we obtain the total demand
D(λ) =
J
¸
j=1
d(j, λ) ≥ 0 , (4.43)
which also satisﬁes
λ
D(λ) =
J
¸
j=1
λ
ω(j) . (4.44)
The most basic question of equilibrium theory is to determine conditions under which there exists a
price vector λ
E
such that the economy is in equilibrium, i.e., S(λ
E
) = D(λ
E
), because if such an
equilibrium price λ
E
exists, then at that price the production plans of all the ﬁrms and the buying
plan of all the consumers can be realized. Unfortunately we must stop at this point since we cannot
proceed further without introducing some more convex analysis and the ﬁxed point theorem. For
a simple treatment the reader is referred to (Dorfman, Samuelson, and Solow [1958], Chapter 13).
For a much more general mathematical treatment see (Nikaido [1968], Chapter V).
4.5 Miscellaneous Comments
4.5. MISCELLANEOUS COMMENTS 47
4.5.1 Some mathematical tricks.
It is often the case in practical decision problems that the objective is not welldeﬁned. There may
be a number of plausible objective functions. In our LP framework this situation can be formulated
as follows. The constraints are given as usual by Ax ≤ b, x ≥ 0. However, there are, say, k
objective functions (c
1
)
x, . . . , (c
k
)
x. It is reasonable then to deﬁne a single objective function
f
0
(x) by f
0
(x) = minimum ¦(c
1
)
x, (c
2
)
x, . . . , (c
k
)
x¦, so that we have the decision problem,
Maximize f
0
(x)
subject to Ax ≤ b, x ≥ 0 .
(4.45)
This is not a LP since f
0
is not linear. However, the following exercise shows how to transform
(4.45) into an equivalent LP.
Exercise 1: Show that (4.45) is equivalent to (4.46) below, in the sense that x
∗
is optimal for (4.45)
iff (x
∗
, y
∗
) = (x
∗
, f
0
(x
∗
)) is optimal for (4.46).
Maximize y
subject to Ax ≤ b, x ≤ 0
y ≤ (c
i
)
x , 1 ≤ i ≤ k .
(4.46)
Exercise 1 will also indicate how to do Exercise 2.
Exercise 2: Obtain an equivalent LP for (4.47):
Maximize
n
¸
j=1
c
i
(x
i
)
subject to Ax ≤ b, x ≤ 0 ,
(4.47)
where c
i
: R → R are concave, piecewiselinear functions of the kind shown in Figure 4.3.
The abovegiven assumption of the concavity of the c
i
is crucial. In the next exercise, the inter
pretation of “equivalent” is purposely left ambiguous.
Exercise 3: Construct an example of the kind (4.47), where the c
i
are piecewise linear (but not
concave), and such that there is no equivalent LP.
It turns out however, that even if the c
i
are not concave, an elementary modiﬁcation of the Simplex
algorithm can be given to obtain a “local” optimal decision. See (Miller [1963]).
4.5.2 Scope of linear programming.
LP is today the single most important optimization technique. This is because many decision prob
lems can be adequately formulated as LPs, and, given the capabilities of modern computers, the
Simplex method (together with its variants) is an extremely powerful technique for solving LPs in
volving thousands of variables. To obtain a feeling for the scope of LP we refer the reader to the
book by one of the originators of LP (Dantzig [1963]).
48 CHAPTER 4. LINEAR PROGRAMMING
.
.
.
c
i
(x
i
)
x
i
Figure 4.3: A function of the form used in Exercise 2.
Chapter 5
OPTIMIZATION OVER SETS
DEFINED BY INEQUALITY
CONSTRAINTS: NONLINEAR
PROGRAMMING
In many decisionmaking situations the assumption of linearity of the constraint inequalities in LP
is quite restrictive. The linearity of the objective function is not restrictive as shown in the ﬁrst
exercise below. In Section 1 we present the general nonlinear programming problem (NP) and
prove the KuhnTucker theorem. Section 2 deals with Duality theory for the case where appropriate
convexity conditions are satisﬁed. Two applications are given. Section 3 is devoted to the important
special case of quadratic programming. The last section is devoted to computational considerations.
5.1 Qualitative Theory of Nonlinear Programming
5.1.1 The problem and elementary results.
The general NP is a decision problem of the form:
Maximize f
0
(x)
subject to (x) ≤ 0 , i = 1, . . . , m,
(5.1)
where x ∈ R
n
, f
i
: R
n
→ R, i = 0, 1, . . . , m, are differentiable functions. As in Chapter 4,
x ∈ R
n
is said to be a feasible solution if it satisﬁes the constraints of (5.1), and Ω ⊂ R
n
is the
subset of all feasible solutions; x
∗
∈ Ω is said to be an optimal decision or optimal solution if
f
0
(x
∗
) ≥ f
0
(x) for x ∈ Ω. From the discussion in 4.1.2 it is clear that equality constraints and sign
constraints on some of the components of x can all be transformed into the form (5.1). The next
exercise shows that we could restrict ourselves to objective functions which are linear; however, we
will not do this.
Exercise 1: Show that (5.2), with variables y ∈ R, x ∈ R
n
, is equivalent to (5.1):
49
50 CHAPTER 5. NONLINEAR PROGRAMMING
Maximize y
subject to f
i
(x) ≤ 0, 1 ≤ i ≤ m, and y −f
0
(x) ≤ 0 .
(5.2)
Returning to problem (5.1), we are interested in obtaining conditions which any optimal decision
must satisfy. The argument parallels very closely that developed in Exercise 1 of 4.1 and Exercise 1
of 4.2. The basic idea is to linearize the functions f
i
in a neighborhood of an optimal decision x
∗
.
Deﬁnition: Let x be a feasible solution, and let I(x) ⊂ ¦1, . . . , m¦ be such that f
i
(x) = 0 for
ı ∈ I(x), f
i
(x) < 0 for i ∈ I(x). (The set I(x) is called the set of active constraints at x.)
Deﬁnition: (i) Let x ∈ Ω. A vector h ∈ R
n
is said to be an admissible direction for Ω at x if there
exists a sequence x
k
, k = 1, 2, . . . , in Ω and a sequence of numbers ε
k
, k = 1, . . . , with ε
k
> 0
for all k such that
lim
k→∞
x
k
= x ,
lim
k→∞
1
ε
k
(x
k
−x) = h .
(ii) Let C(Ω, x) = ¦h[h is an admissible direction for Ω at x¦. C(Ω, x) is called the tangent cone
of Ω at x. Let K(Ω, x) = ¦x + h[h ∈ C(Ω, x)¦. (See Figures 5.1 and 5.2 and compare them with
Figures 4.1 and 4.2.)
If we take x
k
= x and ε
k
= 1 for all k, we see that 0 ∈ C(Ω, x) so that the tangent cone is always
nonempty. Two more properties are stated below.
Exercise 2: (i) Show that C(Ω, x) is a cone, i.e., if h ∈ C(Ω, x) and θ ≥ 0, then θh ∈ C(Ω, x).
(ii) Show that C(Ω, x) is a closed subset of R
n
. (Hint for (ii): For m = 1, 2, . . . , let h
m
and
¦x
mk
, ε
mk
> 0¦
∞
k=1
be such that x
mk
→ x and (1/ε
mk
)(x
mk
−x) → h
m
as k → ∞. Suppose
that h
m
→ h as m → ∞. Show that there exist subsequences ¦x
mkm
, ε
mkm
¦
∞
m=1
such that
x
mkm
→ x and (1/ε
mkm
)(x
mkm
−x) → h as m → ∞.)
In the deﬁnition of C(Ω, x) we made no use of the particular functional description of Ω. The
following elementary result is more interesting in this light and should be compared with (2.18) in
Chapter 2 and Exercise 1 of 4.1.
Lemma 1: Suppose x
∗
∈ Ω is an optimum decision for (5.1).
Then
f
0x
(x
∗
)h ≤ 0 for all h ∈ C(Ω, x
∗
) . (5.3)
Proof: Let x
k
∈ Ω, ε
k
> 0, k = 1, 2, 3, . . . , be such that
5.1. QUALITATIVE THEORY OF NONLINEAR PROGRAMMING 51
,
P
¦x[f
3
(x) = 0¦
Q
x
∗
direction of
increasing
payoff
π(k) =
¦x[f
0
(x) = k¦
¦x[f
1
(x) = 0¦
Ω
R
¦x[f
2
(x) = 0¦
Figure 5.1: Ω = PQR
lim
k→∞
x
k
= x
∗
,
lim
k→∞
1
ε
k
(x
k
−x
∗
) = h . (5.4)
Note that in particular (5.4) implies
lim
k→∞
1
ε
k
[x
k
−x
∗
[ = [h[ . (5.5)
Since f
0
is differentiable, by Taylor’s theorem we have
f
0
(x
k
) = f
0
(x
∗
+ (x
k
−x
∗
)) = f
0
(x
∗
) +f
0x
(x
∗
)(x
k
−x
∗
) +o([x
k
−x
∗
[) . (5.6)
Since x
k
∈ Ω, and x
∗
is optimal, we have f
0
(x
k
) ≤ f
0
(x
∗
), so that
0 ≥ f
0x
(x
∗
)
(x
k
−x
∗
)
ε
k
+
o(x
k
−x
∗
)
ε
k
.
Taking limits as k → ∞, using (5.4) and (5.5), we can see that
0 ≥
=
lim
k→∞
f
0x
(x
∗
)h. ♦
f
0x
(x
∗
)
(x
k
−x
∗
)
ε
k
+
lim
k→∞
o(x
k
−x
∗
)
x
k
−x
∗
 lim
k→∞
x
k
−x
∗

ε
k
52 CHAPTER 5. NONLINEAR PROGRAMMING








x
∗
K(Ω, x
∗
)
0
C(Ω, x
∗
)
Figure 5.2: C(Ω, x
∗
) is the tangent cone of Ω at x
∗
.
The basic problem that remains is to characterize the set C(Ω, x
∗
) in terms of the derivatives of the
functions f
i
. Then we can apply Farkas’ Lemma just as in Exercise 1 of 4.2.
Lemma 2: Let x
∗
∈ Ω. Then
C(Ω, x
∗
) ⊂ ¦h[f
ix
(x
∗
)h ≤ 0 for all i ∈ I(x
∗
)¦ . (5.7)
Proof: Let h ∈ R
n
and x
k
∈ Ω, ε
k
> 0, k = 1, 2, . . . , satisfy (5.4). Since f
i
is differentiable, by
Taylor’s theorem we have
f
i
(x
k
) = f
i
(x
∗
) +f
ix
(x
∗
)(x
k
−x
∗
) +o([x
k
−x
∗
[) .
Since x
k
∈ Ω, f
i
(x
k
) ≤ 0, and if i ∈ I(x
∗
), f
i
(x
∗
) = 0, so that f
i
(x
k
) ≤ f
i
(x
∗
). Following the
proof of Lemma 1 we can conclude that 0 ≥ f
ix
(x
∗
)h. ♦
Lemma 2 gives us a partial characterization of C(Ω, x
∗
). Unfortunately, in general the inclusion
sign in (5.7) cannot be reversed. The main reason for this is that the set ¦f
ix
(x
∗
)[i ∈ I(x
∗
)¦ is not
in general linearly independent.
Exercise 3: Let x ∈ R
2
, f
1
(x
1
, x
2
) = (x
1
−1)
3
+x
2
, and f
2
(x
1
, x
2
) = −x
2
. Let
(x
∗
1
, x
∗
2
) = (1, 0). Then I(x
∗
) = ¦1, 2¦. Show that
5.1. QUALITATIVE THEORY OF NONLINEAR PROGRAMMING 53
C(Ω, x
∗
) = ¦h[f
ix
(x
∗
)h ≤ 0 , i = 1, 2, ¦.
(Note that ¦f
1x
(x
∗
), f
2x
(x
∗
)¦ is not a linearly independent set; see Lemma 4 below.)
5.1.2 KuhnTucker Theorem.
Deﬁnition: Let x
∗
∈ Ω. We say that the constraint qualiﬁcation (CQ) is satisﬁed at x
∗
if
C(Ω, x) = ¦h[f
ix
(x
∗
)h ≤ 0 for all i ∈ I(x
∗
)¦,
and we say that CQ is satisﬁed if CQ is satisﬁed at all x ∈ Ω. (Note that by Lemma 2 C(Ω, x) is
always a subset of the righthand side.)
Compare the next result with Exercise 2 of 4.2.
Theorem 1: (Kuhn and Tucker [1951]) Let x
∗
be an optimum solution of (5.1), and suppose that
CQ is satisﬁed at x
∗
. Then there exist λ
∗
i
≥ 0, for i ∈ I(x
∗
), such that
f
0x
(x
∗
) =
¸
i∈I(x
∗
)
λ
∗
i
f
ix
(x
∗
)
(5.8)
Proof: By Lemma 1 and the deﬁnition of CQ it follows that f
0x
(x
∗
)h ≤ 0 whenever f
ix
(x
∗
)h ≤ 0
for all i ∈ I(x
∗
). By the Farkas’ Lemma of 4.2.1 it follows that there exist λ
∗
i
≥ 0 for i ∈ I(x
∗
)
such that (5.8) holds. ♦
In the original formulation of the decision problem we often have equality constraints of the form
r
j
(x) = 0, which get replaced by r
j
(x) ≤ 0, −r
j
(x) ≤ 0 to give the form (5.1). It is convenient in
application to separate the equality constraints from the rest. Theorem 1 can then be expressed as
Theorem 2.
Theorem 2: Consider the problem (5.9).
Maximize f
0
(x)
subject to f
i
(x) ≤ 0 , i = 1, . . . , m,
r
j
(x) = 0 , j = 1, . . . , k .
(5.9)
Let x
∗
be an optimum decision and suppose that CQ is satisﬁed at x
∗
. Then there exist λ
∗
i
≥ 0, i =
1, . . . , m, and µ
∗
j
, j = 1, . . . , k such that
f
0x
(x
∗
) =
m
¸
i=1
λ
∗
i
f
ix
(x
∗
) +
k
¸
j=1
µ
∗
j
r
jx
(x
∗
) , (5.10)
and
λ
∗
i
= 0 whenever f
i
(x
∗
) < 0 . (5.11)
Exercise 4: Prove Theorem 2.
54 CHAPTER 5. NONLINEAR PROGRAMMING
An alternative form of Theorem 1 will prove useful for computational purposes (see Section 4).
Theorem 3: Consider (5.9), and suppose that CQ is satisﬁed at an optimal solution x
∗
. Deﬁne
ψ : R
n
→ R by
ψ(h) = max ¦−f
0x
(x
∗
)h, f
1
(x
∗
) +f
1x
(x
∗
)h, . . . , f
m
(x
∗
) +f
mx
(x
∗
)h¦ ,
and consider the decision problem
Minimize ψ(h)
subject to −ψ(h) −f
0x
(x
∗
)h ≤ 0,
−ψ(h) +f
i
(x
∗
) +f
ix
(x
∗
)h ≤ 0 , 1 ≤ i ≤ m
−1 ≤ h
i
≤ 1 , i = 1, . . . , n .
(5.12)
Then h = 0 is an optimal solution of (5.12).
Exercise 5: Prove Theorem 3. (Note that by Exercise 1 of 4.5, (5.12) can be transformed into a
LP.)
Remark: For problem (5.9) deﬁne the Lagrangian function L:
(x
1
, . . . , x
n
; λ
1
, . . . , λ
m
; µ
1
, . . . , µ
k
) → f
0
(x) −
m
¸
i=1
λ
i
f
i
(x) −
k
¸
j=1
µ
j
r
j
(x).
Then Theorem 2 is equivalent to the following statement: if CQ is satisﬁed and x
∗
is optimal, then
there exist λ
∗
≥ 0 and µ
∗
such that L
x
(x
∗
, λ
∗
, µ
∗
) = 0 and L(x
∗
, λ
∗
, µ
∗
) ≤ L(x
∗
, λ, µ) for all
λ ≥ 0, µ.
There is a very important special case when the necessary conditions of Theorem 1 are also
sufﬁcient. But ﬁrst we need some elementary properties of convex functions which are stated as an
exercise. Some additional properties which we will use later are also collected here.
Recall the deﬁnition of convex and concave functions in 4.2.3.
Exercise 6: Let X ⊂ R
n
be convex. Let h : X → R be a differentiable function. Then
(i) h is convex iff h(y) ≥ h(x) +h
x
(x)(y −x) for all x, y, in X,
(ii) h is concave iff h(y) ≤ h(x) +h
x
(x)(y −x) for all x, y in X,
(iii) h is concave and convex iff h is afﬁne, i.e. h(x) ≡ α +b
x for some
ﬁxed α ∈ R, b ∈ R
n
.
Suppose that h is twice differentiable. Then
(iv) h is convex iff h
xx
(x) is positive semideﬁnite for all x in X,
(v) h is concave iff h
xx
(x) is negative semideﬁnite for all x in X,
(vi) h is convex and concave iff h
xx
(x) ≡ 0.
Theorem 4: (Sufﬁcient condition) In (5.1) suppose that f
0
is concave and f
i
is convex for
i = 1, . . . , m. Then
(i) Ω is a convex subset of R
n
, and
(ii) if there exist x
∗
∈ Ω, λ
∗
i
≥ 0, i ∈ I(x
∗
), satisfying (5.8), then x
∗
is an optimal solution of
(5.1).
Proof:
(i) Let y, z be in Ω so that f
i
(y) ≤ 0, f
i
(z) ≤ 0 for i = 1, . . . , m. Let 0 ≤ θ ≤ 1. Since f
i
is
convex we have
5.1. QUALITATIVE THEORY OF NONLINEAR PROGRAMMING 55
f
i
(θy + (1 −θ)z) ≤ θf
i
(y) + (1 −θ)f
i
(z) ≤ 0 , 1 ≤ i ≤ m,
so that (θy + (1 −θ)z) ∈ Ω, hence Ω is convex.
(ii) Let x ∈ Ω be arbitrary. Since f
0
is concave, by Exercise 6 we have
f
0
(x) ≤ f
0
(x
∗
) +f
0x
(x
∗
)(x −x
∗
) ,
so that by (5.8)
f
0
(x) ≤ f
0
(x
∗
) +
¸
i∈I(x
∗
)
λ
∗
i
f
ix
(x
∗
)(x −x
∗
) .
(5.13)
Next, f
i
is convex so that again by Exercise 6,
f
i
(x) ≥ f
i
(x
∗
) +f
ix
(x
∗
)(x −x
∗
) ;
but f
i
(x) ≤ 0, and f
i
(x
∗
) = 0 for i ∈ I(x
∗
), so that
f
ix
(x
∗
)(x −x
∗
) ≤ 0 for i ∈ I(x
∗
) . (5.14)
Combining (5.14) with the fact that λ
∗
i
≥ 0, we conclude from (5.13) that f
0
(x) ≤ f
0
(x
∗
), so that
x
∗
is optimal. ♦
Exercise 7: Under the hypothesis of Theorem 4, show that the subset Ω
∗
of Ω, consisting of all the
optimal solutions of (5.1), is a convex set.
Exercise 8: A function h : X → R deﬁned on a convex set X ⊂ R
n
is said to be strictly convex if
h(θy + (1 −θ)z) < θh(y) + (1 −θ)h(z) whenever 0 < θ < 1 and y, z are in X with y = z. h is
said to be strictly concave if −h is strictly convex. Under the hypothesis of Theorem 4, show that
an optimal solution to (5.1) is unique (if it exists) if either f
0
is strictly concave or if the
f
i
, 1 ≤ i ≤ m, are strictly convex. (Hint: Show that in (5.13) we have strict inequality if x = x
∗
.)
5.1.3 Sufﬁcient conditions for CQ.
As stated, it is usually impractical to verify if CQ is satisﬁed for a particular problem. In this
subsection we give two conditions which guarantee CQ. These conditions can often be veriﬁed in
practice. Recall that a function g : R
n
→ R is said to be afﬁne if g(x) ≡ α + b
x for some ﬁxed
α ∈ R and b ∈ R
n
.
We adopt the formulation (5.1) so that
Ω = ¦x ∈ R
n
[f
i
(x) ≤ 0 , 1 ≤ i ≤ m¦ .
Lemma 3: Suppose x
∗
∈ Ω and suppose there exists h
∗
∈ R
n
such that for each i ∈ I(x
∗
), either
f
ix
(x
∗
)h
∗
< 0, or f
ix
(x
∗
)h
∗
= 0 and f
i
is afﬁne. Then CQ is satisﬁed at x
∗
.
Proof: Let h ∈ R
n
be such that f
ix
(x
∗
)h ≤ 0 for i ∈ I(x
∗
). Let δ > 0. We will ﬁrst show that
(h +δh
∗
) ∈ C(Ω, x
∗
). To this end let ε
k
> 0, k = 1, 2, . . . , be a sequence converging to 0 and set
x
k
= x
∗
+ε
k
(h +δh
∗
). Clearly x
k
converges to x
∗
, and (1/ε
k
)(x
k
−x
∗
) converges to (h +δh
∗
).
Also for i ∈ I(x
∗
), if f
ix
(x
∗
)h < 0, then
f
i
(x
k
) = f
i
(x
∗
) +ε
k
f
ix
(x
∗
)(h +δh
∗
) +o(ε
k
[h +δh
∗
[)
≤ δε
k
f
ix
(x
∗
)h
∗
+o(ε
k
[h +δh
∗
[)
< 0 for sufﬁciently large k ,
whereas for i ∈ I(x
∗
), if f
i
is afﬁne, then
56 CHAPTER 5. NONLINEAR PROGRAMMING
f
i
(x
k
) = f
i
(x
∗
) +ε
k
f
ix
(x
∗
)(h +δh
∗
) ≤ 0 for all k .
Finally, for i ∈ I(x
∗
) we have f
i
(x
∗
) < 0, so that f
i
(x
k
) < 0 for sufﬁciently large k. Thus we
have also shown that x
k
∈ Ω for sufﬁciently large k, and so by deﬁnition (h + δh
∗
) ∈ C(Ω, x
∗
).
Since δ > 0 can be arbitrarily small, and since C(Ω, x
∗
) is a closed set by Exercise 2, it follows that
h ∈ C(Ω, x
∗
). ♦
Exercise 9: Suppose x
∗
∈ Ω and suppose there exists ˆ x ∈ R
n
such that for each i ∈ I(x
∗
), either
f
i
(x
∗
) < 0 and f
i
is convex, or f
i
(ˆ x) ≤ 0 and f
i
is afﬁne. Then CQ is satisﬁed at x
∗
. (Hint: Show
that h
∗
= ˆ x −x
∗
satisﬁes the hypothesis of Lemma 3.)
Lemma 4: Suppose x
∗
∈ Ω and suppose there exists h
∗
∈ R
n
such that f
ix
(x
∗
)h
∗
≤ 0 for
i ∈ I(x
∗
), and ¦f
ix
(x
∗
)[i ∈ I(x
∗
), f
ix
(x
∗
)h
∗
= 0¦ is a linearly independent set. Then CQ is
satisﬁed at x
∗
.
Proof: Let h ∈ R
n
be such that f
ix
(x
∗
)h ≤ 0 for all i ∈ I(x
∗
). Let δ > 0. We will show that
(h +δh
∗
) ∈ C(Ω, x
∗
). Let J
δ
= ¦i[i ∈ I(x
∗
), f
ix
(x
∗
)(h +δh
∗
) = 0¦, consist of p elements.
Clearly J
δ
⊂ J = ¦i[i ∈ I(x
∗
), f
i
x(x
∗
)h
∗
= 0¦, so that ¦f
ix
(x
∗
, u
∗
)[i ∈ J
δ
¦ is linearly
independent. By the Implicit Function Theorem, there exist ρ > 0, an open set V ⊂ R
n
containing
x
∗
= (w
∗
, u
∗
), and a differentiable function g : U → R
p
, where U = ¦u ∈ R
n−p
[[u −u
∗
[ < ρ¦,
such that
f
i
(w, u) = 0, i ∈ J
δ
, and (w, u) ∈ V
iff
u ∈ U, and w = g(u) .
Next we partition h, h
∗
as h = (ξ, η), h
∗
= (ξ
∗
, η
∗
) corresponding to the partition of x = (w, u).
Let ε
k
> 0, k = 1, 2 . . . , be any sequence converging to 0, and set u
k
= u
∗
+ ε
k
(η +δη
∗
), w
k
=
g(u
k
), and ﬁnally x
k
= (s
k
, u
k
).
We note that u
k
converges to u
∗
, so w
k
= g(u
k
) converges to w
∗
= g(u
∗
). Thus, x
k
converges
to x
∗
. Now (1/ε
k
)(x
k
−x
∗
) = (1/ε
k
)(w
k
−w
∗
, u
k
−u
∗
) = (1/ε
k
)(g(u
k
) −g(u
∗
), ε
k
(η +δη
∗
)).
Since g is differentiable, it follows that (1/ε
k
)(x
k
−x
∗
) converges to (g
u
(u
∗
)(η +δη
∗
), η +δη
∗
).
But for i ∈ J
δ
we have
0 = f
ix
(x
∗
)(h +δh
∗
) = f
iw
(x
∗
)(ξ +δξ
∗
) +f
iu
(x
∗
)(η +δη
∗
) . (5.15)
Also, for i ∈ J
δ
, 0 = f
i
(g(u), u) for u ∈ U so that 0 = f
iw
(x
∗
)g
u
(u
∗
) +f
iu
(x
∗
), and hence
0 = f
iw
(x
∗
)g
u
(u
∗
)(η +δη
∗
) +f
iu
(x
∗
)(η +δη
∗
) . (5.16)
If we compare (5.15) and (5.16) and recall that ¦f
iw
(x
∗
)[i ∈ J
δ
¦ is a basis in R
p
we can conclude
that (ξ +δξ
∗
) = g
u
(u
∗
)(η +δη
∗
) so that (1/ε
k
)(x
k
−x
∗
) converges to (h +hδh
∗
).
It remains to show that x
k
∈ Ω for sufﬁciently large k. First of all, for i ∈ J
δ
, f
i
(x
k
) =
f
i
(g(u
k
), u
k
) = 0, whereas for i ∈ J
δ
, i ∈ I(x
∗
),
f
i
(x
k
) = f
i
(x
∗
) +f
ix
(x
∗
)(x
k
−x
∗
) +o([x
k
−x
∗
[)
f
i
(x
∗
) +ε
k
f
ix
(x
∗
)(h +δh
∗
) +o(ε
k
) +o([x
k
−x
∗
[),
5.2. DUALITY THEORY 57
and since f
i
(x
∗
) = 0 whereas f
ix
(x
∗
)(h + δh
∗
) < 0, we can conclude that f
i
(x
k
) < 0 for sufﬁ
ciently large k. Thus, x
k
∈ Ω for sufﬁciently large k. Hence, (h +δh
∗
) ∈ C(Ω, x
∗
).
To ﬁnish the proof we note that δ > 0 can be made arbitrarily small, and C(Ω, x
∗
) is closed by
Exercise 2, so that h ∈ C(Ω, x
∗
). ♦
The next lemma applies to the formulation (5.9). Its proof is left as an exercise since it is very
similar to the proof of Lemma 4.
Lemma 5: Suppose x
∗
is feasible for (5.9) and suppose there exists h
∗
∈ R
n
such that the set
¦f
ix
(x
∗
)[i ∈ I(x
∗
), f
ix
(x
∗
)h
∗
= 0¦
¸
¦r
jx
(x
∗
)[j = 1, . . . , k¦ is linearly independent, and f
ix
(x
∗
)h
∗
≤
0 for i ∈ I(x
∗
), r
jx
(x
∗
)h
∗
= 0 for 1 ≤ j ≤ k. Then CQ is satisﬁed at x
∗
.
Exercise 10: Prove Lemma 5
5.2 Duality Theory
Duality theory is perhaps the most beautiful part of nonlinear programming. It has resulted in many
applications within nonlinear programming, in terms of suggesting important computational algo
rithms, and it has provided many unifying conceptual insights into economics and management
science. We can only present some of the basic results here, and even so some of the proofs are
relegated to the Appendix at the end of this Chapter since they depend on advanced material. How
ever, we will give some geometric insight. In 2.3 we give some application of duality theory and in
2.2 we refer to some of the important generalizations. The results in 2.1 should be compared with
Theorems 1 and 4 of 4.2.1 and the results in 4.2.3.
It may be useful to note in the following discussion that most of the results do not require differ
entiability of the various functions.
5.2.1 Basic results.
Consider problem (5.17) which we call the primal problem:
Maximize f
0
(x)
subject to f
i
(x) ≤
ˆ
b
i
, 1 ≤ i ≤ m
x ∈ X ,
(5.17)
where x ∈ R
n
, f
i
: R
n
→ R, 1 ≤ i ≤ m, are given convex functions, f
0
: R
n
→ R is a
given concave function, X is a given convex subset of R
n
and
ˆ
b = (
ˆ
b
1
, . . . ,
ˆ
b
m
)
is a given vector.
For convenience, let f = (f
1
, . . . , f
m
)
: R
n
→ R
m
. We wish to examine the behavior of the
maximum value of (5.17) as
ˆ
b varies. So we deﬁne
Ω(b) = ¦x[x ∈ X, f(x) ≤ b¦, B = ¦b[Ω(b) = φ¦,
and
M : B → R
¸
¦+∞¦ by M(b) = sup¦f
0
(x)[x ∈ X, f(x) ≤ b¦
= sup¦f
0
(x)[x ∈ Ω(b)¦ ,
so that in particular if x
∗
is an optimal solution of (5.17) then M(
ˆ
b) = f
0
(ˆ x). We need to consider
the following problem also. Let λ ∈ R
m
, λ ≥ 0, be ﬁxed.
Maximize f
0
(x) −λ
(f(x) −
ˆ
b)
subject to x ∈ X ,
(5.18)
58 CHAPTER 5. NONLINEAR PROGRAMMING
and deﬁne
m(λ) = sub¦f
0
(x) −λ
(f(x) −
ˆ
b)[x ∈ X¦ .
Problem (5.19) is called the dual problem:
Minimize m(λ)
subject to λ ≥ 0 .
(5.19)
Let m
∗
= inf ¦m(λ)[λ ≥ 0¦.
Remark 1: The set X in (5.17) is usually equal to R
n
and then, of course, there is no reason to
separate it out. However, it is sometimes possible to include some of the constraints in X in such
a way that the calculation of m(λ) by (5.18) and the solution of the dual problem (5.19) become
simple. For example see the problems discussed in Sections 2.3.1 and 2.3.2 below.
Remark 2: It is sometimes useful to know that Lemmas 1 and 2 below hold without any convexity
conditions on f
0
, f, X. Lemma 1 shows that the cost function of the dual problem is convex which
is useful information since there are computation techniques which apply to convex cost functions
but not to arbitrary nonlinear cost functions. Lemma 2 shows that the optimum value of the dual
problem is always an upper bound for the optimum value of the primal.
Lemma 1: m : R
n
+
→ R
¸
¦+∞¦ is a convex function. (Here R
n
+
= ¦λ ∈ R
n
[λ ≥ 0¦.)
Exercise 1: Prove Lemma 1.
Lemma 2: (Weak duality) If x is feasible for (5.17), i.e., x ∈ Ω(
ˆ
b), and if λ ≥ 0, then
f
0
(x) ≤ M(
ˆ
b) ≤ m
∗
≤ m(λ) . (5.20)
Proof: Since f(x) −
ˆ
b ≤ 0, and λ ≥ 0, we have λ
(f(x) −
ˆ
b) ≤ 0. So,
f
0
(x) ≤ f
0
(x) −λ
(f(x) −
ˆ
b), for x ∈ Ω(
ˆ
b), λ ≥ 0 .
Hence
f
0
(x) ≤ sup ¦f
0
(x)[x ∈ Ω(
ˆ
b)¦ = M(
ˆ
b)
≤ sup ¦f
0
(x) −λ
(f(x) −
ˆ
b)[x ∈ Ω(
ˆ
b)¦ and since Ω(
ˆ
b) ⊂ X,
≤ sup ¦f
0
(x) −λ
(f(x) −
ˆ
b)[x ∈ X¦ = m(λ) .
Thus, we have
f
0
(x) ≤ M(
ˆ
b) ≤ m(λ) for x ∈ Ω(
ˆ
b), λ ≥ 0 ,
and since M(
ˆ
b) is independent of λ, if we take the inﬁmum with respect to λ ≥ 0 in the righthand
inequality we get (5.20). ♦
The basic problem of Duality Theory is to determine conditions under which M(
ˆ
b) = m
∗
in
(5.20). We ﬁrst give a simple sufﬁciency condition.
Deﬁnition: A pair (ˆ x,
ˆ
λ) with ˆ x ∈ X, and
ˆ
λ ≤ 0 is said to satisfy the optimality conditions if
5.2. DUALITY THEORY 59
ˆ x is optimal solution of (5.18) with λ =
ˆ
λ, (5.21)
ˆ x is feasible for (5.17), i.e., f
i
(ˆ x) ≤
ˆ
b
i
for i = 1, . . . , m , (5.22)
ˆ
λ
i
= 0 when f
i
(ˆ x) <
ˆ
b
i
, equivalently,
ˆ
λ
(f(ˆ x) −
ˆ
b) = 0. (5.23)
ˆ
λ ≥ 0 is said to be an optimal price vector if there is ˆ x ∈ X such that (ˆ x,
ˆ
λ) satisfy the optimality
condition. Note that in this case ˆ x ∈ Ω(
ˆ
b) by virtue of (5.22).
The next result is equivalent to Theorem 4(ii) of Section 1 if X = R
n
, and f
i
, 0 ≤ i ≤ m, are
differentiable.
Theorem 1: (Sufﬁciency) If (ˆ x,
ˆ
λ) satisfy the optimality conditions, then ˆ x is an optimal solution to
the primal,
ˆ
λ is an optimal solution to the dual, and M(
ˆ
b) = m
∗
.
Proof: Let x ∈ Ω(
ˆ
b), so that
ˆ
λ
(f(x) −
ˆ
b) ≤ 0. Then
f
0
(x) ≤ f
0
(x) −
ˆ
λ
(f(x) −
ˆ
b)
≤ sup¦f
0
(x) −
ˆ
λ
(f(x) −
ˆ
b)[x ∈ X¦
= f
0
(ˆ x) −
ˆ
λ
(f(ˆ x) −
ˆ
b) by (5.21)
= f
0
(ˆ x) by (5.23)
so that ˆ x is optimal for the primal, and hence by deﬁnition f
0
(ˆ x) = M(
ˆ
b). Also
m(
ˆ
λ) = f
0
(ˆ x) −
ˆ
λ
(f(ˆ x) −
ˆ
b)
f
0
(ˆ x) = M(
ˆ
b) ,
so that from Weak Duality
ˆ
λ is optimal for the dual. ♦
We now proceed to a much more detailed investigation.
Lemma 3: B is a convex subset of R
m
, and M : B → R
¸
¦+∞¦ is a concave function.
Proof: Let b,
˜
b belong to B, let x ∈ Ω(b), ˜ x ∈ Ω(
˜
b), let 0 ≤ θ ≤ 1. Then (θx + (1 − θ)˜ x) ∈ X
since X is convex, and
f
i
(θx + (1 −θ)˜ x) ≤ θf
i
(x) + (1 −θ)f
i
(˜ x)
since f
i
is convex, so that
f
i
(θx + (1 −θ)˜ x) ≤ θb + (1 −θ)
˜
b , (5.24)
hence
(θx + (1 −θ)˜ x) ∈ Ω(θb + (1 −θ)
˜
b)
and therefore, B is convex.
Also, since f
0
is concave,
f
0
(θx + (1 −θ)˜ x) ≥ θf
0
(x) + (1 −θ)f
0
(˜ x) .
60 CHAPTER 5. NONLINEAR PROGRAMMING
Since (5.24) holds for all x ∈ Ω(b) and ˜ x ∈ Ω(
˜
b) it follows that
M(θb + (1 −θ)
ˆ
b) ≥ sup ¦f
0
(θx + (1 −θ)˜ x)[x ∈ Ω(b), ˜ x ∈ Ω(
˜
b)¦
≥ sup¦f
0
(x)[x ∈ Ω(b)¦ + (1 −θ) sup ¦f
0
(˜ x)[˜ x ∈ Ω(
˜
b)¦
= θM(b) + (1 −θ)M(
˜
b). ♦
Deﬁnition: Let X ⊂ R
n
and let g : X → R
¸
¦∞, −∞¦. A vector λ ∈ R
n
is said to be a
supergradient (subgradient) of g at ˆ x ∈ X if
g(x) ≤ g(ˆ x) +λ
(x − ˆ x) for x ∈ X.
(g(x) ≥ g(ˆ x) +λ
(x − ˆ x) for x ∈ X.)
(See Figure 53.)
,





.
.
.
.
M(b)
M(
ˆ
b)
b ∈ B
ˆ
b b
M is not stable at
ˆ
b
M(b)
M(
ˆ
b)
ˆ
b
b
M is stable at
ˆ
b
M(b)
M(
ˆ
b) +λ
(b −
ˆ
b)
ˆ
b b
λ is a supergradient at
ˆ
b
Figure 5.3: Illustration of supergradient of stability.
Deﬁnition: The function M : B → R
¸
¦∞¦ is said to be stable at
ˆ
b ∈ B if there exists a real
number K such that
M(b) ≤ M(
ˆ
b) +K[b −
ˆ
b[ for b ∈ B .
(In words, M is stable at
ˆ
b if M does not increase inﬁnitely steeply in a neighborhood of
ˆ
b. See
Figure 5.3.)
A more geometric way of thinking about subgradients is the following. Deﬁne the subset A ⊂
R
1+m
by
5.2. DUALITY THEORY 61
A = ¦(r, b)[b ∈ B, and r ≤ M(b)¦ .
Thus A is the set lying ”below” the graph of M. We call A the hypograph
1
of M. Since M is
concave it follows immediately that A is convex (in fact these are equivalent statements).
Deﬁnition: A vector (λ
0
, λ
1
, . . . , λ
m
) is said to be the normal to a hyperplane supporting A at a
point(ˆ r,
ˆ
b) if
λ
0
ˆ r +
m
¸
i=1
λ
i
ˆ
b
i
≥ λ
0
r +
m
¸
i=1
λ
i
b
i
for all (r, b) ∈ A . (5.25)
(In words, Alies below the hyperplane ˆ π = ¦(r, b)[λ
0
r+
¸
λ
i
b
i
= λ
0
ˆ r+
¸
λ
i
b
i
¦.) The supporting
hyperplane is said to be nonvertical if λ
0
= 0. See Figure 5.4.
Exercise 2: Show that if
ˆ
b ∈ B,
˜
b ≥
ˆ
b, and ˜ r ≤ M(
ˆ
b), then
˜
b ∈ B, M(
˜
b), and (˜ r,
˜
b) ∈ A.
Exercise 3: Assume that
ˆ
b ∈ B, and M(
ˆ
b) < ∞. Show that (i) if λ = (λ
1
, . . . , λ
m
)
is a
supergradient of M at
ˆ
b then λ ≥ 0, and (1, −λ
1
, . . . , −λ
m
)
deﬁnes a nonvertical hyperplane
supporting A at (M(
ˆ
b),
ˆ
b), (ii) if (λ
0
, −λ
1
, . . . , −λ
m
)
deﬁnes a hyperplane supporting A at
(M(
ˆ
b),
ˆ
b) then λ
0
≥ 0, λ
i
≥ 0 for 1 ≤ i ≤ m; futhermore, if the hyperplane is nonvertical then
((λ
1
/λ
0
, . . . , (λ
m
/λ
0
))
is a supergradient of M at
ˆ
b.
We will prove only one part of the next crucial result. The reader who is familiar with the
Separation Theorem of convex sets should be able to construct a proof for the second part based
on Figure 5.4, or see the Appendix at the end of this Chapter.
Lemma 4: (Gale [1967]) M is stable at
ˆ
b iff M has a supergradient at
ˆ
b. Proof: (Sufﬁciency only)
Let λ be a supergradient at
ˆ
b, then
M(b) ≤ M(
ˆ
b) +λ
(b −
ˆ
b)
≤ M(
ˆ
b) +[λ[[b −
ˆ
b[ . ♦
The next two results give important alternative interpretations of supergradients.
Lemma 5: Suppose that ˆ x is optimal for (5.17). Then
ˆ
λ is a supergradient of M at
ˆ
b iff
ˆ
λ is an
optimal price vector, and then (ˆ x,
ˆ
λ) satisfy the optimality conditions.
Proof: By hypothesis, f(ˆ x) = M(
ˆ
b), ˆ x ∈ X, and f(ˆ x) ≤
ˆ
b. Let
ˆ
λ be a supergradient of M at
ˆ
b.
By Exercise 2, (M(
ˆ
b), f(ˆ x)) ∈ A and by Exercise 3,
ˆ
λ ≥ 0 and
M(
ˆ
b) −
ˆ
λ
ˆ
b ≥ M(
ˆ
b) −
ˆ
λ
f(ˆ x) ,
so that
ˆ
λ
(f(ˆ x) −
ˆ
b) ≥ 0. But then
ˆ
λ
(
ˆ
b − f(ˆ x)) = 0, giving (5.23). Next let x ∈ X. Then
(f
0
(x), f(x)) ∈ A, hence again by Exercise 3
M(
ˆ
b) −
ˆ
λ
ˆ
b ≥ f
0
(x) −
ˆ
λ
f(x) .
Since f
0
(ˆ x) = M(
ˆ
b), and
ˆ
λ
(f(ˆ x) −
ˆ
b) = 0, we can rewrite the inequality above as
1
From the Greek “hypo” meaning below or under. This neologism contrasts with the epigraph of a function which is
the set lying above the graph of the function.
62 CHAPTER 5. NONLINEAR PROGRAMMING
.
.
M(
ˆ
b)
M(b)
ˆ
b
A
b
No nonvertical hyperplane supporting A at (M(
ˆ
b),
ˆ
b)
(λ
0
, . . . , λ
m
)
M(
ˆ
b)
M(b)
ˆ π
A
b
ˆ
b
ˆ π is a nonvertical hyperplane supporting A at (M(
ˆ
b),
ˆ
b)
Figure 5.4: Hypograph and supporting hyperplane.
f
0
(ˆ x) +
ˆ
λ
(f(ˆ x) −
ˆ
b) ≥ f
0
(x) −
ˆ
λ
(f(x) −
ˆ
b) ,
so that (5.21) holds. It follows that (ˆ x,
ˆ
λ) satisfy the optimality conditions.
Conversely, suppose ˆ x ∈ X,
ˆ
λ ≥ 0 satisfy (5.21), (5.22), and (5.23). Let x ∈ Ω(b), i.e.,
x ∈ X, f(x) ≤ b. Then
ˆ
λ
(f(x) −b) ≤ 0 so that
f
0
(x) ≤ f
0
(x)
ˆ
λ
(f(x) −b)
= f
0
(x) −
ˆ
λ
(f(x) −
ˆ
b) +
ˆ
λ
(b −
ˆ
b)
≤ f
0
(ˆ x) −
ˆ
λ
(f(ˆ x) −
ˆ
b) +
ˆ
λ
(b −
ˆ
b) by (5.21)
= f
0
(ˆ x) +
ˆ
λ
(b −
ˆ
b) by (5.23)
= M(
ˆ
b) +
ˆ
λ
(b −
ˆ
b) .
Hence
M(b) = sup¦f
0
(x)[x ∈ Ω(b)¦ ≤ M(
ˆ
b) +
ˆ
λ
(b −
ˆ
b) ,
so that
ˆ
λ
is a supergradient of M at
ˆ
b. ♦
Lemma 6: Suppose that
ˆ
b ∈ B, and M(
ˆ
b) < ∞. Then
ˆ
λ is a supergradient of M at
ˆ
b iff
ˆ
λ is an
optimal solution of the dual (5.19) and m(
ˆ
λ) = M(
ˆ
b).
Proof: Let
ˆ
λ be a supergradient of M at
ˆ
b. Let x ∈ X. By Exercises 2 and 3
5.2. DUALITY THEORY 63
M(
ˆ
b) −
ˆ
λ
ˆ
b ≥ f
0
(x) −
ˆ
λ
f(x)
or
M(
ˆ
b) ≥ f
0
(x) −
ˆ
λ
(f(x) −
ˆ
b) ,
so that
M(
ˆ
b) ≥ sup¦f
0
(x) −
ˆ
λ
(f(x) −
ˆ
b)[x ∈ X¦ = m(
ˆ
λ) .
By weak duality (Lemma 2) it follows that M(
ˆ
b) = m(
ˆ
λ) and
ˆ
λ is optimal for (5.19).
Conversely suppose
ˆ
λ ≥ 0, and m(
ˆ
λ) = M(
ˆ
b). Then for any x ∈ X
M(
ˆ
b) ≥ f
0
(x) −
ˆ
λ
(f(x) −
ˆ
b) ,
and if moreover f(x) ≤ b, then
ˆ
λ
(f(x) −b) ≤ 0, so that
M(
ˆ
b) ≥ f
0
(x) −
ˆ
λ
(f(x) −
ˆ
b) +
ˆ
λ
(f(x) −b)
= f
0
(x) −
ˆ
λ
b +
ˆ
λ
ˆ
b for x ∈ Ω(b) .
Hence,
M(b) = sup¦f
0
(x)[x ∈ Ω(b)¦ ≤ M(
ˆ
b) +
ˆ
λ
(b −
ˆ
b) ,
so that
ˆ
λ is a supergradient. ♦
We can now summarize our results as follows.
Theorem 2: (Duality) Suppose
ˆ
b ∈ B, M(
ˆ
b) < ∞, and M is stable at
ˆ
b. Then
(i) there exists an optimal solution
ˆ
λ for the dual, and m(
ˆ
λ) = M(
ˆ
b),
(ii)
ˆ
λ is optimal for the dual iff
ˆ
λ is a supergradient of M at
ˆ
b,
(iii) if
ˆ
λ is any optimal solution for the dual, then ˆ x is optimal for the primal iff (ˆ x,
ˆ
λ) satisfy the
optimality conditions of (5.21), (5.22), and (5.23).
Proof: (i) follows from Lemmas 4,6. (ii) is implied by Lemma 6. The “if” part of (iii) follows from
Theorem 1, whereas the “only if” part of (iii) follows from Lemma 5. ♦
Corollary 1: Under the hypothesis of Theorem 2, if
ˆ
λ is an optimal solution to the dual then
(∂M
+
/∂b
i
)(
ˆ
b) ≤
ˆ
λ
i
≤ (∂M
−
/∂b
i
)(
ˆ
b).
Exercise 4: Prove Corollary 1. (Hint: See Theorem 5 of 4.2.3.)
5.2.2 Interpretation and extensions.
It is easy to see using convexity properties that, if X = R
n
and f
i
, 0 ≤ i ≤ m, are differentiable,
then the optimality conditions (5.21), (5.22), and (5.23) are equivalent to the KuhnTucker condition
(5.8). Thus the condition of stability of M at
ˆ
b plays a similar role to the constraint qualiﬁcation.
However, by Lemmas 4, 6 stability is equivalent to the existence of optimal dual variables, whereas
CQ is only a sufﬁcient condition. In other words if CQ holds at ˆ x then M is stable at
ˆ
b. In particular,
if X = R
n
and the f
i
are differentiable, the various conditions of Section 1.3 imply stability. Here
we give one sufﬁcient condition which implies stability for the general case.
Lemma 7: If
ˆ
b is in the interior of B, in particular if there exists x ∈ X such that f
i
(x) <
ˆ
b
i
for
1 ≤ i ≤ m, then M is stable at
ˆ
b.
64 CHAPTER 5. NONLINEAR PROGRAMMING
The proof rests on the Separation Theorem for convex sets, and only depends on the fact that M
is concave, M(
ˆ
b) < ∞ without loss of generality, and
ˆ
b is the interior of B. For details see the
Appendix.
Much of duality theory can be given an economic interpretation similar to that in Section 4.4.
Thus, we can think of x as the vector of n activity levels, f
0
(x) the corresponding revenue, X as
constraints due to physical or longterm limitations, b as the vector of current resource supplies,
and ﬁnally f(x) the amount of these resources used up at activity levels x. The various convexity
conditions are generalizations of the economic hypothesis of nonincreasing returnstoscale. The
primal problem (5.17) is the shortterm decision problem faced by the ﬁrm. Next, if the current
resources can be bought or sold at prices
ˆ
λ = (λ
1
, . . . , λ
m
)
, the ﬁrm faces the decision problem
(5.18). If for a price system
ˆ
λ, an optimal solution of (5.17) also is an optimal solution for (5.18),
then we can interpret
ˆ
λ as a system of equilibrium prices just as in 4.2. Assuming the realistic
condition
ˆ
b ∈ B, M(
ˆ
b) < ∞ we can see from Theorem 2 and its Corollary 1 that there exists
an equilibrium price system iff (∂M
+
/∂b
i
)(
ˆ
b) < ∞, 1 ≤ i ≤ m; if we interpret (∂M
+
/∂b
i
)(
ˆ
b)
as the marginal revenue of the ith resource, we can say that equilibrium prices exist iff marginal
productivities of every (variable) resource is ﬁnite. These ideas are developed in (Gale [1967]).
.
M(b)
A
b
ˆ
b
M(
ˆ
b)
Figure 5.5: If M is not concave there may be no supporting hyperplane at (M(
ˆ
b),
ˆ
b).
Referring to Figure 5.3 or Figure 5.4, and comparing with Figure 5.5 it is evident that if M is not
concave or, equivalently, if its hypograph Ais not convex, there may be no hyperplane supporting A
at (M(
ˆ
b),
ˆ
b). This is the reason why duality theory requires the often restrictive convexity hypoth
esis on X and f
i
. It is possible to obtain the duality theorem under conditions slightly weaker than
convexity but since these conditions are not easily veriﬁable we do not pursue this direction any fur
ther (see Luenberger [1968]). A much more promising development has recently taken place. The
basic idea involved is to consider supporting A at (M(
ˆ
b),
ˆ
b) by (nonvertical) surfaces ˆ π more gen
eral than hyperplanes; see Figure 5.6. Instead of (5.18) we would then have more general problem
of the form (5.26):
Maximize f
0
(x) −F(f(x) −
ˆ
b)
subject to x ∈ X ,
(5.26)
5.2. DUALITY THEORY 65
where F : R
m
→ R is chosen so that ˆ π (in Figure 5.6) is the graph of the function b → M(
ˆ
b) −
F(b −
ˆ
b). Usually F is chosen from a class of functions φ parameterized by µ = (µ
1
, . . . , µ
k
) ≥ 0.
Then for each ﬁxed µ ≥ 0 we have (5.27) instead of (5.26):
Maximize f
0
(x) −φ(µ; f(x) −
ˆ
b)
subject to x ∈ X .
(5.27)
.
M(b)
ˆ π
A
b
ˆ
b
M(
ˆ
b)
Figure 5.6: The surface ˆ π supports A at (M(
ˆ
b),
ˆ
b).
If we let
ψ(µ) =sup¦f
0
(x) −φ(µ; f(x) −
ˆ
b)[x ∈ X¦ .
then the dual problem is
Minimize ψ(µ)
subject to µ ≥ 0 ,
in analogy with (5.19).
The economic interpretation of (5.27) would be that if the prevailing (nonuniform) price system
is φ(µ; ) then the resources f(x) −
ˆ
b can be bought (or sold) for the amount φ(µ; f(x) −
ˆ
b). For
such an interpretation to make sense we should have φ(µ; b) ≥ 0 for b ≥ 0, and φ(µ; b) ≥ φ(µ;
˜
b)
whenever b ≥
˜
b. A relatively unnoticed, but quite interesting development along these lines is
presented in (Frank [1969]). Also see (Arrow and Hurwicz [1960]).
For noneconomic applications, of course, no such limitation on φ is necessary. The following
references are pertinent: (Gould [1969]), (Greenberg and Pierskalla [1970]), (Banerjee [1971]). For
more details concerning the topics of 2.1 see (Geoffrion [1970a]) and for a mathematically more
elegant treatment see (Rockafellar [1970]).
5.2.3 Applications.
Decentralized resource allocation.
Parts (i) and (iii) of Theorem 2 make duality theory attractive for computation purposes. In particular
from Theorem 2 (iii), if we have an optimal dual solution
ˆ
λ then the optimal primal solutions are
those optimal solutions of (5.18) for λ =
ˆ
λ which also satisfy the feasibility condition (5.22) and
66 CHAPTER 5. NONLINEAR PROGRAMMING
the “complementary slackness” condition (5.23). This is useful because generally speaking (5.18)
is easier to solve than (5.17) since (5.18) has fewer constraints.
Consider a decision problem in a large system (e.g., a multidivisional ﬁrm). The system is
made up of k subsystems (divisions), and the decision variable of the ith subsystem is a vector
x
i
∈ R
n
i
, 1 ≤ i ≤ k. The subsystem has individual constraints of the form x
i
∈ X
i
where x
i
is
a convex set. Furthermore, the subsystems share some resources in common and this limitation is
expressed as f
1
(x
1
) +. . . +f
k
(x
k
) ≤
ˆ
b where f
i
: R
n
i
→ R
m
are convex functions and
ˆ
b ∈ R
m
is the vector of available common resources. Suppose that the objective function of the large system
is additive, i.e. it is the form f
1
0
(x
1
) + . . . + f
k
0
(x
k
) where f
i
0
: R
n
i
→ R are concave functions.
Thus we have the decision problem (5.28):
Maximize
k
¸
i=1
f
i
0
(x
i
)
subject to x
i
∈ X
i
, 1 ≤ i ≤ k,
k
¸
i=1
f
i
(x
i
) ≤
ˆ
b .
(5.28)
For λ ∈ R
m
, λ ≥ 0, the problem corresponding to (5.19) is
Maximize f
i
0
(x
i
) −λ
f
i
(x
i
) −λ
(
k
¸
i=1
f
i
(x
i
) −
ˆ
b)
subject to x
i
∈ X
i
, 1 ≤ i ≤ k ,
which decomposes into k separate problems:
Maximize f
i
0
(x
i
) −λ
f
i
(x
i
)
subject to x
i
∈ X
i
, 1 ≤ i ≤ k .
(5.29)
If we let m
i
(λ) = sup¦f
i
0
(x
i
) −λ
f
i
(x
i
)[x
i
∈ X
i
¦, and m(λ) =
k
¸
i=1
m
i
(λ) +λ
ˆ
b, then the dual
problem is
Minimize m(λ) ,
subject to λ ≥ 0 .
(5.30)
Note that (5.29) may be much easier to solve than (5.28) because, ﬁrst of all, (5.29) involves fewer
constraints, but perhaps more importantly the decision problems in (5.29) are decentralized whereas
in (5.28) all the decision variables x
1
, . . . , x
k
are coupled together; in fact, if k is very large it may
be practically impossible to solve (5.28) whereas (5.29) may be trivial if the dimensions of x
i
are
small.
Assuming that (5.28) has an optimal solution and the stability condition is satisﬁed, we need to
ﬁnd an optimal dual solution so that we can use Theorem 2(iii). For simplicity suppose that the
f
i
0
, 1 ≤ i ≤ k, are strictly concave, and also suppose that (5.29) has an optimal solution for every
λ ≥ 0. Then by Exercise 8 of Section 1, for each λ ≥ 0 there is a unique optimal solution of (5.29),
say x
i
(λ). Consider the following algorithm.
5.2. DUALITY THEORY 67
Step 1. Select λ
0
≥ 0 arbitrary. Set p = 0, and go to Step 2.
Step 2. Solve (5.29) for λ = λ
p
and obtain the optimal solution x
p
= (x
1
(λ
p
), . . . , x
k
(λ
p
)).
Compute e
p
=
k
¸
i=1
f
i
(x
i
(λ
p
)) −
ˆ
b. If e
p
≥ 0, x
p
is feasible for (5.28) and can easily be seen to be
optimal.
Step 3. Set λ
p=1
according to
λ
p+1
i
=
λ
p
i
if e
p
i
≥ 0
λ
p
i
−d
p
e
p
i
if e
p
i
< 0
where d
p
> 0 is chosen a priori. Set p = p + 1 and return to Step 3.
It can be shown that if the step sizes d
p
are chosen properly, x
p
will converge to the optimum
solution of (5.28). For more detail see (Arrow and Hurwicz [1960]), and for other decentralization
schemes for solving (5.28) see (Geoffrion [1970b]).
Control of water quality in a stream.
The discussion in this section is mainly based on (Kendrick, et al., [1971]). For an informal discus
sion of schemes of pollution control which derive their effectiveness from duality theory see (Solow
[1971]). See (Dorfman and Jacoby [1970].)
Figure 5.7 is a schematic diagram of a part of a stream into which n sources (industries and
municipalities) discharge polluting efﬂuents. The pollutants consist of various materials, but for
simplicity of exposition we assume that their impact on the quality of the stream is measured in
terms of a single quantity, namely the biochemical oxygen demand (BOD) which they place on the
dissolved oxygen (DO) in the stream. Since the DO in the stream is used to breakdown chemically
the pollutants into harmless substances, the quality of the stream improves with the amount of
DO and decreases with increasing BOD. It is a welladvertized fact that if the DO drops below a
certain concentration, then life in the stream is seriously threatened; indeed, the stream can “die.”
Therefore, it is important to treat the efﬂuents before they enter the stream in order to reduce the
BOD to concentration levels which can be safely absorbed by the DO in the stream. In this example
we are concerned with ﬁnding the optimal balance between costs of waste treatment and costs of
high BOD in the stream.
We ﬁrst derive the equations which govern the evolution in time of BOD and DO in the n areas
of the streams. The ﬂuctuations of BOD and DO will be cyclical with a period of 24 hours. Hence,
it is enough to study the problem over a 24hour period. We divide this period into T intervals,
t = 1, . . . , T. During interval t and in area i let
z
i
(t) = concentration of BOD measured in mg/liter,
q
i
(t) = concentration of DO measured in mg/liter,
s
i
(t) = concentration of BOD of efﬂuent discharge in mg/liter, and
m
i
(t) = amount of efﬂuent discharge in liters.
The principle of conservation of mass gives us equations (5.31) and (5.32):
z
i
(t + 1) −z
i
(t) = −α
i
z
i
(t) +
ψ
i−1
z
i−1
(t)
v
i
−
ψ
i
z
i
(t)
v
i
+
s
i
(t)m
i
(t)
v
i
, (5.31)
q
i
(t + 1) −q
i
(t) = β
i
(q
s
i
−q
i
(t)) +
ψ
i−1
q
i−1
(t)
v
i
−
ψ
i
q
i
(t)
v
i
+α
i
z
i
(t) −η
i
v
i
, t = 1, . . . , T and i = 1, . . . , N.
(5.32)
68 CHAPTER 5. NONLINEAR PROGRAMMING
. . . . . .
. . .
direction of ﬂow
0
z
0
q
0
1
z
1
q
1
i −1
z
i−1
q
i−1
i
z
i
q
i
i + 1
z
i+1
q
i+1
N
z
N
q
N
N + 1
given
(1 −π
1
)s
i
s
i
1 −π
i−1
s
i−a
s
i
(1 −π
i
)s
i
s
i+1
(1 −π
i+1
)s
i+1
s
N
(1 −π
N
)s
N
Figure 5.7: Schematic of stream with efﬂuent discharges.
Here, v
i
= volume of water in area i measured in liters, ψ
i
= volume of water which ﬂows from
area i to are i +1 in each period measured in liters. α
i
is the rate of decay of BOD per interval. This
decay occurs by combination of BOD and DO. β
i
is the rate of generation of DO. The increase in
DO is due to various natural oxygenproducing biochemical reactions in the stream and the increase
is proportional to (q
s
− q
i
) where q
s
is the saturation level of DO in the stream. Finally, η
i
is the
DO requirement in the bottom sludge. The v
i
, ψ
i
, α
i
, η
i
, q
s
are parameters of the stream and are
assumed known. They may vary with the time interval t. Also z
0
(t), q
0
(t) which are the concen
trations immediately upstream from area 1 are assumed known. Finally, the initial concentrations
z
i
(1), q
i
(1), i = 1, . . . , N are assumed known.
Now suppose that the waste treatment facility in area i removes in interval t a fraction π
i
(t) of
the concentration s
i
(t) of BOD. Then (5.31) is replaced by
z
i
(t + 1) −z
i
(t) = −α
i
z
i
(t) +
ψ
i
z
i−1
v
i
−
ψ
i
z
i
(t)
v
i
+
(1−π
i
(t))s
i
(t)m
i
(t)
v
i
. (5.33)
We now turn to the costs associated with waste treatment and pollution. The cost of waste treat
ment can be readily identiﬁed. In period t the ith facility treats m
i
(t) liters of efﬂuent with a BOD
concentration s
i
(t) mg/liter of which the facility removes a fraction π
i
(t). Hence, the cost in period
t will be f
i
(π
i
(t), s
i
(t), m
i
(t)) where the function must be monotonically increasing in all of its
arguments. We further assume that f is convex.
The costs associated with increased amounts of BOD and reduced amounts of DO are much
more difﬁcult to quantify since the stream is used by many institutions for a variety of purposes
(e.g., agricultural, industrial, municipal, recreational), and the disutility caused by a decrease in
the water quality varies with the user. Therefore, instead of attempting to quantify these costs let
us suppose that some minimum water quality standards are set. Let q be the minimum acceptable
DO concentration and let ¯ z be the maximum permissible BOD concentration. Then we face the
5.2. DUALITY THEORY 69
following NP:
Maximize −
N
¸
i=1
T
¸
t=1
f
i
(π
i
(t), s
i
(t), m
i
(t))
subject to (5.32), (5.33), and
−q
i
(t) ≤ −q , i = 1, . . . , N; t = 1, . . . , T,
z
i
(t) ≤ ¯ z , i = 1, . . . , N; t = 1, . . . , T,
0 ≤ π
i
(t) ≤ 1 , i = 1, . . . , N; t = 1, . . . , T.
(5.34)
Suppose that all the treatment facilities are in the control of a single public agency. Then assuming
that the agency is required to maintain the standards (q, ¯ z) and it does this at a minimum cost it will
solve the NP (5.34) and arrive at an optimal solution. Let the minimum cost be m(q, ¯ z). But if
there is no such centralized agency, then the individual polluters may not (and usually do not) have
any incentive to cooperate among themselves to achieve these standards. Furthermore, it does not
make sense to enforce legally a minimum standard q
i
(t) ≥ q, z
i
(t) ≤ ¯ z on every polluter since the
pollution levels in the ith area depend upon the pollution levels on all the other areas lying upstream.
On the other hand, it may be economically and politically acceptable to tax individual polluters in
proportion to the amount of pollutants discharged by the individual. The question we now pose
is whether there exist tax rates such that if each individual polluter minimizes its own total cost
(i.e., cost of waste treatment + tax on remaining pollutants), then the resulting water quality will be
acceptable and, furthermore, the resulting amount of waste treatment is carried out at the minimum
expenditure of resources (i.e., will be an optimal solution of (5.34)).
It should be clear from the duality theory that the answer is in the afﬁrmative. To see this let
w
i
(t) = (z
i
(t), −q
i
(t))
, let w(t) = (w
1
(t), . . . , w
N
(t)), and let w = (w(1), . . . , w(t)). Then we
can solve (5.32) and (5.33) for w and obtain
w = b +Ar , (5.35)
where the matrix A and the vector b depend upon the known parameters and initial conditions, and
r is the NTdimensional vector with components (1 − π
i
(t))s
i
(t)m
i
(t). Note that the coefﬁcients
of the matrix must be nonnegative because an increase in any component of r cannot decrease the
BOD levels and cannot increase the DO levels. Using (5.35) we can rewrite (5.34) as follows:
Maximize −
¸
i
¸
t
f
i
(π
i
(t), s
i
(t), m
i
(t))
subject to b +Ar ≤ ¯ w ,
0 ≤ π
i
(t) ≤ 1 , i = 1, . . . , N; t = 1, . . . , T,
(5.36)
where the 2NTdimensional vector ¯ w has its components equal to −q or ¯ z in the obvious manner.
By the duality theorem there exists a 2NTdimensional vector λ
∗
≥ 0, and an optimal solution
π
∗
i
(t), i = 1, . . . , N, t = 1, . . . , T, of the problem:
Maximize −
¸
i
¸
t
f
i
(π
i
(t), s
i
(t), m
i
(t)) −λ
∗
(b +Ar −w)
subject to 0 ≤ π
i
(t) ≤ 1, i = 1, . . . , N; t = 1, . . . , T ,
(5.37)
such that ¦π
∗
i
(t)¦ is also an optimal solution of (5.36) and, furthermore, the optimal values of (5.36)
and (5.37) are equal. If we let p
∗
= A
λ
∗
≥ 0, and we write the components of p
∗
as p
∗
i
(t) to match
70 CHAPTER 5. NONLINEAR PROGRAMMING
with the components (1−π
i
(t))s
i
(t)m
i
(t) of r we can see that (5.37) is equivalent to the set of NT
problems:
Maximize −f
i
(π
i
(t), s
i
(t), m
i
(t)) −p
∗
i
(t)(1 −π
i
(t))s
i
(t)m
i
(t)
0 ≤ π
i
(t) ≤ 1 ,
i = 1, . . . , N; t = 1, . . . , T .
(5.38)
Thus, p
∗
i
(t) is optimum tax per mg of BOD in area i during period t.
Before we leave this example let us note that the optimum dual variable or shadow price λ
∗
plays an important role in a larger framework. We noted earlier that the quality standard (q, ¯ z)
was somewhat arbitrary. Now suppose it is proposed to change the standard in the ith area during
period t to q +∆q
i
(t) and ¯ z +∆z
i
(t). If the corresponding components of λ
∗
are λ
q∗
i
(t) and λ
z∗
i
(t),
then the change in the minimum cost necessary to achieve the new standard will be approximately
λ
q∗
i
(t)∆q
i
(t) + λ
z∗
i
(t)∆z
i
(t). This estimate can now serve as a basis in making a beneﬁts/cost
analysis of the proposed new standard.
5.3 Quadratic Programming
An important special case of NP is the quadratic programming (QP) problem:
Maximize c
x −
1
2
x
Px
subject to Ax ≤ b, x ≥ 0 ,
(5.39)
where x ∈ R
n
is the decision variable , c ∈ R
n
, b ∈ R
m
are ﬁxed, A is a ﬁxed m n matrix and
P = P
is a ﬁxed positive semideﬁnite matrix.
Theorem 1: A vector x
∗
∈ R
n
is optimal for (5.39) iff there exist λ
∗
∈ R
m
, µ
∗
∈ R
n
, such that
Ax
∗
≤ b, x
∗
≥ 0
c −Px
∗
= A
λ
∗
−µ
∗
, λ
∗
≥ 0, µ
∗
≥ 0 ,
(λ
∗
)
(Ax
∗
−b) = 0 , (µ
∗
)
x
∗
= 0 .
(5.40)
Proof: By Lemma 3 of 1.3, CQ is satisﬁed, hence the necessity of these conditions follows from
Theorem 2 of 1.2. On the other hand, since P is positive semideﬁnite it follows from Exercise 6
of Section 1.2 that f
0
: x → c
x −1/2 x
Px is a concave function, so that the sufﬁciency of these
conditions follows from Theorem 4 of 1.2. ♦
From (5.40) we can see that x
∗
is optimal for (5.39) iff there is a solution (x
∗
, y
∗
, λ
∗
, µ
∗
) to
(5.41), (5.42), and (5.43):
Ax +I
m
Y = b
−Px −A
λ +I
n
µ = −c ,
(5.41)
x ≥ 0 y ≥ 0, λ ≥ 0, µ ≥ 0 , (5.42)
µ
x = 0 , λ
y = 0 . (5.43)
Suppose we try to solve (5.41) and (5.42) by Phase I of the Simplex algorithm (see 4.3.2). Then we
must apply Phase II to the LP:
Maximize −
m
¸
i=1
z
i
−
n
¸
j=1
ξ
j
5.4. COMPUTATIONAL METHOD 71
subject to
Ax +I
m
y +z = b
−Px −A
λ +I
n
µ +ξ = −c
x ≥ 0, y ≥ 0, λ ≥ 0, µ ≥ 0, z ≥ 0, ξ ≥ 0,
(5.44)
starting with a basic feasible solution z = b, ξ = −c. (We have assumed, without loss of generality,
that b ≥ 0 and −c ≥ 0.) If (5.41) and (5.42) have a solution then the maximum value in (5.44) is 0.
We have the following result.
Lemma 1: If (5.41), (5.42), and (5.43) have a solution, then there is an optimal basic feasible solution
of (5.44) which is also a solution f (5.41), (5.42), and (5.43).
Proof: Let ˆ x, ˆ y,
ˆ
λ, ˆ µ be a solution of (5.41), (5.42), and (5.43). Then ˆ x, ˆ y,
ˆ
λ, ˆ µ, ˆ z = 0,
ˆ
ξ = 0 is
an optimal solution of (5.44). Furthermore, from (5.42) and (5.43) we see that at most (n + m)
components of (ˆ x, ˆ y,
ˆ
λ, ˆ µ) are nonzero. But then a repetition of the proof of Lemma 1 of 4.3.1 will
also prove this lemma. ♦
This lemma suggests that we can apply the Simplex algorithm of 4.3.2 to solve (5.44), starting
with the basic feasible solution z = b, ξ = −c, in order to obtain a solution of (5.41), (5.42), and
(5.43). However, Step 2 of the Simplex algorithm must be modiﬁed as follows to satisfy (5.43):
If a variable x
j
is currently in the basis, do not consider µ
j
as a candidate for entry into the basis;
if a variable y
i
is currently in the basis, do not consider λ
i
as a candidate for entry into the basis. If
it not possible to remove the z
i
and ξ
j
from the basis, stop.
The above algorithm is due to Wolfe [1959]. The behavior of the algorithm is summarized below.
Theorem 2: Suppose P is positive deﬁnite. The algorithm will stop in a ﬁnite number of steps at an
optimal basic feasible solution (ˆ x, ˆ y,
ˆ
λ, ˆ µ, ˆ z,
ˆ
ξ) of (5.44). If ˆ z = 0 and
ˆ
ξ = 0 then (ˆ x, ˆ y,
ˆ
λ, ˆ µ) solve
(5.41), (5.42), and (5.43) and ˆ x is an optimal solution of (5.39). If ˆ z = 0 or
ˆ
ξ = 0, then there is no
solution to (5.41), (5.42), (5.43), and there is no feasible solution of (5.39).
For a proof of this result as well as for a generalization of the algorithm which permits positive
semideﬁnite P see (Cannon, Cullum, and Polak [1970], p. 159 ff).
5.4 Computational Method
We return to the general NP (5.45),
Maximize f
0
(x)
subject to f
i
(x) ≤ 0, i = 1, . . . , m ,
(5.45)
where x ∈ R
n
, f
i
: R
n
→ R, 0 ≤ i ≤ m, are differentiable. Let Ω ⊂ R
n
denote the set of
feasible solutions. For ˆ x ∈ Ω deﬁne the function ψ(ˆ x) : R
n
→ R by
ψ(ˆ x)(h) = max¦−f
0x
(ˆ x)h, f
1
(ˆ x) +f
1x
(ˆ x)h, . . . , f
m
(ˆ x) +f
mx
(ˆ x)h¦.
Consider the problem:
Minimize ψ(ˆ x)(h)
subject to −ψ(ˆ x)(h) −f
0x
(ˆ x)h ≤ 0 ,
−ψ(ˆ x)(h) +f
i
(ˆ x)f
ix
h ≤ 0 ,
1 ≤ i ≤ m , −1 ≤ h
j
≤ 1 , 1 ≤ j ≤ n .
(5.46)
72 CHAPTER 5. NONLINEAR PROGRAMMING
.
f
0
(x) = F
0
(x
∗
) > f
0
(x
k
)
f
0
(x) = f
0
(x
k
)
f
2
= 0
f
1
= 0
Ω
f
3
= 0
f
2
(x
k
)
x
k
f
3
(x
k
)
f
1
(x
k
)
f
0
(x
k
)
h(x
k
)
Figure 5.8: h(x
k
) is a feasible direction.
Call h(ˆ x) an optimum solution of (5.46) and let h
0
(ˆ x) = ψ(ˆ x)(h(ˆ x)) be the minimum value at
tained. (Note that by Exercise 1 of 4.5.1 (5.46) can be solved as an LP.)
The following algorithm is due to Topkis and Veinott [1967].
Step 1. Find x
0
∈ Ω, set k = 0, and go to Step 2.
Step 2. Solve (5.46) for ˆ x = x
k
and obtain h
0
(x
k
), h(x
k
). If h
0
(x
k
) = 0, stop, otherwise go to Step
3.
Step 3. Compute an optimum solution µ(x
k
) to the onedimensional problem,
Maximize f
0
(x
k
+µh(x
k
)) ,
subject to (x
k
+µh(x
k
)) ∈ Ω, µ ≥ 0 ,
and go to Step 4.
Step 4. Set x
k+1
= x
k
+µ(x
k
)h(x
k
), set k = k + 1 and return to Step 2.
The performance of the algorithm is summarized below.
Theorem 1: Suppose that the set
Ω(x
0
) = ¦x[x ∈ Ω, f
0
(x) ≥ f
0
(x
0
)¦
is compact, and has a nonempty interior, which is dense in Ω(x
0
). Let x
∗
be any limit point of
the sequence x
0
, x
1
, . . . , x
k
, . . . , generated by the algorithm. Then the KuhnTucker conditions are
satisﬁed at x
∗
.
For a proof of this result and for more efﬁcient algorithms the reader is referred to (Polak [1971]).
Remark: If h
0
(x
k
) < 0 in Step 2, then the direction h(x
k
) satisﬁes f
0x
(x
k
)h(x
k
) > 0, and f
i
(x
k
)+
f
ix
(x
K
)h(x
k
) < 0, 1 ≤ i ≤ m. For this reason h(x
k
) is called a (desirable) feasible direction.
(See Figure 5.8.)
5.5. APPENDIX 73
5.5 Appendix
The proofs of Lemmas 4,7 of Section 2 are based on the following extremely important theorem
(see Rockafeller [1970]).
Separation theorem for convex sets. Let F, Gbe convex subsets of R
n
such that the relative interiors
of F, G are disjoint. Then there exists λ ∈ R
n
, λ = 0, and θ ∈ R such that
λ
g ≤ θ for all g ∈ G
λ
f ≥ θ for all f ∈ F .
Proof of Lemma 4: Since M is stable at
ˆ
b there exists K such that
M(b) −M(
ˆ
b) ≤ K[b −
ˆ
b[ for all b ∈ B . (5.47)
In R
1+m
consider the sets
F = ¦(r, b)[b ∈ R
m
, r > K[b −
ˆ
b[¦ ,
G = ¦(r, b)[b ∈ B, r ≤ M(b) −M(
ˆ
b)¦ .
It is easy to check that F, G are convex, and (5.47) implies that F ∩ G = φ. Hence, there exist
(λ
0
, . . . , λ
m
) = 0, and θ such that
λ
0
r +
m
¸
i=1
λ
i
b
i
≤ θ for (r, b) ∈ G ,
λ
0
r +
m
¸
i=1
λ
i
b
i
≥ θ for (r, b) ∈ F .
(5.48)
From the deﬁnition of F, and the fact that (λ
0
, . . . , λ
m
) = 0, it can be veriﬁed that (5.49) can hold
only if λ
0
> 0. Also from (5.49) we can see that
m
¸
i=1
λ
i
ˆ
b
i
≥ θ, whereas from (5.48)
m
¸
i=1
λ
i
ˆ
b
i
≤ θ,
so that
m
¸
i=1
λ
i
ˆ
b
i
= θ. But then from (5.48) we get
M(b) −M(
ˆ
b) ≤
1
λ
0
[θ −
m
¸
i=1
λ
i
b
i
] =
m
¸
i=1
(−
λ
i
λ
0
)(b
i
−
ˆ
b). ♦
Proof of Lemma 7: Since
ˆ
b is in the interior of B, there exists ε > 0 such that
b ∈ B whenever [b −
ˆ
b[ < ε . (5.49)
In R
1+m
consider the sets
F = ¦(r,
ˆ
b)[r > M(
ˆ
b¦
G = ¦(r, b)[b ∈ B, r ≤ M(b)¦ .
Evidently, F, G are convex and F ∩ G = φ, so that there exist (λ
0
, . . . , λ
m
) = 0, and θ such that
λ
0
r +
m
¸
i=1
λ
i
ˆ
b
i
≥ θ , for r > M(
ˆ
b) , (5.50)
74 CHAPTER 5. NONLINEAR PROGRAMMING
λ
0
r +
m
¸
i=1
λ
i
ˆ
b
i
≤ θ , for (r, b) ∈ G . (5.51)
From (5.49), and the fact that (λ
0
, . . . , λ
m
) = 0 we can see that (5.50) and (5.51) imply λ
0
> 0.
From (5.50),(5.51) we get
λ
0
M(
ˆ
b) +
m
¸
i=1
λ
i
ˆ
b
i
= θ ,
so that (5.52) implies
M(b) ≤ (
ˆ
b) +
m
¸
i=1
(−
λ
i
λ
0
)(b
i
−
ˆ
b
i
) . ♦
Chapter 6
SEQUENTIAL DECISION PROBLEMS:
DISCRETETIME OPTIMAL
CONTROL
In this chapter we apply the results of the last two chapters to situations where decisions have to be
made sequentially over time. A very important class of problems where such situations arise is in
the control of dynamical systems. In the ﬁrst section we give two examples, and in Section 2 we
derive the main result.
6.1 Examples
The trajectory of a vertical sounding rocket is controlled by adjusting the rate of fuel ejection which
generates the thrust force. Speciﬁcally suppose that the equations of motion are given by (6.1).
˙ x
1
(t) = x
2
(t)
˙ x
2
(t) = −
C
D
x
3
(t)
ρ(x
1
(t))x
2
2
(t) −g +
C
T
x
3
(t)
u(t)
˙ x
3
(t) = −u(t) ,
(6.1)
where x
1
(t) is the height of the rocket from the ground at time t, x
2
(t) is the (vertical) speed at
time t, x
3
(t) is the weight of the rocket (= weight of remaining fuel) at time t. The “dot” denotes
differentiation with respect to t. These equations can be derived from the force equations under the
assumption that there are four forces acting on the rocket, namely: inertia = x
3
¨ x
1
= x
3
˙ x
2
; drag
force = C
D
ρ(x
1
)x
2
2
where C
D
is constant, ρ(x
1
) is a friction coefﬁcient depending on atmospheric
density which is a function of x
1
; gravitational force = gx
3
with g assumed constant; and thrust
force C
T
˙ x
3
, assumed proportional to rate of fuel ejection. See Figure 6.1. The decision variable at
time t is u(t), the rate of fuel ejection. At time 0 we assume that (x
1
(0), x
2
(0), x
3
(0)) = (0, 0, M);
that is, the rocket is on the ground, at rest, with initial fuel of weight M. At a prescribed ﬁnal time
t
f
, it is desired that the rocket be at a position as high above the ground as possible. Thus, the
75
76 CHAPTER 6. DISCRETETIME OPTIMAL CONTROL
decision problem can be formalized as (6.2).
Maximize x
1
(t
f
)
subject to ˙ x(t) = f(x(t), u(t)), 0 ≤ t ≤ t
f
x(0) = (0, 0, M)
u(t) ≥ 0, x
3
(t) ≥ 0, 0 ≤ t ≤ t
f
,
(6.2)
where x = (x
1
, x
2
, x
3
)
, f : R
3+1
→ R
3
is the righthand side of (6.1). The constraint inequalities
u(t) ≥ 0 and x
3
(t) ≥ 0 are obvious physical constraints.
x
3
¨ x
1
= inertia
C
D
ϕ(x
1
)x
2
2
= drag
gx
3
= gravitational force
C
R
˙ x
3
= thrust
Figure 6.1: Forces acting on the rocket.
The decision problem (6.2) differs from those considered so far in that the decision variables,
which are functions u : [0, t
f
] → R, cannot be represented as vectors in a ﬁnitedimensional
space. We shall treat such problems in great generality in the succeeding chapters. For the moment
we assume that for computational or practical reasons it is necessary to approximate or restrict
the permissible function u() to be constant over the intervals [0, t
1
), [t
1
, t
2
), . . . , [t
N−1
, t
f
), where
t
1
, t
2
, . . . , t
N−1
are ﬁxed a priori. But then if we let u(i) be the constant value of u() over [t
i
, t
i+1
),
we can reformulate (6.2) as (6.3):
Maximize x
1
(t
N
)(t
N
= t
f
)
subject to x(t
i+1
) = g(i, x(t
i
), u(i)), i = 0, 1, . . . , N −1
x(t
0
) = x(0) = (0, 0, M)
u(i) ≥ 0, x
3
(t
i
) ≥ 0, i = 0, 1, . . . , N .
(6.3)
In (6.3) g(i, x(t
1
), u(i)) is the state of the rocket at time t
i+1
when it is in state x(t
i
) at time t
i
and
u(t) ≡ u(i) for t
i
≤ t < t
i+1
.
As another example consider a simple inventory problem where time enters discretely in a natural
fashion. The Squeezme Toothpaste Company wants to plan its production and inventory schedule
for the coming month. It is assumed that the demand on the ith day, 0 ≤ i ≤ 30, is d
1
(i) for
6.2. MAIN RESULT 77
their orange brand and d
2
(i) for their green brand. To meet unexpected demand it is necessary that
the inventory stock of either brand should not fall below s > 0. If we let s(i) = (s
1
(i), s
2
(i))
denote the stock at the beginning of the ith day, and m(i) = (m
1
(i), m
2
(i))
denote the amounts
manufactured on the ith day, then clearly
s(i + 1) +s(i) +m(i) −d(i) ,
where d(i) = (d
1
(i), d
2
(i))
. Suppose that the initial stock is ˆ s, and the cost of storing inventory s
for one day is c(s) whereas the cost of manufacturing amount mis b(m). The the costminimization
decision problem can be formalized as (6.4):
Maximize
30
¸
i=0
(c(s(i)) +b(m(i)))
subject to s(i + 1) = s(i) +m(i) −d(i), 0 ≤ i ≤ 29
s(0) = ˆ s
s(i) ≥ (s, s)
, m(i) ≥ 0, 0 ≤ i ≤ 30 .
(6.4)
Before we formulate the general problem let us note that (6.3) and (6.4) are in the form of non
linear programming problems. The reason for treating these problems separately is because of their
practical importance, and because the conditions of optimality take on a special form.
6.2 Main Result
The general problem we consider is of the form (6.5).
Maximize
N−1
¸
i=0
f
0
(i, x(i), u(i))
subject to
dynamics : x(i + 1) −x(i) = f(i, x(i), u(i)), i = 0, . . . , N −1 ,
initial condition: q
0
(x(0) ≤ 0, g
0
(x(0)) = 0 ,
ﬁnal condition: q
N
(x(N)) ≤ 0, g
N
(x(N)) = 0 ,
statespace constraint: q
i
(x(i)) ≤ 0, i = 1, . . . , N −1 ,
control constraint: h
i
(u(i)) ≤ 0, i = 0, . . . , N −1 .
(6.5)
Here x(i) ∈ R
n
, u(i) ∈ R
p
, f
0
(i, , ) : R
n+p
→ R, f(i, , ) : R
n+p
→ R
n
, q
i
: R
n
→
R
m
i
, g
i
: R
n
→ R
i
, h
i
: R
p
→ R
s
i
are given differentiable functions. We follow the control
theory terminology, and refer to x(i) as the state of the system at time i, and u(i) as the control or
input at time i.
We use the formulation mentioned in the Remark following Theorem 3 of V.1.2, and construct
the Lagrangian function L by
L(x(0), . . . , x(N); u(0), . . . , u(N −1); p(1), . . . , p(N);
λ
0
, . . . , λ
N
; α
0
, α
N
; γ
0
, . . . , γ
N−1
)
78 CHAPTER 6. DISCRETETIME OPTIMAL CONTROL
=
N−1
¸
i=0
f
0
(i, x(i), u(i)) −
N−1
¸
i=0
(p(i + 1))
(x(i + 1) −x(i) −f(i, x(i), u(i)))+
N
¸
i=0
(λ
i
)
q
i
(x(i)) + (α
0
)
g
0
(x(0)) + (α
N
)
g
N
(x(N)) +
N−1
¸
i=0
(γ
i
)
h
i
(u(i))
¸
.
Suppose that CQ is satisﬁed for (6.5), and x
∗
(0), . . . , x
∗
(N); u
∗
(0), . . . , u
∗
(N −1), is an optimal
solution. Then by Theorem 2 of 5.1.2, there exist p
∗
(i) in R
n
for 1 ≤ i ≤ N, λ
i∗
≥ 0 in R
m
i
for
0 ≤ i ≤ N, α
i∗
in R
i
for i = 0, N, and γ
i∗
≥ 0 in R
s
i
for 0 ≤ i ≤ N −1, such that
(A) the derivative of L evaluated at these points vanishes,
and
(B) λ
i∗
q
i
(x
∗
(i)) = 0 for 0 ≤ i ≤ N, γ
i∗
h
i
(u
∗
(i)) = 0 for 0 ≤ i ≤ N −1 .
We explore condition (A) by taking various partial derivatives.
Differentiating L with respect to x(0) gives
f
0x
(0, x
∗
(0), u
∗
(0)) −¦−(p
∗
(1))
−(p
∗
(1))
[f
x
(0, x
∗
(0), u
∗
(0))]
+(λ
0∗
)
[q
0x
(x
∗
(0))] + (α
0∗
)
[g
0x
(x
∗
(0))]¦ = 0 ,
or
p
∗
(0) −p
∗
(1) = [f
x
(0, x
∗
(0), u
∗
(x))]
p
∗
(1)
+[f
0x
(0, x
∗
(0), u
∗
(0))]
−[q
0x
(x
∗
(0))]
λ
0∗
,
(6.6)
where we have deﬁned
p
∗
(0) = [g
0x
(x
∗
(x))]
α
0∗
. (6.7)
Differentiating L with respect to x(i), 1 ≤ i ≤ N −1, and rearranging terms gives
p
∗
(i) −p
∗
(i + 1) = [f
x
(i, x
∗
(i), u
∗
(i))]
p
∗
(i + 1)
+[f
0x
(i, x
∗
(i), u
∗
(i))]
−[q
ix
(x
∗
(i))]
λ
i∗
.
(6.8)
Differentiating L with respect to x(N) gives,
p
∗
(N) = −[g
Nx
(x
∗
(N))]
α
N∗
−[q
Nx
(x
∗
(N))]
λ
N∗
.
It is convenient to replace α
N∗
by −α
N∗
so that the equation above becomes (6.9)
p
∗
(N) = [g
Nx
(x
∗
(N))]
α
N∗
−[q
Nx
(x
∗
(N))]
λ
N∗
. (6.9)
Differentiating L with respect to u(i), 0 ≤ i ≤ N −1 gives
[f
0u
(i, x
∗
(i), u
∗
(i))]
+ [f
u
(i, x
∗
(i), u
∗
(i))]
p
∗
(i +l) −[h
iu
(u
∗
(i))]
γ
i∗
= 0 . (6.10)
We summarize our results in a convenient form in
Table 6.1
Remark 1: Considerable elegance and mnemonic simpliﬁcation is achieved if we deﬁne the
Hamiltonian function H by
6
.
2
.
M
A
I
N
R
E
S
U
L
T
7
9
Suppose x
∗
(0), . . . , x
∗
(N);
u
∗
(0), . . . , u
∗
(N −1) maximizes
N−1
¸
i=0
f
0
(i, x(i), u(i)) subject
to the constraints below
then there exist p
∗
(N); λ
0∗
, . . . , λN
∗
; α
0∗
, α
N∗;
γ
0∗
, . . . , γ
N−1
∗
, such that
dynamics: i = 0, . . . , N −1
x(i + 1) −x(i) = f(i, x(i), u(i))
initial condition:
q
0
(x
∗
(0)) ≤ 0, g
0
(x
∗
(0)) = 0
ﬁnal conditions:
q
N
(x
∗
(N)) ≤ 0, g
N
(x
∗
(N)) = 0
state space constraint:
i = 1, . . . , N −1
q
i
(x
∗
(i)) ≤ 0
control constraint:
i = 0, . . . , N −1
h
i
(u
∗
(i)) ≤ 0
adjoint equations: i = 0, . . . , N −1
p
∗
(i) −p
∗
(i + 1) = [f
x
(i, x
∗
(i), u
∗
(i)]
p
∗
(i + 1)
+[f
0x
(i, x
∗
(i), u
∗
(i)]
−[q
ix
(x
∗
(i)]
γ
i∗
transversality conditions:
p
∗
(0) = [g
0x
(x
∗
(0))]
α
0∗
p
∗
(N) = [g
Nx
(x
∗
(N))]
α
N∗
−[q
Nx
(x
∗
(N))]
λ
N∗
[f
0u
(i, x
∗
(i), u
∗
(i))]
+ [f
u
(i, x
∗
(i)u
∗
(i))]
.
p
∗
(i
1
) = [h
iu
(u
∗
(i))]
γ
i∗
λ
0∗
≥ 0,
(λ
0∗
)
q
0
(x
∗
(0)) = 0
λ
N∗
≥ 0,
(λ
N∗)
q
N
(x
∗
(N)) = 0
λ
i∗
≥ 0,
(λ
i∗
)
q
i
(x
∗
(i)) = 0
γ
i∗
≥ 0
(γ
i∗)
h
i
(u
∗
(i) = 0
T
a
b
l
e
6
.
1
:
80 CHAPTER 6. DISCRETETIME OPTIMAL CONTROL
H(i, x, u, p) = f
0
(i, x, u) +p
f(i, x, u) .
The dynamic equations then become
x
∗
(i + 1) −x
∗
(i) = [H
p
(i, x
∗
(i), u
∗
(i), p
∗
(i + 1))]
,
0 ≤ i ≤ N −1 .
(6.11)
and the adjoint equations (6.6) and (6.8) become
p
∗
(i) −p
∗
(i + 1) = [H
x
(i, x
∗
(i), u
∗
(i), u
∗
(i), p
∗
(i + 1))]
−[q
ix
(x
∗
(i))]
λ
i∗
,
0 ≤ i ≤ N −1 ,
whereas (6.10) becomes
[h
iu
(u
∗
(i))]
γ
i∗
= [H
u
(i, x
∗
(i), u
∗
(i), p
∗
(i + 1))]
, 0 ≤ i ≤ N −1 . (6.12)
Remark 2: If we linearize the dynamic equations about the optimal solution we obtain
δx(i + 1) −δx(i) = [f
x
(i, x
∗
(i), u
∗
(i))]δx(i) + [f
u
(i, x
∗
(i, x
∗
, (i), u
∗
(i))]δu(i) ,
whose homogeneous part is
z(i + 1) −z(i) = [f
x
(i, x
∗
(i), u
∗
(i))]z(i) ,
which has for it adjoint the system
r(i) −r(i + 1) = [f
x
(i, x
∗
(i), u
∗
(i))]
r(i + 1) . (6.13)
Since the homogeneous part of the linear difference equations (6.6), (6.8) is (6.13), we call (6.6),
(6.8) the adjoint equations, and the p
∗
(i) are called adjoint variables.
Remark 3: If the f
0
(i, , ) are concave and the remaining function in (6.5) are linear, then CQ is
satisﬁed, and the necessary conditions of Table 6.1 are also sufﬁcient. Furthermore, in this case we
see from (6.13) that u
∗
(i) is an optimal solution of
Maximize H(i, x
∗
(i), u, p
∗
(i + 1)),
subject to h
i
(u) ≤ 0 .
For this reason the result is sometimes called the maximum principle.
Remark 4: The conditions (6.7), (6.9) are called transversality conditions for the following reason.
Suppose q
0
≡ 0, q
N
≡ 0, so that the initial and ﬁnal conditions read g
0
(x(0)) = 0, g
N
(x(N)) = 0,
which describe surfaces in R
n
. Conditions (6.7), (6.9) become respectively p
∗
(0) = [g
0x
(x
∗
(0))]
α
0∗
, p
∗
(N) =
[g
Nx
(x(N))]
α
N∗
which means that p
∗
(0) and p
∗
(N) are respectively orthogonal or transversal to
the initial and ﬁnal surfaces. Furthermore, we note that in this case the initial and ﬁnal conditions
specify (
0
+
n
) conditions whereas the transversality conditions specify (n−
0
)+(n−
n
) condi
tions. Thus, we have a total of 2n boundary conditions for the 2ndimensional system of difference
equations (6.5), (6.12); but note that these 2n boundary conditions are mixed, i.e., some of them
refer to the initial time 0 and the rest refer to the ﬁnal time.
6.2. MAIN RESULT 81
Exercise 1: For the regulator problem,
Maximize
1
2
N−1
¸
i=0
x(i)
Qx(i) +
1
2
N−1
¸
i=0
u(i)
Pu(i)
subject to x(i + 1) −x(i) = Ax(i) +Bu(i), 0 ≤ i ≤ N −1
x(0) = ˆ x(0),
u(i) ∈ R
p
, 0 ≤ i ≤ N −1 ,
where x(i) ∈ R
n
, A and B are constant matrices, ˆ x(0) is ﬁxed, Q = Q
is positive semideﬁnite,
and P = P
is positive deﬁnite, show that the optimal solution is unique and can be obtained by
solving a 2ndimensional linear difference equation with mixed boundary conditions.
Exercise 2: Show that the minimal fuel problem,
Minimize
N−1
¸
i=0
¸
P
¸
j=1
[(u(i))
j
[
¸
,
subject to x(i + 1) −x(i) = Ax(i) +Bu(i), 0 ≤ i ≤ N −1
x(0) = ˆ x(0), x(N) = ˆ x(N) ,
u(i) ∈ R
p
, [u(i))
j
[ ≤ 1, 1 ≤ j ≤ p, 0 ≤ i ≤ N −1
can be transformed into a linear programming problem. Here ˆ x(0), ˆ x(N) are ﬁxed, A and B are as
in Exercise 1.
82 CHAPTER 6. DISCRETETIME OPTIMAL CONTROL
Chapter 7
SEQUENTIAL DECISION PROBLEMS:
CONTINUOUSTIME OPTIMAL
CONTROL OF LINEAR SYSTEMS
We will investigate decision problems similar to those studied in the last chapter with one (math
ematically) crucial difference. A choice of control has to be made at each instant of time t where
t varies continuously over a ﬁnite interval. The evolution in time of the state of the systems to be
controlled is governed by a differential equation of the form:
˙ x(t) = f(t, x(t), u(t)) ,
where x(t) ∈ R
n
and u(t) ∈ R
p
are respectively the state and control of the system at time t.
To understand the main ideas and techniques of analysis it will prove proﬁtable to study the linear
case ﬁrst. The general nonlinear case is deferred to the next chapter. In Section 1 we present the
general linear problem and study the case where the initial and ﬁnal conditions are particularly
simple. In Section 2 we study more general boundary conditions.
7.1 The Linear Optimal Control Problem
We consider a dynamical system governed by the linear differential equation (7.1):
˙ x(t) = A(t)x(t) +B(t)u(t), t ≥ t
0
. (7.1)
Here A() and B() are n n and n pmatrix valued functions of time; we assume that they are
piecewise continuous functions. The control u() is constrained to take values in a ﬁxed set Ω ⊂ R
p
,
and to be piecewise continuous.
Deﬁnition: A piecewise continuous function u : [t
0
, ∞) → Ω will be called an admissible control.
 denotes the set of all admissible controls.
Let c ∈ R
n
, x
0
∈ R
n
be ﬁxed and let t
f
≥ t
0
be a ﬁxed time. We are concerned with the
83
84 CHAPTER 7. CONTINUOUSTIME LINEAR OPTIMAL CONTROL
decision problem (7.2).
Maximize c
x(t
f
),
subject to
dynamics: ˙ x(t) = A(t)x(t) +B(t)u(t) , t
0
≤ t ≤ t
f
,
initial condition: x(t
0
) = x
0
,
ﬁnal condition: x(t
f
) ∈ R
n
,
control constraint: u() ∈  .
(7.2)
Deﬁnition: (i) For any piecewise continuous function u() : [t
0
, t
f
] → R
p
, for any z ∈ R
n
, and
any t
0
≤ t
1
≤ t
2
≤ t
f
let
φ(t
2
, t
1
, z, u)
denote the state of (7.1) at time t
2
, if a time t
1
it is in state z, and the control u() is applied.
(ii) Let
K(t
2
, t
1
, z) = ¦φ(t
2
, t
1
, z, u)[u ∈ ¦ .
Thus, K(t
2
, t
1
, z) is the set of states reachable at time t
2
starting at time t
1
in state z and using
admissible controls. We call K the reachable set.
Deﬁnition: Let Φ(t, τ), t
0
≤ τ ≤ t ≤ t
f
, be the transitionmatrix function of the homogeneous
part of (7.1), i.e., Φ satisﬁes the differential equation
∂Φ
∂t
(t, τ) = A(t)Φ(t, τ) ,
and the boundary condition
Φ(t, t) ≡ I
n
.
The next result is wellknown. (See Desoer [1970].)
Lemma 1: φ(t
2
, t
1
, z, u) = Φ(t
2
, t
1
)z +
t
2
t
1
Φ(t
2
, τ)B(τ)u(τ)dτ.
Exercise 1: (i) Assuming that Ω is convex, show that  is a convex set. (ii) Assuming that  is
convex show that K(t
2
, t
1
, z) is a convex set. (It is a deep result that K(t
2
, t
1
, z) is convex even if
Ω is not convex (see Neustadt [1963]), provided we include in  any measurable function
u : [t
0
, ∞) → Ω.)
Deﬁnition: Let K ⊂ R
n
, and let x
∗
∈ K. We say that c is the outward normal to a hyperplane
supporting K at x
∗
if c = 0, and
c
x
∗
≥ c
x for all x ∈ K .
The next result gives a geometric characterization of the optimal solutions of (2).
Lemma 2: Suppose c = 0. Let u
∗
() ∈  and let x
∗
(t) = φ(t, t
0
, x
0
, u
∗
). Then u
∗
is an optimal
solution of (2) iff
(i) x
∗
(t
f
) is on the boundary of K = K(t
f
, t
0
, x
0
), and
(ii) c is the outward normal to a hyperplane supporting K at x
∗
. (See Figure 7.1.)
Proof: Clearly (i) is implied by (ii) because if x
∗
(t
f
) is in the interior of K there is δ > 0 such
that (x
∗
(t
f
) +δc) ∈ K; but then
7.1. THE LINEAR OPTIMAL CONTROL PROBLEM 85
x
3
c
x
2
x
1
c
x
∗
(t
f
)
π
∗
= ¦x[c
x = c
x
∗
(t
f
)¦
K
Figure 7.1: c is the outward normal to π
∗
supporting K at x
∗
(t
f
)
.
c
(x
∗
(t
f
) +δc) = c
x
∗
(t
f
) +δ[c[
2
> c
x
∗
(t
f
) .
Finally, from the deﬁnition of K it follows immediately that u
∗
is optimal iff c
x
∗
(t
f
) ≥ c
x for all
x ∈ K . ♦
The result above characterizes the optimal control u
∗
in terms of the ﬁnal state x
∗
(t
f
). The beauty
and utility of the theory lies in the following result which translates this characterization directly in
terms of u
∗
.
Theorem 1: Let u
∗
() ∈  and let x
∗
(t) = φ(t, t
0
, x
0
, u
∗
), t
0
≤ t ≤ t
f
. Let p
∗
(t) be the solution
of (7.3) and (7.4):
adjoint equation: ˙ p
∗
(t) = −A
(t)p
∗
(t) , t
0
≤ t ≤ t
f
. (7.3)
ﬁnal condition: p
∗
(t
f
) = c . (7.4)
Then u
∗
() is optimal iff
(p
∗
(t))
B(t)u
∗
(t) = sup¦(p
∗
(t))
B(t)v[v ∈ Ω¦ , (7.5)
for all t ∈ [t
0
, t
f
], except possibly for a ﬁnite set.
Proof: u
∗
() is optimal iff for every u() ∈ 
(p
∗
(t
f
))
[Φ(t
f
, t
0
)x
0
+
t
f
t
0
Φ(t
f
, τ)B(τ)u
∗
(τ)dτ]
≥ (p
∗
(t
f
))
[Φ(t
f
, t
0
)x
0
+
t
f
t
0
Φ(t
f
, τ)B(τ)u(τ)dτ] ,
which is equivalent to (7.6).
t
f
t
0
(p
∗
(t
f
))
Φ(t
f
, τ)B(τ)u
∗
(τ)dτ
≥
t
f
t
0
(p
∗
(t
f
))
Φ(t
f
, τ)B(τ)u(τ)dτ
(7.6)
86 CHAPTER 7. CONTINUOUSTIME LINEAR OPTIMAL CONTROL
Now by properties of the adjoint equation we know that p
∗
(t))
= (p
∗
(t
f
))
Φ(t
f
, t) so that (7.6) is
equivalent to (7.7),
t
f
t
0
(p
∗
(τ))
B(τ)u
∗
(τ)dτ ≥
t
f
t
0
(p
∗
(τ))
B(τ)u(τ)dτ, (7.7)
and the sufﬁciency of (7.5) is immediate.
To prove the necessity let D be the ﬁnite set of points where the function B() or u
∗
() is discon
tinuous. We shall show that if u
∗
() is optimal then (7.5) is satisﬁed for t ∈ D. Indeed if this is not
the case, then there exists t
∗
∈ [t
0
, t
f
], t
∗
∈ D, and v ∈ Ω such that
(p
∗
(t
∗
))
B(t
∗
)u
∗
(t
∗
) < (p
∗
(t
∗
))
B(t
∗
)v ,
and since t
∗
is a point of continuity of B() and u
∗
(), it follows that there exists δ > 0 such that
(p
∗
(t))
B(t)u
∗
(t) < (p
∗
(t))
B(t)v, for [t −t
∗
[ < δ . (7.8)
Deﬁne ˜ u() ∈  by
˜ u(t) =
v [t −t
∗
[ < δ, t ∈ [t
0
, t
f
]
u
∗
(t) otherwise .
Then (7.8) implies that
t
f
t
0
(p
∗
(t))
B(t)˜ u(t)dt >
t
f
t
0
(p
∗
(t))
B(t)u
∗
(t)dt .
But then from (7.7) we see that u
∗
() cannot be optimal, giving a contradiction. ♦
Corollary 1: For t
0
≤ t
1
≤ t
2
≤ t
f
,
(p
∗
(t
2
))x
∗
(t
2
) ≥ (p
∗
(t
2
))
x for all x ∈ K(t
2
, t
1
, x
∗
(t
1
)). (7.9)
Exercise 2: Prove Corollary 1.
Remark 1: The geometric meaning of (7.9) is the following. Taking t
1
= t
0
in (7.9), we see that if
u
∗
() is optimal, i.e., if c = p
∗
(t
f
) is the outward normal to a hyperplane supporting K(t
f
, t
0
, x
0
)
at x
∗
(t
f
), then x
∗
(t) is on the boundary of K(t, t
0
, x
0
) and p
∗
(t) is the normal to a hyperplane
supporting K(t, t
0
, x
0
) at x
∗
(t). This normal is obtained by transporting backwards in time, via
the adjoint differential equation, the outward normal p
∗
(t
f
) at time t
f
. The situation is illustrated
in Figure 7.2.
Remark 2: If we deﬁne the Hamiltonian function H by
H(t, x, u, p) = p
(A(t)x +B(t)u) ,
and we deﬁne M by
M(t, x, p) = sup¦H(t, x, u, p)[u ∈ Ω¦,
then (7.5) can be rewritten as
H(t, x
∗
(t), u
∗
(t), p
∗
(t)) = M(t, x
∗
(t), p
∗
(t)) . (7.10)
This condition is known as the maximum principle.
7.2. MORE GENERAL BOUNDARY CONDITIONS 87
Exercise 3: (i) Show that m(t) = M(t, x
∗
(t), p
∗
(t)) is a Lipschitz function of t. (ii) If A(t), B(t)
are constant, show that m(t) is constant. (Hint: Show that (dm/dt) ≡ 0.)
The next two exercises show how we can obtain important qualitative properties of an optimal
control.
Exercise 4: Suppose that Ω is bounded and closed. Show that there exists an optimal control u
∗
()
such that u
∗
(t) belongs to the boundary of Ω for all t.
Exercise 5: Suppose Ω = [α, β], so that B(t) is an n 1 matrix. Suppose that A(t) ≡ A and
B(t) ≡ B are constant matrices and A has n real eigenvalues. Show that there is an optimal
control u
∗
() and t
0
≤ t
1
≤ t
2
≤ . . . ≤ t
n
≤ t
f
such that u
∗
(t) ≡ α or β on [t
i
, t
i+1
), 0 ≤ i ≤ n.
(Hint: ﬁrst show that (p
∗
(t))
B = γ
1
exp(δ
1
t) +. . . +γ
n
exp(δ
n
(t)) for some γ
i
, δ
i
in R.)
Exercise 6: Assume that K(t
f
, t
0
, x
0
) is convex (see remark in Exercise 1 above). Let
f
0
: R
n
→ R be a differentiable function and suppose that the objective function in (7.2) is
f
0
(x(t
f
)) instead of c
x(t
f
). Suppose u
∗
() is an optimal control. Show that u
∗
() satisﬁes the
maximum principle (7.10) where p
∗
() is the solution of the adjoint equation (7.3) with the ﬁnal
condition
p
∗
(t
f
) = f
0
(x
∗
(t
f
)) .
Also show that this condition is sufﬁcient for optimality if f
0
is concave. (Hint: Use Lemma 1 of
5.1.1 to show that if u
∗
() is optimal, then f
0x
(x
∗
(t
f
)(x
∗
(t
f
) −x) ≤ for all x ∈ K(t
f
, t
0
, x
0
).)
7.2 More General Boundary Conditions
We consider the following generalization of (7.2). The notion of the previous section is retained.
Maximize c
x(t
f
)
subject to
dynamics: ˙ x(t) = A(t)x(t) +B(t)u(t), t
0
≤ t ≤ t
f
,
initial condition: G
0
x(t
0
) = b
0
,
ﬁnal condition: G
f
x(t
f
) = b
f
,
control constraint: u() ∈ , i.e., ¯ : [.
, .
{
] → ⊗ and
u()piecewise continuous.
(7.11)
In (7.11) G
0
and G
f
are ﬁxed matrices of dimensions
0
xn and
f
n respectively, while b
0
∈
R
0
, b
f
∈ R
f
are ﬁxed vectors.
We will analyze the problem in the same way as before. That is, we ﬁrst characterize optimality
in terms of the state at the ﬁnal time, and then translate these conditions in terms of the control. For
convenience let
T
0
= ¦z ∈ R
n
[G
0
z = b
0
¦ ,
T
f
= ¦z ∈ R
n
[G
f
z = b
f
¦ .
Deﬁnition: Let p ∈ R
n
. Let z
∗
∈ T
0
. We say that p is orthogonal to T
0
at z
∗
and we write
p ⊥ T
0
(z
∗
) if
88 CHAPTER 7. CONTINUOUSTIME LINEAR OPTIMAL CONTROL
R
n
x
0
=
x
∗
(
t
0
)
=
K
(
t
0
,
t
0
,
x
0
) R
n
p
∗
(
t
1
)
x
∗
(
t
1
)
K
(
t
1
,
t
0
,
x
0
)
R
n
p
∗
(
t
2
)
x
∗
(
t
2
)
K
(
t
2
,
t
0
,
x
0
)
R
n
p
∗
(
t
f
)
=
c
x
∗
(
t
f
)
K
(
t
f
,
t
0
,
x
0
)
t
t
f
t
2
t
1
t
0
Figure 7.2: Illustration of (7.9) for t
1
= t
0
.
7.2. MORE GENERAL BOUNDARY CONDITIONS 89
p
(z −z
∗
) = 0 for all z ∈ T
0
.
Similarly if z
∗
∈ T
f
, p ⊥ T
f
(z
∗
) if
p
(z −z
∗
) = 0 for all z ∈ T
f
.
Deﬁnition: Let X(t
f
) = ¦Φ(t
f
, t
0
)z +w[z ∈ T
0
, w ∈ K(t
f
, t
0
, 0)¦.
Exercise 1: X(t
f
) = ¦Φ(t
f
, t
0
, z, u)[z ∈ T
0
, u() ∈ ¦.
Lemma 1: Let x
∗
(t
0
) ∈ T
0
and u
∗
() ∈ . Let x
∗
(t) = φ(t, t
0
, x
∗
(t
0
), u
∗
), and suppose that
x
∗
(t
f
) ∈ T
f
.
(i) Suppose the Ω is convex. If u
∗
() is optimal, there exist ˆ p
0
∈ R, ˆ p
0
≥ 0 and ˆ p ∈ R
n
, not both
zero, such that
(ˆ p
0
c + ˆ p)
x
∗
(t
f
) ≥ (ˆ p
0
c + ˆ p)
x for all x ∈ X(t
f
) , (7.12)
ˆ p ⊥ T
f
(x
∗
(t
f
)) , (7.13)
[Φ(t
f
, t
0
)]
(ˆ p
0
c + ˆ p) ⊥ T
0
(x
∗
(t
0
)) . (7.14)
(ii) Conversely if there exist ˆ p
0
> 0, and ˆ p such that (7.12) and (7.13) are satisﬁed, then u
∗
() is
optimal and (7.14) is also satisﬁed.
Proof: Clearly u
∗
() is optimal iff
c
x
∗
(t
f
) ≥ c
x for all x ∈ X(t
f
) ∩ T
f
. (7.15)
(i) Suppose that u
∗
() is optimal. In R
1+m
deﬁne sets S
1
, S
2
by
S
1
= ¦(r, x)[r > c
x
∗
(t
f
), x ∈ T
f
¦ , (7.16)
S
2
= ¦(r, x)[r = c
x , x ∈ X(t
f
)¦ . (7.17)
First of all S
1
∩S
2
= φ because otherwise there exists x ∈ X(t
f
) ∩T
f
such that c
x > c
x
∗
(t
f
)
contradicting optimality of u
∗
() by (7.15).
Secondly, S
1
is convex since T
f
is convex. Since Ω is convex by hypothesis it follows by Exercise
1 of Section 1 that S
2
is convex.
But then by the separation theorem for convex sets (see 5.5) there exists ˆ p
0
∈ R, ˆ p ∈ R
n
, not
both zero, such that
ˆ p
0
r
1
+ ˆ p
x
1
≥ ˆ p
0
r
2
+ ˆ p
x
2
for all (r
i
, x
i
) ∈ S
i
, i = 1, 2. (7.18)
In particular (7.18) implies that
ˆ p
0
r + ˆ p
x
∗
(t
f
) ≥ ˆ p
0
c
x + ˆ p
x for all x ∈ X(t
f
), r > c
x
∗
(t
f
). (7.19)
Letting r → ∞ we conclude that (7.19) can hold only if ˆ p
0
≥ 0. On the other hand letting r →
c
x
∗
(t
f
) we see that (7.19) can hold only if
ˆ p
0
c
x
∗
(t
f
) + ˆ p
x
∗
(t
f
) ≥ ˆ p
0
c
x + ˆ p
x for all x ∈ X(t
f
) , (7.20)
which is the same as (7.12). Also from (7.18) we get
90 CHAPTER 7. CONTINUOUSTIME LINEAR OPTIMAL CONTROL
ˆ p
0
r + ˆ p/x ≥ ˆ p
0
c
x
∗
(t
f
) + ˆ p
x
∗
(t
f
) for all r > c
x
∗
(t
f
), x ∈ T
f
,
which can hold only if
ˆ p
1
c
x
∗
(t
f
) + ˆ p
x ≥ ˆ p
0
c
x
∗
(t
f
) + ˆ p
x
∗
(t
f
) for all x ∈ T
f
,
or
ˆ p
(x −x
∗
(t
f
)) ≥ 0 for all x ∈ T
f
(7.21)
But ¦x −x
∗
(t
f
)[x ∈ T
f
¦ = ¦z[G
f
z = 0¦ is a subspace of R
n
, so that (7.21) can hold only if
ˆ p
(x −x
∗
(t
f
)) = 0 for all x ∈ T
f
,
which is the same as (7.13). Finally (7.12) always implies (7.14), because by the deﬁnition of X(t
f
)
and Exercise 1, ¦Φ(t
f
, t
0
)(z − x
∗
(t
0
)) + x
∗
(t
f
)¦ ∈ X(t
f
) for all z ∈ T
0
, so that from (7.12) we
get
0 ≥ (ˆ p
0
c + ˆ p)
Φ(t
f
, t
0
)(z −x
∗
(t
0
)) for all z ∈ T
0
,
which can hold only if (7.14) holds.
(ii) Now suppose that ˆ p
0
> 0 and ˆ p are such that (7.12), (7.13) are satisﬁed. Let ˜ x ∈ X(t
f
) ∩ T
f
.
Then from (7.13) we conclude that
ˆ p
x
∗
(t
f
) = ˆ p
˜ x ,
so that from (7.12) we get
ˆ p
0
c
x
∗
(t
f
) ≥ ˆ p
0
c
˜ x ;
but then by (7.15) u
∗
() is optimal. ♦
Remark 1: If it is possible to choose ˆ p
0
> 0 then ˆ p
0
= 1, ˆ p = (ˆ p/ˆ p
0
) will also satisfy (7.12),
(7.13), and (7.14). In particular, in part (ii) of the Lemma we may assume ˆ p
0
= 1.
Remark 2: it would be natural to conjecture that in part (i) ˆ p
0
may be chosen > 0. But in Figure
7.3 below, we illustrate a 2dimensional situation where T
0
= ¦x
0
¦, T
f
is the vertical line, and
T
f
∩ X(t
f
) consists of just one vector. It follows that the control u
∗
() ∈  for which
x
∗
(t
f
) = φ(t
f
, t
0
, x
0
, u
∗
) ∈ T
f
is optimal for any c. Clearly then for some c (in particular for the
c in Figure 7.3) we are forced to set ˆ p
0
= 0. In higher dimensions the reasons may be more
complicated, but basically if T
f
is “tangent” to X(t
f
) we may be forced to set ˆ p
0
= 0 (see
Exercise 2 below). Finally, we note that part (i) is not too useful if ˆ p
0
= 0, since then (7.12), (7.13),
and (7.14) hold for any vector c whatsoever. Intuitively ˆ p
0
= 0 means that it is so difﬁcult to satisfy
the initial and ﬁnal boundary conditions in (7.11) that optimization becomes a secondary matter.
Remark 3: In (i) the convexity of Ω is only used to guarantee that K(t
f
, t
0
, 0) is convex. But it is
known that K(t
f
, t
0
, 0) is convex even if Ω is not (see Neustadt [1963]).
Exercise 2: Suppose there exists z in the interior of X(t
f
) such that z ∈ T
f
. Then in part (i) we
must have ˆ p
0
> 0.
We now translate the conditions obtained in Lemma 1 in terms of the control u
∗
.
7.2. MORE GENERAL BOUNDARY CONDITIONS 91
Theorem 1: Let x
∗
(t
0
) ∈ T
0
and u
∗
() ∈ . Let x
∗
(t) = φ(t, t
0
, x
∗
(t
0
), u
∗
) and suppose that
x
∗
(t
f
) ∈ T
f
.
(i) Suppose that Ω is convex. If u
∗
() is optimal for (7.11), then there exist a number p
∗
0
≥ 0, and a
function p
∗
: [t
0
, t
f
] → R
n
, not both identically zero, satisfying
adjoint equation: ˙ p
∗
(t) = −A
(t)p
∗
(t) , t
0
≤ t ≤ t
f
(7.22)
initial condition: p
∗
(t
0
)⊥T
0
(x
∗
(t
0
)) (7.23)
ﬁnal condition: (p
∗
(t
f
) −p
∗
0
c)⊥T
f
(x
∗
(t
f
)) , (7.24)
and the maximum principle
H(t, x
∗
(t), u
∗
(t), p
∗
(t)) = M(t, x
∗
(t), p
∗
(t)) , (7.25)
holds for all t ∈ [t
0
, t
f
] except possibly for a ﬁnite set.
(ii) Conversely suppose there exist p
∗
0
> 0 and p
∗
() satisfying (7.22), (7.23), (7.24), and (7.25).
Then u
∗
() is optimal.
[Here
H(t, x, u, p) = p
(A(t)x +B(t)u), M(t, x, p) = sup¦H(t, x, v, p)[v ∈ Ω¦.]
Proof: A repetition of a part of the argument in the proof of Theorem 1 of Section 1 show that if p
∗
satisﬁes (7.22), then (7.25) is equivalent to (7.26):
(p
∗
(t
f
))
x
∗
(t
f
) ≥ (p
∗
(t
f
))
x for all x ∈ K(t
f
, t
0
, x
∗
(t
0
)) . (7.26)
(i) Suppose u
∗
() is optimal and Ω is convex. Then by Lemma 1 there exist ˆ p ≥ 0, ˆ p ∈ R
n
, not
both zero, such that (7.12), (7.13) and (7.14) are satisﬁed. Let p
∗
0
= ˆ p
0
and let p
∗
() be the solution
of (7.22) with the ﬁnal condition
p
∗
(t
f
) = p
∗
0
c + ˆ p = ˆ p
0
c + ˆ p .
Then (7.14) and (7.13) are respectively equivalent to (7.23) and (7.24), whereas since K(t
f
, t
0
, x
∗
(t
0
)) ⊂
X(t
f
), (7.26) is implied by (7.12).
(ii) Suppose p
∗
0
> 0 and (7.22), (7.23), (7.24), and (7.26) are satisﬁed. Let ˆ p
0
= p
∗
0
and ˆ p =
p
∗
(t
f
) −p
∗
0
c, so that (7.24) becomes equivalent to (7.13). Next if x ∈ X(t
f
) we have
(ˆ p
0
c + ˆ p)
x = (p
∗
(t
f
))
x
= (p
∗
(t
f
))
(Φ(t
f
, t
0
)z +w) ,
92 CHAPTER 7. CONTINUOUSTIME LINEAR OPTIMAL CONTROL
.
x
0
=
T
0
T
f
x
∗
(
t
f
)
=
X
(
t
f
)
¸
T
f
K
(
t
f
,
t
0
,
x
0
)
=
X
(
t
f
)
c
t
Figure 7.3: Situation where ˆ p
0
= 0
7.2. MORE GENERAL BOUNDARY CONDITIONS 93
for some z ∈ T
0
and some w ∈ K(t
f
, t
0
, 0). Hence
(ˆ p
0
c + ˆ p)
x = (p
∗
(
f
))
Φ(t
f
, t
0
)(z −x
∗
(t
0
))
+(p
∗
(t
f
))
(w +φ(t
f
, t
0
)x
∗
(t
0
))
= (p
∗
(t
0
))
(z −x
∗
(t
0
))
+(p
∗
(t
f
))
(w + Φ(t
f
, t
0
)x
∗
(t
0
)) .
But by (7.23) the ﬁrst term on the right vanishes, and since (w+φ(t
f
, t
0
)x
∗
(t
0
)) ∈ K(t
f
, t
0
, x
∗
(t
0
)),
it follows from (7.26) that the second term is bounded by (p
∗
(t
f
))
x
∗
(t
f
). Thus
(ˆ p
0
c + ˆ p)
x
∗
(t
f
) ≥ (ˆ p
0
c + ˆ p)
x for all x ∈ X(t
f
) ,
and so u
∗
() is optimal by Lemma 1. ♦
Exercise 3: Suppose that the control constraint set is Ω(t) which varies continuously with t, and
we require that u(t) ∈ Ω(t) for all t. Show that Theorem 1 also holds for this case where, in (7.25),
M(t, x, p) =sup¦H(t, x, v, p)[v ∈ Ω(t)¦.
Exercise 4: How would you use Exercise 3 to solve Example 3 of Chapter 1?
94 CHAPTER 7. CONTINUOUSTIME LINEAR OPTIMAL CONTROL
Chapter 8
SEQUENTIAL DECISION PROBLEMS:
CONTINUOUSTIME OPTIMAL
CONTROL OF NONLINEAR SYSTEMS
We now present a sweeping generalization of the problem studied in the last chapter. Unfortunately
we are forced to omit the proofs of the results since they require a level of mathematical sophis
tication beyond the scope of these Notes. However, it is possible to convey the main ideas of the
proofs at an intuitive level and we shall do so. (For complete proofs see (Lee and Markus [1967]
or Pontryagin, et al., [1962].) The principal result, which is a direct generalization of Theorem 1 of
7.2 is presented in Section 1. An alternative form of the objective function is discussed in Section
2. Section 3 deals with the minimumtime problem and Section 4 considers the important special
case of linear systems with quadratic cost. Finally, in Section 5 we discuss the socalled singular
case and also analyze Example 4 of Chapter 1.
8.1 Main Results
8.1.1 Preliminary results based on differential equation theory.
We are interested in the optimal control of a system whose dynamics are governed by the nonlinear
differential equation
˙ x(t) = f(t, x, (t), u(t)) , t
0
≤ t ≤ t
f
, (8.1)
where x(t) ∈ R
n
is the state and u(t) ∈ R
p
is the control. Suppose u
∗
() is an optimal control
and x
∗
() is the corresponding trajectory. In the case of linear systems we obtained the necessary
conditions for optimality by comparing x
∗
() with trajectories x() corresponding to other admis
sible controls u(). This comparison was possible because we had an explicitly characterization of
x() in terms of u(). Unfortunately when f is nonlinear such a characterization is not available.
Instead we shall settle for a comparison between the trajectory x
∗
() and trajectories x() obtained
by perturbing the control u
∗
() and the initial condition x
∗
(t
0
). We can then estimate the difference
between x() and x
∗
() by the solution to a linear differential equation as shown in Lemma 1 below.
But ﬁrst we need to impose some regularity conditions on the differential equation (8.1). We assume
throughout that the function f : [t
0
, t
f
] R
n
R
p
→ R
n
satisﬁes the following conditions:
95
96 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
1. for each ﬁxed t ∈ [t
0
, t
f
], f(t, , ) : R
n
xR
p
→ R
n
is continuously differentiable in the
remaining variables (x, u),
2. except for a ﬁnite subset D ⊂ [t
0
, t
f
], the functions f, f
x
, f
u
are continuous on [t
0
, t
f
] R
n
R
p
, and
3. for every ﬁnite α, there exist ﬁnite number β and γ such that
[f(t, x, u)[ ≤ β +γ[x[ for all t ∈ [t
0
, t
f
], x ∈ R
n
, u ∈ R
p
with [u[ ≤ α .
The following result is proved in every standard treatise on differential equations.
Theorem 1: For every z ∈ R
n
, for every t
1
∈ [t
0
, t
f
], and every piecewise continuous function
u() : [t
0
, t
f
] → R
p
, there exists a unique solution
x(t) = φ(t, t
1
, z, u()) , t
1
≤ t ≤ t
f
,
of the differential equation
˙ x(t) = f(t, x(t), u(t)) , t
1
≤ t ≤ t
f
,
satisfying the initial condition
x(t
1
) = z .
Furthermore, for ﬁxed t
1
≤ t
2
in [t
0
, t
f
] and ﬁxed u(), the function φ(t
2
, t
1
, , u()) : R
n
→ R
n
is
differentiable. Moreover, the n n matrixvalued function Φ deﬁned by
Φ(t
2
, t
1
, z, u()) =
∂φ
∂z
(t
2
, t
1
, z, u())
is the solution of the linear homogeneous differential equation
∂Φ
∂t
(t, t
1
, z, u, ()) = [
∂f
∂x
(t, x, (t), u(t))]Φ(t, t
1
, z, u()), t
1
≤ t ≤ t
f
,
and the initial condition
Φ(t
1
, t
1
, z, u()) = I
n
.
Now let Ω ⊂ R
p
be a ﬁxed set and let  be set of all piecewise continuous functions u() :
[t
0
, t
f
] → Ω. Let u
∗
() ∈  be ﬁxed and let D
∗
be the set of discontinuity points of u
∗
(). Let
x
∗
0
∈ R
n
be a ﬁxed initial condition.
Deﬁnition: π = (t
1
, . . . , t
m
;
1
, . . . ,
m
; u
1
, . . . , u
m
) is said to be a perturbation data for u
∗
() if
1. m is a nonnegative integer,
2. t
0
< t
1
< t
2
< . . . t
m
< t
f
, and t
i
∈ D
∗
¸
D, i = 1, . . . , m (recall that D is the set of
discontinuity points of f),
3.
i
≥ 0, i = 1, . . . , m, and
4. u
i
∈ Ω, i = 1, . . . , m.
8.1. MAIN RESULTS 97
Let ε(π) > 0 be such that for 0 ≤ ε ≤ ε(π) we have [t
i
− ε
i
, t
i
] ⊂ [t
0
, t
f
] for all i, and
[t
i
−ε
i
, t
i
]
¸
[t
j
−ε
j
, t
j
] = φ for i = j. Then for 0 ≤ ε ≤ ε(π),the perturbed control u
(π,ε)
() ∈ 
corresponding to π is deﬁned by
u
(π,ε)
(t) =
u
i
for all t ∈ [t
i
−ε
i
, t
i
] , i = 1, . . . , m
u
∗
(t) otherwise .
Deﬁnition: Any vector ξ ∈ R
n
is said to be a perturbation for x
∗
0
, and a function x
(ξ,ε)
deﬁned for
ε > 0 is said to be a perturbed initial condition if
lim
ε→0
x
(ξ,ε)
= x
∗
0
,
and
lim
ε→0
1
ε
(x
(ξ,ε)
−x
∗
0
) = ξ .
Now let x
∗
(t) = φ(t, t
0
, x
∗
0
, u
∗
()) and let x
ε
(t) = φ(t, t
0
, x
(ξ,ε)
, u
(π,ε)
()). Let Φ(t
2
, t
1
) =
Φ(t
2
, t
1
, x
∗
(t
1
), u
∗
()). The following lemma gives an estimate of x
∗
(t) − x
ε
(t). The proof of the
lemma is a straightforward exercise in estimating differences of solutions to differential equations,
and it is omitted (see for example (Lee and Markus [1967])).
Lemma 1: lim
ε→0
[x
ε
(t) −x
∗
(t) −εh
(π,ε)
(t)[ = 0 for t ∈ [t
0
, t
1
], where h
(π,ε)
() is given by
h
(π,ε)
(t) = Φ(t, t
0
)ξ , t ∈ [t
0
, t
1
)
= Φ(t, t
0
)ξ + Φ(t, t
1
)[f(t
1
, x
∗
(t
1
), u
1
) −f(t
1
, x
∗
(t
1
), u
∗
(t
1
))]
1
, t ∈ [t
1
, t
2
)
= Φ(t, t
0
)ξ +
i
¸
j=1
Φ(t, t
j
)[f(t
j
, x
∗
(t
j
), u
j
) −f(t
j
, x
∗
(t
j
), u
∗
(t
j
))]
j
, t ∈ [t
i
, t
i+1
)
= Φ(t, t
0
)ξ +
m
¸
j=1
Φ(t, t
m
)[f(t
j
, x
∗
(t
j
), u
j
) −f(t
j
, x
∗
(t
j
), u
∗
t
j
))]
j
, t ∈ [t
m
, t
f
] .
(See Figure 8.1.)
We call h
(π,ξ)
() the linearized (trajectory) perturbation corresponding to (π, ξ).
Deﬁnition: For z ∈ R
n
, t ∈ [t
0
, t
f
] let
K(t, t
0
, z) = ¦φ(t, t
0
, z, u())[u() ∈ ¦
be the set of states reachable at time t, starting at time t
0
in state z, and using controls u() ∈ .
Deﬁnition: For each t ∈ [t
0
, t
f
], let
Q(t) = ¦h
(π,0)
(t)[πis a perturbation data for u
∗
(), and
h
(π,0)
()is the linearized perturbation
corresponding to(π, 0)¦ .
Remark: By Lemma 1 (x
∗
(t)+εh
(π,ξ)
) belongs to the set K(t, t
0
, x
(ξ,ε)
) up to an error of order o(ε).
In particular for ξ = 0, the set x
∗
(t) + Q(t) can serve as an approximation to the set K(t, t
0
, x
∗
0
).
More precisely we have the following result which we leave as an exercise.
98 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
      


 
u
u
1
u
(πε)
()
u
2
u
∗
()
u
3
ε
3
ε
2
ε
1
t
0
t
1
t
2
t
3
t
f
x
x
(
ξ, ε)
x
ε
()
x
∗
()
εh
πξ
t
1
t
2
t
3
t
f
Figure 8.1: Illustration for Lemma 1.
Exercise 1: (Recall the deﬁnition of the tangent cone in 5.1.1.) Show that
Q(t) ⊂ C(K(t, t
0
, x
∗
0
), x
∗
(t)) . (8.2)
We can now prove a generalization of Theorem 1 of 7.1.
Theorem 2: Consider the optimal control problem (8.3):
Maximize ψ(x(t
f
))
subject to
dynamics: ˙ x(t) = f(t, x(t), u(t)) , t
0
≤ t ≤ t
f
,
initial condition: x(t
0
) = x
∗
0
,
ﬁnal condition: x(t
f
) ∈ R
n
,
control constraint: u() ∈ , i.e., u : [t
0
, t
f
] → Ω and
u() piecewise continuous ,
(8.3)
where ψ : R
n
→ R is differentiable and f satisﬁes the conditions listed earlier.
Let u
∗
() ∈  be an optimal control and let x
∗
(t) = φ(t, t
0
, x
∗
0
, u
∗
()), t
0
≤ t ≤ t
f
, be the
corresponding trajectory. Let p
∗
(t), t
0
≤ t ≤ t
f
, be the solution of (8.4) and (8.5):
adjoint equation: ˙ p
∗
(t) = −[
∂f
∂x
(t, x
∗
(t), u
∗
(t))]
p
∗
(t), t
0
≤ t ≤ t
f
, (8.4)
8.1. MAIN RESULTS 99
ﬁnal condition: p
∗
(t
f
) = ψ(x
∗
(t
f
)) . (8.5)
Then u
∗
() satisﬁes the maximum principle
H(t, x
∗
(t), u
∗
(t), p
∗
(t)) = M(t, x
∗
(t), p
∗
(t)) (8.6)
for all t ∈ [t
0
, t
f
] except possibly for a ﬁnite set. [Here H(t, x, u, p) = p
f(t, x, u, ), M(t, x, p) =
sup¦H(t, x, v, p)[v ∈ Ω¦].
Proof: Since u
∗
() is optimal we must have
ψ(x
∗
(t
f
)) ≥ ψ(z) for all z ∈ K(t
f
, t
0
, x
∗
0
) ,
and so by Lemma 1 of 5.1.1
ψ(x
∗
(t
f
))h ≤ 0 for all h ∈ C(K(t
f
, t
0
, x
∗
0
), x
∗
(t
f
)) ,
and in particular from (8.2)
ψ
x
(x
∗
(t
f
))h ≤ 0 for all h ∈ Q(t
f
) . (8.7)
Now suppose that (8.6) does not hold from some t
∗
∈ D
∗
∪ D. Then there exists v ∈ Ω such that
p
∗
(t
∗
)
[f(t
∗
, x(t
∗
), v) −f(t
∗
, x(t
∗
), u
∗
(t
∗
))] > 0 . (8.8)
If we consider the perturbation data π = (t
∗
; 1; v), then (8.8) is equivalent to
p
∗
(t
∗
)
h
(π,0)
(t
∗
) > 0 . (8.9)
Now from (8.4) we can see that p
∗
(t
∗
)
= p
∗
(t
f
)
Φ(t
f
, t
∗
). Also h
(π,0)
(t
f
) = Φ(t
f
, t
∗
)h
(π,0)
(t
∗
)
so that (8.9) is equivalent to
p
∗
(t
f
)
h
(π,0)
(t
f
) > 0
which contradicts (8.7). ♦
8.1.2 More general boundary conditions.
In Theorem 2 the initial condition is ﬁxed and the ﬁnal condition is free. The problem involving
more general boundary conditions is much more complicated and requires more reﬁned analysis.
Speciﬁcally, Lemma 1 needs to be extended to Lemma 2 below. But ﬁrst we need some simple
properties of the sets Q(t) which we leave as exercises.
Exercise 2: Show that
(i) Q(t) is a cone, i.e., if h ∈ Q(t) and λ ≥ 0, then λh ∈ Q(t),
(ii) for t
0
≤ t
1
≤ t
2
≤ t
f
, Φ(t
2
, t
1
)Q(t
1
) ⊂ Q(t
2
) .
Deﬁnition: Let C(t) denote the closure of Q(t).
Exercise 3: Show that
(i) C(t) is a convex cone,
(ii) for t
0
≤ t
1
≤ t
2
≤ t
f
, Φ(t
2
, t
1
)C(t
1
) ⊂ C(t
2
) .
Remark: From Lemma 1 we know that if h ∈ C(t) then (x
∗
(t) +εh) belongs to K(t, t
0
, x
∗
(t
0
)) up
to an error of order o(ε). Lemma 2, below, asserts further that if h is in the interior of C(t) then in
fact (x
∗
(t) +εh) ∈ K(t, t
0
, x
∗
(t
0
)) for ε > 0 sufﬁciently small. The proof of the lemma depends
upon some deep topological results and is omitted. Instead we offer a plausibility argument.
Lemma 2: Let h belong to the interior of the cone C(t). Then for all ε > 0 sufﬁciently small,
100 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
(x
∗
(t) +εh) ∈ K(t, t
0
, x
∗
0
) . (8.10)
Plausibility argument. (8.10) is equivalent to
εh ∈ K(t, t
0
, x
∗
(t
0
)) −¦x
∗
(t)¦ , (8.11)
where we have moved the origin to x
∗
(t). The situation is depicted in Figure 8.2.
0
ˆ
C(ε)
ˆ
K(ε)
o(ε)
K(t
1
, t
0
, x
∗
) −¦x
∗
(t)¦
h
C(t)
δε
εh
Figure 8.2: Illustration for Lemma 2.
Let
ˆ
C(ε) be the crosssection of C(t) by a plane orthogonal to h and passing through εh. Let
ˆ
K(ε) be the crosssection of K(t, t
0
, x
∗
0
) −¦x
∗
(t
0
)¦ by the same plane. We note the following:
(i) by Lemma 1 the distance between
ˆ
C(ε) and
ˆ
K(ε) is of the order o(ε);
(ii) since h is in the interior of C(t), the minimum distance between εh and
ˆ
C(ε) is δε where
δ > 0 is independent of ε.
Hence for ε > 0 sufﬁciently small εh must be trapped inside the set
ˆ
K(ε).
(This would constitute a proof except that for the argument to work we need to show that there
are no “holes” in
ˆ
K(ε) through which εh can “escape.” The complications in a rigorous proof arise
precisely from this drawback in our plausibility argument.) ♦
Lemmas 1 and 2 give us a characterization of K(t, t
0
, x
∗
0
) in a neighborhood of x
∗
(t) when we
perturb the control u
∗
() leaving the initial condition ﬁxed. Lemma 3 extends Lemma 2 to the case
when we also allow the initial condition to vary over a ﬁxed surface in a neighborhood of x
∗
0
.
Let g
0
: R
n
→ R
0
be a differentiable function such that the
0
n matrix g
0
x
(x) has rank
0
for all x. Let b
0
∈ R
n
be ﬁxed and let T
0
= ¦x[g
0
(x) − b
0
¦. Suppose that x
∗
0
∈ T
0
and let
T
0
(x
∗
0
) = ¦ξ[g
0
x
(x
∗
0
)ξ = 0¦. Thus, T
0
(x
∗
0
) + ¦x
∗
0
¦ is the plane through x
∗
0
tangent to the surface
T
0
. The proof of Lemma 3 is similar to that of Lemma 2 and is omitted also.
Lemma 3: Let h belong to the interior of the cone ¦C(t)+Φ(t, t
0
)T
0
(x
∗
0
)¦. For ε ≥ 0 let h(ε) ∈ R
n
be such that lim h(ε) = 0, and lim
ε→0
(
1
ε
)h(ε) = h. Then for ε > 0 sufﬁciently small there exists
x
0
(ε) ∈ T
0
such that
(x
∗
(t) +h(ε)) ∈ K(t, t
0
, x
0
(ε)) .
8.1. MAIN RESULTS 101
We can now prove the main result of this chapter. We keep all the notation introduced above.
Further, let g
f
: R
n
→ R
f
be a differentiable function such that g
f
x
(x) has rank
f
for all x.
Let b
f
∈ R
n
be ﬁxed and let T
f
= ¦x[g
f
(x) − b
f
¦. Finally, if x
∗
(t
f
) ∈ T
f
let T
f
(x
∗
(t
f
)) =
¦ξ[g
f
x
(x
∗
(t
f
))ξ = 0¦.
Theorem 3: Consider the optimal control problem (8.12):
Maximize ψ(x(t
f
))
subject to
dynamics: ˙ x(t) = f(t, x(t), u(t)) , t
0
≤ t ≤ t
f
,
initial conditions: g
0
(x(t
0
)) = b
0
,
ﬁnal conditions: g
f
(x(t
f
)) = b
f
,
control constraint: u() ∈ , i.e., u : [t
0
, t
f
] → Ω and
u() piecewise continuous .
(8.12)
Let u
∗
() ∈ , let x
∗
0
∈ T
0
and let x
∗
(t) = φ(t, t
0
, x
∗
0
, u
∗
()) be the corresponding trajectory.
Suppose that x
∗
(t
f
) ∈ T
f
, and suppose that (u
∗
(), x
∗
0
) is optimal. Then there exist a number
p
∗
0
≥ 0, and a function p
∗
: [t
0
, t
f
] → R
n
, not both identically zero, satisfying
adjoint equation: ˙ p
∗
(t) = −[
∂f
∂x
(t, x
∗
(t), u
∗
(t))]
p
∗
(t), t
0
≤ t ≤ t
f
, (8.13)
initial condition: p
∗
(t
0
)⊥T
0
(x
∗
0
) , (8.14)
ﬁnal condition: (p
∗
(t
f
) −p
∗
0
∇ψ(x
∗
(t
f
)))⊥T
f
(x
∗
(t
f
)) . (8.15)
Furthermore, the maximum principle
H(t, x
∗
(t), u
∗
(t), p
∗
(t)) = M(t, x
∗
(t), p
∗
(t)) (8.16)
holds for all t ∈ [t
0
, t
f
] except possibly for a ﬁnite set. [Here H(t, x, p, u) = p
f(t, x, u, ), M(t, x, p) =
sup¦H(t, x, v, p)[v ∈ Ω¦].
Proof: We break the proof up into a series of steps.
Step 1. By repeating the argument presented in the proof of Theorem 2 we can see that (8.15) is
equivalent to
p
∗
(t
f
)
h ≤ 0 for all h ∈ C(t
f
) . (8.17)
Step 2. Deﬁne two convex sets S
1
, S
2
in R
1+m
as follows:
S
1
= ¦(y, h)[y > 0, h ∈ T
f
(x
∗
(t
f
))¦,
S
2
= ¦(y, h)[y = ψ
x
(x
∗
(t
f
))h, h ∈ ¦C(t
f
) + Φ(t
f
, t
0
)T
0
(x
∗
0
)¦¦ .
We claim that the optimality of (u
∗
(), x
∗
0
) implies that S
1
∩ Relative Interior (S
2
) = φ. Suppose
this is not the case. Then there exists h ∈ T
f
(x
∗
(t
f
)) such that
ψ
x
(x
∗
(t
f
))h > 0 , (8.18)
102 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
h ∈ Interior¦C(t
f
) + Φ(t
f
, t
0
)T
0
(x
∗
0
)¦ . (8.19)
Now by assumption g
f
x
(x
∗
(t
f
) has maximum rank. Since g
f
x
(x
∗
(t
f
))h = 0 it follows that the
Implicit Function Theorem that for ε > 0 sufﬁciently small there exists h(ε) ∈ R
n
such that
g
f
(x
∗
(t
f
) +h(ε)) = b
f
, (8.20)
and, moreover, h(ε) → 0, (1/ε)h(ε) → h as ε → 0. From (8.18) and Lemma 3 it follows that for
ε > 0 sufﬁciently small there exists x
0
(ε) ∈ T
0
and u
ε
() ∈  such that
x
∗
(t
f
) +h(ε) = φ(t
f
, t
0
, x
0
(ε), u
ε
()) .
Hence we can conclude from (8.20) that the pair (x
0
(ε), u
ε
()) satisﬁes the initial and ﬁnal condi
tions, and the corresponding value of the objective function is
ψ(x
∗
(t
f
) +h(ε)) = ψ(x
∗
(t
f
)) +ψ
x
(x
∗
(t
f
))h(ε) +o([h(ε)[) ,
and since h(ε) = εh +o(ε) we get
ψ(x
∗
(t
f
) +h(ε)) = ψ(x
∗
(t
f
)) +ε)ψ
x
(x
∗
(t
f
))h +o(ε) ;
but then from (8.18)
ψ(x
∗
(t
f
) +h(ε)) > ψ(x
∗
(t
f
))
for ε > 0 sufﬁciently small, thereby contradicting the optimality of (u
∗
(), x
∗
0
).
Step 3. By the separation theorem for convex sets there exist ˆ p
0
∈ R, ˆ p
1
∈ R
n
, not both zero, such
that
ˆ p
0
y
1
+ ˆ p
1
h
1
≥ ˆ p
0
y
2
+ ˆ p
1
h
2
for all (y
i
, h
i
) ∈ S
1
, i = 1, 2 . (8.21)
Arguing in exactly the same fashion as in the proof of Lemma 1 of 7.2 we can conclude that (8.21)
is equivalent to the following conditions:
ˆ p
0
≥ 0 ,
ˆ p
1
⊥T
f
(x
∗
(t
f
)) ,
(8.22)
Φ(t
f
, t
0
)
(ˆ p
0
∇ψ(x
∗
(t
f
)) + ˆ p
1
)⊥T
0
(x
∗
0
) , (8.23)
and
(ˆ p
0
ψ
x
(x
∗
(t
f
)) + ˆ p
1
)h ≤ 0 for all h ∈ C(t
f
) . (8.24)
If we let ˆ p
∗
0
= ˆ p
0
and p
∗
(t
f
) = ˆ p
0
∇ψ(x
∗
(t
f
)) + ˆ p
1
then (8.22), (8.23), and (8.24) translate respec
tively into (8.15), (8.14), and (8.17). ♦
8.2. INTEGRAL OBJECTIVE FUNCTION 103
8.2 Integral Objective Function
In many control problems the objective function is not given as a function ψ(x(t
f
)) of the ﬁnal
state, but rather as an integral of the form
t
f
t
0
f
0
(t, x(t), u(t))dt . (8.25)
The dynamics of the state, the boundary conditions, and control constraints are the same as before.
We proceed to show how such objective functions can be treated as a special case of the problems
of the last section. To this end we deﬁned the augmented system with state variable ˜ x = (x
0
, x) ∈
R
1+m
as follows:
·
˜ x=
¸
˙ x
0
(t)
˙ x(t)
=
˜
f(t, ˜ x(t), u(t)) =
¸
f
0
(t, x(t), u(t))
f(t, x(t), u(t))
.
The initial and ﬁnal conditions which are of the form
g
0
(x) = b
0
, g
f
(x) = b
f
are augmented ˜ g
0
(˜ x) =
¸
x
0
g
0
(x)
=
˜
b
0
=
¸
0
b
0
and ˜ g
f
(˜ x) = g
f
(x) = b
f
. Evidently then the problem of maximizing (8.25) is equivalent to the
problem of maximizing
ψ(˜ x(t
f
)) = x
0
(t
f
) ,
subject to the augmented dynamics and constraints which is of the form treated in Theorem 3 of
Section 1, and we get the following result.
Theorem 1: Consider the optimal control problem (8.26):
Maximize
t
f
t
0
f
0
(t, x(t), u(t))dt
subject to
dynamics: ˙ x(t) = f(t, x(t), u(t)), t
0
≤ t ≤ t
f
,
initial conditions: g
0
(x(t
0
)) = b
0
,
ﬁnal conditions: g
f
(x(t
f
)) = b
f
,
control constraint: u() ∈  .
(8.26)
Let u
∗
() ∈ , let x
∗
0
∈ T
o
and let x
∗
(t) = φ(t, t
0
, x
∗
0
, u
∗
()), and suppose that x
∗
(t
f
) ∈ T
f
. If
(u
∗
(), x
∗
0
) is optimal, then there exists a function ˜ p
∗
= (p
∗
0
, p
∗
) : [t
0
, t
f
] → R
1+m
, not identically
zero, and with p
∗
0
(t) ≡ constant and p
∗
0
(t) ≥ 0, satisfying
(augmented) adjoint equation:
·
˜ p
∗
(t) = −[
∂
˜
f
∂˜ x
(t, x
∗
(t), u
∗
(t))]
˜ p
∗
(t) ,
initial condition: p
∗
(t
0
)⊥T
0
(x
∗
0
) ,
ﬁnal condition: p
∗
(t
f
)⊥T
f
(x
∗
(t
f
)) .
Futhermore, the maximum principle
˜
H(t, x
∗
(t), ˜ p
∗
(t), u
∗
(t)) =
˜
M(t, x
∗
(t), ˜ p
∗
(t))
holds for all t ∈ [t
0
, t
f
] except possibly for a ﬁnite set. [Here
˜
H(t, x, ˜ p, u) = ˜ p
˜
f(t, x, u) =
p
0
f
0
(t, x, u) +p
f(t, x, u), and
˜
M(t, x, ˜ p) = sup¦
˜
H(t, x, ˜ p, v)[v ∈ Ω¦.]
Finally, if f
0
and f do not explicitly depend on t, then
˜
M(t, x
∗
(t), ˜ p
∗
(t)) ≡ constant.
Exercise 1: Prove Theorem 1. (Hint: For the ﬁnal part show that (d/dt)
˜
M(t, x
∗
(t), ˜ p
∗
(t)) ≡ 0.)
104 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
8.3 Variable Final Time
8.3.1 Main result.
In the problem considered up to now the ﬁnal time t
f
is assumed to be ﬁxed. In many important
cases the ﬁnal time is itself a decision variable. One such case is the minimumtime problem where
we want to transfer the state of the system from a given initial state to a speciﬁed ﬁnal state in
minimum time. More generally, consider the optimal control problem (8.27).
Maximize
t
f
t
0
f
0
(t, x(t), u(t))dt
subject to
dynamics: ˙ x(t) = f(t, x, (t), u(t)), , t
0
≤ t ≤ t
f
,
initial condition: g
0
(x(t
0
)) = b
0
,
ﬁnal condition: g
f
(x(t)f)) = b
f
,
control constraint: u() ∈  ,
ﬁnaltime constraint: t
f
∈ (t
0
, ∞) .
(8.27)
We analyze (8.27) by converting the variable time interval [t
0
, t
f
] into a ﬁxedtime interval [0, 1].
This change of timescale is achieved by regarding t as a new state variable and selecting a new
time variable s which ranges over [0, 1]. The equation for t is
dt(s)
ds
= α(s) , 0 ≤ s ≤ 1 ,
with initial condition
t(0) = t
0
.
Here α(s) is a new control variable constrained by α(s) ∈ (0, ∞). Now if x() is the solution of
˙ x(t) = f(t, x(t), u(t)) , t
0
≤ t ≤ t
f
, x(t
0
) = x
0
(8.28)
and if we deﬁne
z(s) = x(t(s)), v(s) = u(t(s)) , 0 ≤ s ≤ 1 ,
then it is easy to see that z() is the solution of
dz
ds
(s) = α(s)f(s, z(s), v(s)) , 0 ≤ s ≤ 1 z(0) = x
0
. (8.29)
Conversely from the solution z() of (8.29) we can obtain the solution x() of (8.28) by
x(t) = z(s(t)) , t
0
≤ t ≤ t
f
,
where s() : [t
0
, t
f
] → [0, 1] is the functional inverse of s(t); in fact, s() is the solution of the
differential equation ˙ s(t) = 1/α(s(t)), s(t
0
) = 0.
8.3. VARIABLE FINAL TIME 105
With these ideas in mind it is natural to consider the ﬁxedﬁnaltime optimal control problem
(8.30), where the state vector (t, z) ∈ R
1+m
, and the control (α, v) ∈ R
1+p
:
Maximize
1
0
f
0
(t(s), z(s), v(s))α(s)ds
subject to
dynamics: ( ˙ z(s),
˙
t(s)) = (f(t(s), z(s), v(s))α(s), α(s)),
initial constraint: g
0
(z(0)) = b
0
, t(0) = t
0
,
ﬁnal constraint: g
f
(z(1)) = b
f
, t(1) ∈ R ,
control constraint: (v(s), α(s)) ∈ Ω (0, ∞)
for 0 ≤ s ≤ 1 and v(), α() piecewise continuous.
(8.30)
The relation between problems (8.27) and (8.30) is established in the following result.
Lemma 1: (i) Let x
∗
0
∈ T
0
, u
∗
() ∈ , t
∗
f
∈ (t
0
, ∞) and let x
∗
(t) = φ(t, t
0
, x
∗
0
, u
∗
()) be the
corresponding trajectory. Suppose that x
∗
(t
∗
f
) ∈ T
f
, and suppose that (u
∗
(), x
∗
0
, t
∗
f
) is optimal for
(8.27). Deﬁne z
∗
0
, v
∗
(), and α
∗
() by
z
∗
0
= x
∗
0
v
∗
(s) = u
∗
(t
0
+s(t
∗
f
−t
0
))
α
∗
(s) = (t
∗
f
−t
0
)
, 0 ≤ s ≤ 1 ,
, 0 ≤ s ≤ 1 .
Then ((v
∗
(), α
∗
()), z
∗
0
) is optimal for (8.30).
(ii) Let z
∗
0
∈ T
0
, and let (v
∗
(), α
∗
()) be an admissible control for (8.30) such that the correspond
ing trajectory (t
∗
(), z
∗
()) satisﬁes the ﬁnal conditions of (8.30). Suppose that ((v
∗
(), α
∗
()), z
∗
0
)
is optimal for (8.30). Deﬁne x
∗
0
, u
∗
() ∈ , and t
∗
f
by
x
∗
0
= z
∗
0
,
u
∗
(t) = v
∗
(s
∗
(t)) , t
0
≤ t ≤ t
∗
f
,
t
∗
f
= t
∗
(1) ,
where s
∗
() is functional inverse of t
∗
(). Then (u
∗
(), z
∗
0
, t
∗
f
) is optimal for (8.27).
Exercise 1: Prove Lemma 1.
Theorem 1: Let u
∗
() ∈ , let x
∗
0
∈ T
0
, let t
∗
f
∈ (0, ∞), and let
x
∗
(t) = φ(t, t
0
, x
∗
0
, u
∗
()), t
0
≤ t ≤ t
f
, and suppose that x
∗
(t
∗
f
) ∈ T
f
. If (u
∗
(), x
∗
0
, t
∗
f
) is optimal
for (8.27), then there exists a function ˜ p
∗
= (p
∗
0
, p
∗
) : [t
0
, t
∗
f
] → R
1+m
, not identically zero, and
with p
∗
0
(t) ≡ constant and p
∗
0
(t) ≥ 0, satisfying
(augmented) adjoint equation:
·
˜ p
∗
(t) = −[
∂
˜
f
∂˜ x
(t, x
∗
(t), u
∗
(t))]
˜ p
∗
(t) ,
(8.31)
initial condition: p
∗
(t
0
)⊥T
0
(x
∗
0
) , (8.32)
106 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
ﬁnal condition: p
∗
(t
∗
f
)⊥T
f
(x
∗
(t
∗
f
)) . (8.33)
Also the maximum principle
˜
H(t, x
∗
(t), ˜ p
∗
(t), u
∗
(t)) =
˜
M(t, x
∗
(t), ˜ p
∗
(t)) , (8.34)
holds for all t ∈ [t
0
, t
f
] except possibly for a ﬁnite set. Furthermore, t
∗
f
must be such that
ˆ
H(t
∗
f
, x
∗
(t
∗
f
), ˜ p
∗
(t
∗
f
), u
∗
(t
∗
f
)) = 0 . (8.35)
Finally, if f
0
and f do not explicitly depend on t, then
ˆ
M(t, x
∗
(t), ˜ p
∗
(t)) ≡ 0.
Proof: By Lemma 1, z
∗
0
= x
∗
0
, v
∗
(s) = u
∗
(t
0
+s(t
∗
f
−t
0
)) and α
∗
(s) = (t
∗
f
−t
0
) for 0 ≤ s ≤ 1
constitute an optimal solution for (8.30). The resulting trajectory is
z
∗
(s) = x
∗
(t
0
+s(t
∗
f
−t
0
)), t
∗
(s) = t
0
+s(t
∗
f
−t
0
), 0 ≤ s ≤ 1 , so that in particular
z
∗
(1) = x
∗
(t
∗
f
).
By Theorem 1 of Section 2, there exists a function
˜
λ
∗
= (λ
∗
0
, λ
∗
, λ
∗
n+1
) : [0, 1] → R
1+n+1
, not
identically zero, and with λ
∗
0
(s) ≡ constant and λ
∗
0
(s) ≥ 0, satisfying
adjoint equation:
˙
λ
∗
0
(t)
˙
λ
∗
(t)
˙
λ
∗
n+1
(t)
¸
¸
¸
¸
¸
¸
= −
0
¦[
∂f
0
∂z
(t
∗
(s), z
∗
(s), v
∗
(s))]
λ
∗
0
(s)
+[
∂f
∂z
(t
∗
(s), z
∗
(s), v
∗
(s))]
λ
∗
(s)¦α
∗
(s)
¦[
∂f
0
∂t
(t
∗
(s), z
∗
(s), v
∗
(s))]
λ
∗
0
(s)
+[
∂f
∂t
(t
∗
(s), z
∗
(s), v
∗
(s))]
λ
∗
(s)¦α
∗
(s)
¸
¸
¸
¸
¸
¸
¸
(8.36)
initial condition: λ
∗
(0)⊥T
0
(z
∗
0
) (8.37)
ﬁnal condition: λ
∗
(1)⊥T
f
(z
∗
(1)) , λ
∗
n+1
(1) = 0 . (8.38)
Furthermore, the maximum principle
λ
∗
0
(s)f
0
(t
∗
(s), z
∗
(s), v
∗
(s))α
∗
(s)
+λ
∗
(s)
f(t
∗
(s), z
∗
(s), v
∗
(s))α
∗
(s) +λ
∗
n+1
(s)α
∗
(s)
= sup¦[λ
∗
0
(s)f
0
(t
∗
(s), z
∗
(s), w)β
+λ
∗
(s)
f(t
∗
(s), z
∗
(s), w)β +λ
∗
n+1
(s)β][w ∈ Ω, β ∈ (0, ∞)¦
(8.39)
holds for all s ∈ [0, 1] except possibly for a ﬁnite set.
Let s
∗
(t) = (t −t
0
)/(t
∗
f
−t
0
), t
0
≤ t ≤ t
∗
f
, and deﬁne ˜ p
∗
= (p
∗
0
, p
∗
) : [t
0
, t
∗
f
] → R
1+n
by
p
∗
0
(t) = λ
∗
0
(s
∗
(t)), p
∗
(t) = λ
∗
(s
∗
(t)), t
0
≤ t ≤ t
∗
f
. (8.40)
First of all, ˜ p
∗
is not identically zero. Because if ˜ p
∗
≡ 0, then from (8.40) we have (λ
∗
0
, λ
∗
) ≡ 0 and
then from (8.36), λ
∗
n+1
≡ constant, but from (8.38), λ
∗
n+1
(1) = 0 so that we would have
˜
λ
∗
≡ 0
8.3. VARIABLE FINAL TIME 107
which is a contradiction. It is trivial to verify that ˜ p
∗
() satisﬁes (8.31), and, on the other hand (8.37)
and (8.38) respectively imply (8.32) and (8.33). Next, (8.39) is equivalent to
λ
∗
0
(s)f
0
(t
∗
(s), z
∗
(s), v
∗
(s))
+λ
∗
(s)
f(t
∗
(s), z
∗
(s), v
∗
(s)) +λ
∗
n+1
(s) = 0
(8.41)
and
λ
∗
0
(s)f
0
(t
∗
(s), z
∗
(s), v
∗
(s)) +λ
∗
(s)
f(t
∗
(s), z
∗
(s), v
∗
(s))
= Sup ¦[λ
∗
0
(s)f
0
(t
∗
(s), z
∗
(s), w) +λ
∗
(s)
f(t
∗
(s), z
∗
(s), w)][w ∈ Ω¦.
(8.42)
Evidently (8.42) is equivalent to (8.34) and (8.35) follows from (8.41) and the fact that λ
∗
n+1
(1) = 0.
Finally, the last assertion of the Theorem follows from (8.35) and the fact that
˜
M(t, x
∗
(t), ˜ p
∗
(t)) ≡
constant if f
0
, f are not explicitly dependent on t. ♦
8.3.2 Minimumtime problems
.
We consider the following special case of (8.27):
Maximize
t
f
t
0
(−1)dt
subject to
dynamics: ˙ x(t) = f(t, x(t), u(t)), t
0
≤ t ≤ t
f
initial condition: x(t
0
) = x
0
,
ﬁnal condition: x(t
f
) = x
f
,
control constraint: u() ∈  ,
ﬁnaltime constraint: t
f
∈ (t
0
, ∞) .
(8.43)
In (8.43), x
0
, x
f
are ﬁxed, so that the optimal control problem consists of ﬁnding a control which
transfers the system from state x
0
at time t
0
to state x
f
in minimum time. Applying Theorem 1 to
this problem gives Theorem 2.
Theorem 2: Let t
∗
f
∈ (t
0
, ∞) and let u
∗
: [t
0
, t
∗
f
] → Ω be optimal. Let x
∗
() be the corresponding
trajectory. Then there exists a function p
∗
: [t
0
, t
∗
f
] → R
n
, not identically zero, satisfying
adjoint equation: ˙ p
∗
(t) = −[
∂f
∂x
(t, x
∗
(t), u
∗
(t))]
p
∗
(t), t
0
≤ t ≤ t
∗
f
,
initial condition: p
∗
(t
0
) ∈ R
n
,
ﬁnal condition: p
∗
(t
∗
f
) ∈ R
n
.
Also the maximum principle
H(t, x
∗
(t), p
∗
(t), u
∗
(t)) = M(t, x
∗
(t), p
∗
(t)) (8.44)
holds for all t ∈ [t
0
, t
∗
f
] except possibly for a ﬁnite set.
Finally,
M(t
∗
f
, x
∗
(t
f
), p
∗
(t
f
)) ≥ 0 (8.45)
and if f does not depend explicitly on t then
M(t, x
∗
(t), p
∗
, (t)) ≡ constant . (8.46)
108 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
Exercise 2: Prove Theorem 2.
We now study a simple example illustrating Theorem 2. Example 1: The motion of a particle is
described by
m¨ x(t) +σ ˙ x(t) = u(t) ,
where m = mass, σ = coefﬁcient of friction, u = applied force, and x = position of the particle. For
simplicity we suppose that x ∈ R, u ∈ R and u(t) constrained by [u(t)[ ≤ 1. Starting with an
initial condition x(0) = x
01
, ˙ x(0) = x
02
we wish to ﬁnd an admissible control which brings the
particle to the state x = 0, ˙ x = 0 in minimum time.
Solution: Taking x
1
= x, x
2
= ˙ x we rewrite the particle dynamics as
¸
˙ x
1
(t)
˙ x
2
(t)
=
¸
0 1
0 −α
¸
x
1
(t)
x
2
(t)
+
¸
0
b
u(t) , (8.47)
where α = (σ/m) > 0 and b = (1/m) > 0. The control constraint set is Ω = [−1, 1].
Suppose that u
∗
() is optimal and x
∗
() is the corresponding trajectory. By Theorem 2 there exists
a nonzero solution p
∗
() of
¸
˙ p
∗
1
(t)
˙ p
∗
2
(t)
= −
¸
0 0
1 −α
¸
p
∗
1
(t)
p
∗
2
(t)
(8.48)
such that (8.44), (8.45), and (8.46) hold. Now the transition matrix function of the homogeneous
part of (8.47) is
Φ(t, τ) =
¸
1
1
α
(1 −e
−α(t−τ)
)
0 e
−α(t−τ)
,
so that the solution of (8.48) is
¸
p
∗
1
(t)
p
∗
2
(t)
=
¸
1 0
1
α
(1 −e
αt
) e
αt
¸
p
∗
1
(0)
p
∗
2
(0)
,
or
p
∗
1
(t) ≡ p
∗
1
(0) ,
and
p
∗
2
(t) =
1
α
p
∗
1
(0) +e
αt
(−
1
α
p
∗
1
(0) +p
∗
2
(0)) . (8.49)
The Hamiltonian H is given by
H(x
∗
(t), p
∗
(t), v) = (p
∗
1
(t) −αp
∗
2
(t))x
∗
2
(t) +bp
∗
2
(t)v
= e
αt
(p
∗
1
(0) −αp
∗
2
(0))x
∗
2
(t) +pb
∗
2
(t)v ,
8.3. VARIABLE FINAL TIME 109
so that from the maximum principle we can immediately conclude that
u
∗
(t) =
+1 if p
∗
2
(t) > 0,
−1 if p
∗
2
(t) < 0,
? if p
∗
2
(t) = 0 .
(8.50)
Furthermore, since the righthand side of (8.47) does not depend on t explicitly we must also have
e
αt
(p
∗
1
(0) −αp
∗
2
(0))x
∗
2
(t) +bp
∗
2
(t)u
∗
(t) ≡ constant. (8.51)
We now proceed to analyze the consequences of (8.49) and (8.50). First of all since p
∗
1
(t) ≡
p
∗
1
(0), p
∗
2
() can have three qualitatively different forms.
Case 1. −p
∗
1
(0) + αp
∗
2
(0) > 0: Evidently then, from (8.49) we see that p
∗
2
(t) must be a strictly
monotonically increasing function so that from (8.50) u
∗
() can behave in one of two ways:
either
u
∗
(t) =
−1 for t <
ˆ
t and p
∗
2
(t) < 0 for t <
ˆ
t,
+1 for t >
ˆ
t and p
∗
2
(t) > 0 for t >
ˆ
t,
or
u
∗
(t) ≡ +1 and p
∗
2
(t) > 0 for all t.
Case 2. −p
∗
1
(0) +αp
∗
2
(0) < 0 : Evidently u
∗
() can behave in one of two ways:
either
u
∗
(t) =
+1 for t <
ˆ
t and p
∗
2
(t) > 0 for t <
ˆ
t,
−1 for t >
ˆ
t and p
∗
2
(t) < 0 for t >
ˆ
t,
or
u
∗
(t) ≡ −1 and p
∗
(t) < 0 for all t.
Case 3. −p
∗
1
(0) + αp
∗
2
(0) = 0 : In this case p
∗
2
(t) ≡ (1/α)p
∗
1
(0). Also since p
∗
(t) ≡ 0, we must
have in this case p
∗
1
(0) = 0. Hence u
∗
() we can behave in one of two ways:
either
u
∗
(t) ≡ +1 and p
∗
2
(t) ≡
1
α
p
∗
1
(0) > 0 ,
or
u
∗
(t) ≡ −1 and p
∗
2
(t) ≡
1
α
p
∗
1
(0) < 0 ,
110 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
Thus, the optimal control u
∗
is always equal to +1 or 1 and it can switch at most once between
these two values. The optimal control is given by
u
∗
(t) = sgn p
∗
2
(t)
= sgn [
1
α
p
∗
1
(0) +e
αt
(−
1
α
p
∗
1
(0) +p
∗
2
(0))] .
Thus the search for the optimal control reduces to ﬁnding p
∗
1
(0), p
∗
2
(0) such that the solution of the
differential equation
˙ x = x
2
˙ x
2
= −αx
2
+b sgn[
1
α
p
∗
1
(0) +e
αt
(−
1
α
p
∗
1
(0) +p
∗
2
(0))] ,
(8.52)
with initial condition
x
1
(0) = x
10
, x
20
= x
20
(8.53)
also satisﬁes the ﬁnal condition
x
1
(t
∗
f
) = 0, x
2
(t
∗
f
) = 0 , (8.54)
for some t
∗
f
> 0; and then t
∗
f
is the minimum time.
There are at least two ways of solving the twopoint boundary value problem (8.52), (8.53), and
(8.54). One way is to guess at the value of p
∗
(0) and then integrate (8.52) and (8.53) forward in time
and check if (8.54) is satisﬁed. If (8.54) is not satisﬁed then modify p
∗
(0) and repeat. An alternative
is to guess at the value of p
∗
(0) and then integrate (8.52) and (8.54) backward in time and check of
(8.53) is satisﬁed. The latter approach is more advantageous because we know that any trajectory
obtained by this procedure is optimal for initial conditions which lie on the trajectory. Let us follow
this procedure.
Suppose we choose p
∗
(0) such that −p
∗
1
(0) = αp
∗
2
(0) = 0 and p
∗
2
(0) > 0. Then we must have
u
∗
(t) ≡ 1. Integrating (8.52) and (8.54) backward in time give us a trajectory ξ(t) where
˙
ξ
1
(t) = −
˙
ξ
2
(t)
˙
ξ
2
(t) = αξ
2
(t) −b ,
with
ξ
1
(0) −ξ
2
(0) = 0 .
This gives
ξ
1
(t) =
b
α
(−t +
e
αt
−1
α
) , ξ
2
(t) =
b
α
(1 −e
αt
) ,
which is the curve OA in Figure 8.3.
On the other hand, if p
∗
(0) is such that −p
∗
1
(0) + αp
∗
2
(0) = 0 and p
∗
2
(0) < 0, then u
∗
(t) ≡ −1
and we get
ξ
1
(t) = −
b
α
(−t +
e
αt
−1
α
) , ξ
2
(t) = −
b
α
(1 −e
αt
) ,
which is the curve OB.
8.3. VARIABLE FINAL TIME 111
B
u
∗
≡ −1
D
u
∗
≡ 1
C
ξ
1
O
ξ
2
u
∗
≡ 1
A
E
u
∗
≡ −1
F
Figure 8.3: Backward integration of (8.52) and (8.54).
Next suppose p
∗
(0) is such that −p
∗
1
(0) + αp
∗
2
(0) > 0, and p
∗
2
(0) < 0. Then [(1/α)p
∗
1
(0) +
e
αt
(−(1/α)p
∗
1
(0) + p
∗
2
(0))] will have a negative value for t ∈ (0,
ˆ
t) and a positive value for t ∈
(
ˆ
t, ∞). Hence, if we integrate (8.52), (8.54) backwards in time we get trajectory ξ(t) where
˙
ξ(t) = −ξ
2
(t)
˙
ξ
2
(t) = αξ
2
(t)+
−b for t <
ˆ
t
b for t >
ˆ
t ,
with ξ
1
(0) = 0, ξ
2
(0) = 0. This give us the curve OCD. Finally if p
∗
(0) is such that −p
∗
1
(0) +
αp
∗
2
(0) < 0, and p
∗
2
(0) < 0, then u
∗
(t) = 1 for t <
ˆ
t and u
∗
(t) = −1 for t >
ˆ
t, and we get the
curve OEF.
We see then that the optimal control u
∗
() has the following characterizing properties:
u
∗
(t) =
1 if x
∗
(t) is above BOA or on OA
−1 if x
∗
(t) is below BOA or on OB .
Hence we can synthesize the optimal control in feedback from: u
∗
(t) = ψ(x
∗
(t)) where the
B
u
∗
≡ −1
x
2
u
∗
≡ 1
x
1
A
u
∗
≡ 1
O
u
∗
≡ −1
Figure 8.4: Optimal trajectories of Example 1.
112 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
function ψ : R
2
→ ¦1, −1¦ is given by (see Figure 8.4)
ψ(x
1
, x
2
) =
1 if (x
1
, x
2
) is above BOA or on OA
−1 if (x
1
, x
2
) is below BOA or on OB .
8.4 Linear System, Quadratic Cost
An important class of problems which arise in practice is the case when the dynamics are linear and
the objective function is quadratic. Speciﬁcally, consider the optimal control problem (8.55):
Minimize
T
0
1
2
[x
(t)P(t)x(t) +u
(t)Q(t)u(t)]dt
subject to
dynamics: ˙ x(t) = A(t)x(t) +B(t)u(t), 0 ≤ t ≤ T ,
initial condition: x(0) = x
0
,
ﬁnal condition: G
f
x(t) = b
f
,
control constraint: u(t) ∈ R
p
, u() piecewise continuous.
(8.55)
In (8.56) we assume that P(t) is an n n symmetric, positive semideﬁnite matrix whereas Q(t) is
a p p symmetric, positive deﬁnite matrix. G
f
is a given
f
n matrix, and x
0
∈ R
n
, b
f
∈ R
f
are given vectors. T is a ﬁxed ﬁnal time.
We apply Theorem 1 of Section 2, so that we must search for a number p
∗
0
≥ 0 and a function
p
∗
: [0, T] → R
n
, not both zero, such that
˙ p
∗
(t) = −p
∗
0
(−P(t)x
∗
(t)) −A
(t)p
∗
(t) , (8.56)
and
p
∗
(t)⊥T
f
(x
∗
(t)) = ¦ξ[G
f
ξ = 0¦ . (8.57)
The Hamiltonian function is
H(t, x
∗
(t), ˜ p
∗
(t), v) = −
1
2
p
∗
0
[x
∗
(t)
P(t)x
∗
(t) +v
Q(t)v]
+p
∗
(t)
[A(t)x
∗
(t) +B(t)v]
so that the optimal control u
∗
(t) must maximize
−
1
2
p
∗
0
v
Q(t)v +p
∗
(t)
B(t)v for v ∈ R
p
. (8.58)
If p
∗
0
> 0, this will imply
u
∗
(t) =
1
p
∗
0
Q
−1
(t)B
(t)p
∗
(t) ,
(8.59)
whereas if p
∗
0
= 0, then we must have
p
∗
(t)
B(t) ≡ 0 (8.60)
because otherwise (8.58) cannot have a maximum.
8.5. THE SINGULAR CASE 113
We make the following assumption about the system dynamics.
Assumption: The control system ˙ x(t) = A(t)x(t) + B(t)u(t) is controllable over the interval
[0, T]. (See (Desoer [1970]) for a deﬁnition of controllability and for the properties we use below.)
Let Φ(t, τ) be the transition matrix function of the homogeneous linear differential equation ˙ x(t) =
A(t)x(t). Then the controllability assumption is equivalent to the statement that for any ξ ∈ R
n
ξ
Φ(t, τ)B(τ) = 0 , 0 ≤ τ ≤ T , implies ξ = 0 . (8.61)
Next we claim that if the system is controllable then p
∗
0
= 0, because if p
∗
0
= 0 then from (8.56)
we can see that
p
∗
(t) = (Φ(T, t))
p
∗
(T)
and hence from (8.60)
(p
∗
(t))
Φ(T, t)B(t) = 0 , 0 ≤ t ≤ T ,
but then from (8.61) we get p
∗
(T) = 0. Hence if p
∗
0
= 0, then we must have ˜ p
∗
(t) ≡ 0 which is a
contradiction. Thus, under the controllability assumption, p
∗
0
> 0, and hence the optimal control is
given by (8.59). Now if p
∗
0
> 0 it is trivial that ˆ p
∗
(t) = (1, (p
∗
(t)/p
∗
0
)) will satisfy all the necessary
conditions so that we can assume that p
∗
0
= 1. The optimal trajectory and the optimal control is
obtained by solving the following twopoint boundary value problem:
˙ x
∗
(t) = A(t)x
∗
(t) +B(t)Q
−1
(t)B
(t)p
∗
(t)
˙ p(t) = P(t)x
∗
(t) −A
(t)p
∗
(t)
x
∗
(0) = x
0
, G
f
x
∗
(T) = b
f
, p
∗
(T)⊥T
f
(x
∗
(T)) .
For further details regarding the solution of this boundary value problem and for related topics see
(See and Markus [1967]).
8.5 The Singular Case
In applying the necessary conditions derived in this chapter it sometimes happens that H(t, x
∗
(t), p
∗
(t), v)
is independent of v for values of t lying in a nonzero interval. In such cases the maximum principle
does not help in selecting the optimal value of the control. We are faced with the socalled singular
case (because we are in trouble–not because the situation is rare). We illustrate this by analyzing
Example 4 of Chapter 1.
The problem can be summarized as follows:
Maximize
T
0
c(t)dt =
T
0
(1 −s(t))f(k(t))dt
subject to
dynamics:
˙
k(t) = s(t)f(k(t)) −µk(t) , 0 ≤ t ≤ T
initial constraint: k(0) = k
0
,
ﬁnal constraint: k(t) ∈ R ,
control constraint: s(t) ∈ [0, 1], s() piecewise continuous.
114 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
We make the following assumptions regarding the production function f:
f
k
(k) > 0, f
kk
(K) < 0 for all k , (8.62)
lim
k→0
f
k
(k) = ∞ .
(8.63)
Assumption (8.62) says that the marginal product of capital is positive and this marginal product
decreases with increasing capital. Assumption (8.63) is mainly for technical convenience and can
be dispensed with without difﬁculty.
Now suppose that s
∗
: [0, T] → [0, 1] is an optimal savings policy and let k
∗
(t), 0 ≤ t ≤ T,
be the corresponding trajectory of the capitaltolabor ratio. Then by Theorem 1 of Section 2, there
exist a number p
∗
0
≥ 0, and a function p
∗
: [0, T] → R, not both identically zero, such that
˙ p
∗
(t) = −p
∗
0
(1 −s
∗
(t))f
k
(k
∗
(t)) −p
∗
(t)[s
∗
(t)f
k
(k
∗
(t)) −µ] (8.64)
with the ﬁnal condition
p
∗
(T) = 0 , (8.65)
and the maximum principle holds. First of all, if p
∗
0
= 0 then from (8.64) and (8.65) we must also
have p
∗
(t) ≡ 0. Hence we must have p
∗
0
> 0 and then by replacing (p
∗
0
, p
∗
) by (1/p
∗
0
)(p
∗
0
, p
∗
) we
can assume without losing generality that p
∗
0
= 1, so that (8.64) simpliﬁes to
˙ p
∗
(t) = −1(1 −s
∗
(t))f
k
(k
∗
(t)) −p
∗
(t)[s
∗
(t)f
k
(k
∗
(t)) −µ] . (8.66)
The maximum principle says that
H(t, k
∗
(t), p
∗
(t), s) = (1 −s)f(k
∗
(t)) +p
∗
(t)[sf(k
∗
(t)) −µk
∗
(t)]
is maximized over s ∈ [0, 1] at s
∗
(t), which immediately implies that
s
∗
(t) =
1 if p
∗
(t) > 1
0 if p
∗
(t) < 1
? if p
∗
(t) = 1
(8.67)
We analyze separately the three cases above.
Case 1. p
∗
(t) > 1, s
∗
(t) = 1 : Then the dynamic equations become
˙
k
∗
(t) = f(k
∗
(t)) −µk
∗
(t) ,
˙ p
∗
(t) = −p
∗
(t)[f
k
(k
∗
(t)) −µ] .
(8.68)
The behavior of the solutions of (8.68) is depicted in the (k, p)−, (k, t)− and (p, t)−planes in
Figure 8.5. Here k
G
, k
H
are the solutions of f
k
(k
G
) −µ = 0 and f(k
M
) −µk = 0. Such solutions
exist and are unique by virtue of the assumptions (8.62) and (8.63). Futhermore, we note from
(8.62) that k
G
< k
M
, and f
k
(k) − µ
<
> 0 according as k
<
> k
G
whereas f(k) − µk
>
< 0 according
as k
<
> k
M
. (See Figure 8.6.)
8.5. THE SINGULAR CASE 115
p
f
k
> µ
f
k
< µ
f < µk f > µk
k
k
M
k
G
l
k
k
M
t
p
l
t
Figure 8.5: Illustration for Case 1.
Case 2. p
∗
(t) < 1, s
∗
(t) = 0: Then the dynamic equations are
˙
k
∗
(t) = −µk
∗
(t) ,
˙ p
∗
(t) = −f
k
(k
∗
(t)) +µp
∗
(t) ,
giving rise to the behavior illustrated in Figure 8.7.
Case 3. p
∗
(t) = 1, s
∗
(t) =?: (Possibly singular case.) Evidently if p
∗
(t) = 1 only for a ﬁnite set of
times t then we do not have to worry about this case. We face the singular case only if p
∗
(t) = 1
for t ∈ I, where I is a nonzero interval. But then we have ˙ p
∗
(t) = 0 for t ∈ I so that from (8.66)
we get
−(1 −s
∗
(t))f
k
(k
∗
(t)) −[s
∗
(t)f
k
(k
∗
(t)) −µ] = 0 for t ∈ I ,
so
−f
k
(k
∗
(t)) +µ = 0 for t ∈ I ,
or
k
∗
(t) = k
G
for t ∈ I . (8.69)
In turn then we must have
˙
k
∗
(t) = 0 for t ∈ I so that
s
∗
(t)f(k
G
) −µK
G
= 0 for t ∈ I ,
and hence,
s
∗
(t) = µ
k
G
f(k
G
)
for t ∈ I . (8.70)
116 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
.
f
line of slope µ
µk
f(k)
k
k
M
k
G
Figure 8.6: Illustration for assumptions (8.62), (8.63).
Thus in the singular case the optimal solution is characterized by (8.69) and (8.70), as in Figure 8.8.
We can now assemble separate cases to obtain the optimal control. First of all, from the ﬁnal
condition (8.65) we know that for t close to T, p
∗
(t) < 1 so that we are in Case 2. We face two
possibilities: Either (A)
p
∗
(t) < 1 for all t < [0, T]
and then s
∗
(t) = 0, k
∗
(t) = k
0
e
−µt
, for 0 ≤ t ≤ T, or (B)
there exists t
2
∈ (0, T) such that p
∗
(t
2
) = 1 and p
∗
(t) < 1 for t
2
< t ≤ T .
We then have three possibilities depending on the value of k
∗
(t
2
):
(Bi) k
∗
(t
2
) < k
G
: then ˙ p
∗
(t
2
) < 0 so that p
∗
(t) > 1 for t < t
2
and we are in Case 1 so that
s
∗
(t) = 1 for t < t
2
. In particular we must have k
0
< k
G
.
(Bii) k
∗
(t
2
) > k
G
: then ˙ p
∗
(
2
) > 0 but then p
∗
(t
2
+ ε) > 1 for ε > 0 sufﬁciently small and since
p
∗
(T) = 0 there must exist t
3
∈ (t
2
, T) such that p
∗
(t
3
) = 1. This contradicts the deﬁnition of t
2
so that this possibility cannot arise.
(Biii) k
∗
(t
2
) − k
G
: then we can have a singular arc in some interval (t
1
, t
2
) so that p
∗
(t) =
1, k
∗
(t) = k
G
, and s
∗
(t) = µ(k
G
/f(k
G
)) for t ∈ (t
1
, t
2
). For t < t
1
we either have p
∗
(t) >
1, s
∗
(t) > 1 if k
0
< k
G
, or we have p
∗
(t) < 1, s
∗
(t) = 0 if k > k
G
.
The various possibilities are illustrated in Figure 8.9.
The capitaltolabor ratio k
G
is called the golden mean and the singular solution is called the
golden path. The reason for this term is contained in the following exercise.
Exercise 1: A capitaltolabor ratio
ˆ
k is said to be sustainable if there exists ˆ s ∈ [0, 1] such that
ˆ sf(
ˆ
k) −µ
ˆ
k = 0. Show that k
G
is the unique sustainable capitaltolabor ratio which maximizes
sustainable consumption (1 −s)f(k).
8.6. BIBLIOGRAPHICAL REMARKS 117
p
k
k
G
l
k
t
p
t
l
Figure 8.7: Illustration for Case 2.
8.6 Bibliographical Remarks
The results presented in this chapter appeared in English in full detail for the ﬁrst time in 1962 in the
book by Pontryagin, et al., cited earlier. That book contains many extensions and many examples
and it is still an important source. However, the derivation of the maximum principle given in the
book by Lee and Markus is more satisfactory. Several important generalizations of the maximum
principle have appeared. On the one hand these include extensions to inﬁnitedimensional state
spaces and on the other hand they allow for constraints on the state more general than merely initial
and ﬁnal constraints. For a uniﬁed, but mathematically difﬁcult, treatment see (Neustadt [1969]).
For a less rigorous treatment of statespace constraints see (Jacobson, et al, [1971]), whereas for a
discussion of the singular case consult (Kelley, et al. [1968]).
For an applicationsoriented treatment of this subject the reader is referred to (Athans and Falb
[1966]) and (Bryson and Ho [1969]). For applications of the maximum principle to optimal eco
nomic growth see (Shell [1967]). There is no single source of computational methods for optimal
control problems. Among the many useful techniques which have been proposed see (Lasdon, et
al., [1967]), (Kelley [1962]), (McReynolds [1966]), and (Balakrishnan and Neustadt [1964]); also
consult (Jacobson and Mayne [1970]), and (Polak [1971]).
118 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
.
p
k
k
G
1
k
t
k
G
p
t
1
Figure 8.8: Case 3. The singular case.
8.6. BIBLIOGRAPHICAL REMARKS 119
. .
.
.
.
.
. .
p
∗
1
t
T
s
∗
1
t
T
k
∗
t
T
p
∗
Case (A)
t
T t
2
t
1
s
∗
1
µk
G
f(k
G
)
t
k
∗
k
G
k
0
t
p
∗
1
t
T
t
2
s
∗
1
t
T
t
2
k
∗
t
T
t
2
p
∗
Case (Bi)
t
T
t
2
t
1
s
∗
t
k
∗
t
Case (Biii)
Figure 8.9: The optimal solution of example.
120 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
Chapter 9
Dynamic programing
SEQUENTIAL DECISION PROBLEMS: DYNAMIC PROGRAMMING FORMULATION
The sequential decision problems discussed in the last three Chapters were analyzed by varia
tional methods, i.e., the necessary conditions for optimality were obtained by comparing the op
timal decision with decisions in a small neighborhood of the optimum. Dynamic programming
(DP is a technique which compares the optimal decision with all the other decisions. This global
comparison, therefore, leads to optimality conditions which are sufﬁcient. The main advantage of
DP, besides the fact that it give sufﬁciency conditions, is that DP permits very general problem for
mulations which do not require differentiability or convexity conditions or even the restriction to a
ﬁnitedimensional state space. The only disadvantage (which unfortunately often rules out its use)
of DP is that it can easily give rise to enormous computational requirements.
In the ﬁrst section we develop the main recursion equation of DP for discretetime problems. The
second section deals with the continuoustime problem. Some general remarks and bibliographical
references are collected in the ﬁnal section.
9.1 Discretetime DP
We consider a problem formulation similar to that of Chapter VI. However, for notational conve
nience we neglect ﬁnal conditions and statespace constraints.
Maximize
N−1
¸
i=0
f
0
(i, x(i), u(i)) + Φ(x(N))
subject to
dynamics: x(i + 1) = f(i, x(i), u(i)) , i = 0, 1, . . . , N −1 ,
initial condition: x(0) = x
0
,
control constraint: u(i) ∈ Ω
i
, i = 0, 1, . . . , N −1 .
(9.1)
In (9.1), the state x(i) and the control u(i) belong to arbitrary sets X and U respectively. X and U
may be ﬁnite sets, or ﬁnitedimensional vector spaces (as in the previous chapters), or even inﬁnite
dimensional spaces. x
0
∈ X is ﬁxed. The Ω
i
are ﬁxed subsets of U. Finally f
0
(i, , ) : X U →
R, Φ : X → R, f(i, , ) : X U → X are ﬁxed functions.
121
122 CHAPTER 9. DYNAMIC PROGRAMING
The main idea underlying DP involves embedding the optimal control problem (9.1), in which
the system starts in state x
0
at time 0, into a family of optimal control problems with the same
dynamics, objective function, and control constraint as in (9.1) but with different initial states and
initial times. More precisely, for each x ∈ X and k between ) and N − 1, consider the following
problem:
Maximize
N−1
¸
i=k
f
0
(i, x(i), u(i)) + Φ(x(N)) ,
subject to
dynamics: x(i + 1) = f(i, x(i), u(i)), i = k, k + 1, . . . , N −1,
initial condition: x(k) = x,
control constraint: u(i) ∈ Ω
i
, i = k, k + 1, , N −1 .
(9.2)
Since the initial time k and initial state x are the only parameters in the problem above, we will
sometimes use the index (9.2)
k,x
to distinguish between different problems. We begin with an
elementary but crucial observation.
Lemma 1: Suppose u
∗
(k), . . . , u
∗
(N − 1) is an optimal control for (9.2)
k,x
, and let x
∗
(k) =
x, x
∗
(k + 1), . . . , x
∗
(N) be the corresponding optimal trajectory. Then for any , k ≤ ≤ N −
1, u
∗
(), . . . , u
∗
(N −1) is an optimal control for (9.2)
,x
∗
()
.
Proof: Suppose not. Then there exists a control ˆ u(), ˆ u( + 1), . . . , ˆ u(N − 1), with corresponding
trajectory ˆ x() = x
∗
(), ˆ x( + 1), . . . , ˆ x(N), such that
N−1
¸
i=
f
0
(i, ˆ x(i), ˆ u(i)) + Φ(ˆ x(N))
>
N−1
¸
i=
f
0
(i, x
∗
(i), u
∗
(i)) + Φ(x
∗
(N)) .
(9.3)
But then consider the control ˜ u(k), . . . , ˜ u(N −1) with
˜ u(i)
u
∗
(i) , i = k, . . . , −1
ˆ u(i) , i = , . . . , N −1 ,
and the corresponding trajectory, starting in state x at time k, is ˜ x(k), . . . , ˜ x(N) where
˜ x(i) =
x
∗
(i) , i = k, . . . ,
ˆ x(i) , i = + 1, . . . , N .
The value of the objective function corresponding to this control for the problem (9.2)
k,x
is
N−1
¸
i=k
f
0
(i, ˜ x(i), ˜ u(i)) + Φ(˜ x(n))
=
−1
¸
i=k
f
0
(i, x
∗
(i), u
∗
(i)) +
N−1
¸
i=
f
0
(i, ˆ x(i), ˆ u(i)) + Φ(ˆ x(N))
>
N−1
¸
i=k
f
0
(i, x
∗
(i), u
∗
(i)) + Φ(x
∗
(N)) ,
9.1. DISCRETETIME DP 123
by (9.3), so that u
∗
(k), . . . , u
∗
(N −1) cannot be optimal for 9.2)
k
, x, contradicting the hypothesis.
(end theorem)
From now on we assume that an optimal solution to (9.2)
k,x
exists for all 0 ≤ k ≤ N −1, and all
x ∈ X. Let V (k, x) be the maximum value of (9.2)
k,x
. We call V the (maximum) value function.
Theorem 1: Deﬁne V (N, ) by (V (N, x) = Φ(x). V (k, x) satisﬁes the backward recursion equa
tion
V (k, x) = Max¦f
0
, (k, x, u) +V (k
1
, f(k, x, u, ))[u ∈ Ω
k
¦, 0 ≤ k ≤ N −1 . (9.4)
Proof: Let x ∈ X, let u
∗
(k), . . . , u
∗
(N − 1) be an optimal control for (9.2)
k,x
, and let x
∗
(k) =
x, . . . , x
∗
(N) be the corresponding trajectory be x(k) = x, . . . , x(N). We have
N−1
¸
i=k
f
0
(i, x
∗
(i), u
∗
(i)) + Φ(x
∗
(N))
≥
N−1
¸
i=k
f
0
(i, x(i), u(i)) + Φ(x(N)) .
(9.5)
By Lemma 1 the lefthand side of (9.5) is equal to
f
0
(k, x, u
∗
(k)) +V (k + 1, f(k, x
∗
, u
∗
(k)) .
On the other hand, by the deﬁnition of V we have
N−1
¸
i=k
f
0
(i, x(i), u(i)) + Φ(x(N)) = f
0
(k, x, u(k))
+¦
N
¸
i=k+1
f
0
(i, x(i), u(i)) + Φ(x(N)) ≤ f
0
(k, x, u, (k)) +V (k + 1, f(k, x, u(k))¦ ,
with equality if and only if u(k +1), . . . , u(N −1) is optimal for (9.2)
k+1,x(k+1)
. Combining these
two facts we get
f
0
(k, xu
∗
(k)) +V (k + 1, f(k, x, u
∗
(k)))
≥ f
0
(k, x, u(k)) +V (k + 1, f(x, k, u(k))) ,
for all u(k) ∈ Ω
k
, which is equivalent to (9.4).(end theorem)
Corollary 1: Let u(k), . . . , u(N − 1) be any control for the problem (9.2)
k,x
and let x(k) =
x, . . . , x(N) be the corresponding trajectory. Then
V (, x()) ≤ f
0
(, x(), u()) +V ( + 1, f(, x(), u()), k ≤ ≤ N −1 ,
and equality holds for all k ≤ ≤ N −1 if and only if the control is optimal for (9.2)
k,x
.
Corollary 2: For k = 0, 1, . . . , N −1, let ψ(k, ) : X → Ω
k
be such that
f
0
(k, x, ψ(k, x)) +V (k + 1, f(k, x, ψ(k, x))
= Max¦f
0
(k, x, u) +V (k + 1, f(k, x, u))[u ∈ Ω
k
¦ .
Then ψ(k, ), k = 0, . . . , N − 1 is an optimal feedback control, i.e., for any k, x the control
u
∗
(k), . . . , u
∗
(N −1) deﬁned by u
∗
() = ψ(, x
∗
()), k ≤ ≤ N −1, where
124 CHAPTER 9. DYNAMIC PROGRAMING
x
∗
( + 1) = f(, x
∗
(), ψ(, x
∗
()), k ≤ ≤ N −1 , x
∗
(k) = x ,
is optimal for (α)
k,x
.
Remark: Theorem 1 and Corollary 2 are the main results of DP. The recursion equation (9.4) al
lows us to compute the value function, and in evaluating the maximum in (9.4) we also obtain the
optimum feedback control. Note that this feedback control is optimum for all initial conditions.
However, unless we can ﬁnd a “closedform” analytic solution to (9.4), the DP formulation may
necessitate a prohibitive amount of computation since we would have to compute and store the val
ues of V and ψ for all k and x. For instance, suppose n = 10 and the statespace X is a ﬁnite set
with 20 elements. Then we have to compute and store 10 20 values of V , which is a reasonable
amount. But now suppose X = R
n
and we approximate each dimension of x by 20 values. Then
for N = 10, we have to compute and store 10x(20)
n
values of V . For n = 3 this number is 80,000,
and for n = 5 it is 32,000,000, which is quite impractical for existing computers. This “curse of
dimensionality” seriously limits the applicability of DP to problems where we cannot solve (9.4)
analytically.
• Exercise 1: An instructor is preparing to lead his class for a long hike. He assumes that each
person can take up to W pounds in his knapsack. There are N possible items to choose from.
Each unit of item i weighs w
i
pounds. The instructor assigns a number U
i
> 0 for each
unit of item i. These numbers represent the relative utility of that item during the hike. How
many units of each item should be placed in each knapsack so as to maximize total utility?
Formulate this problem by DP.
9.2 Continuoustime DP
We consider a continuoustime version of (9.2):
Maximize
t
f
0
f
0
(t, x(t), u(t))dt + Φ(x(t
f
))
subject to
dynamics: ˙ x(t) = f(t, x(t), u(t)) , t
0
≤ t ≤ t
f
initial condition: x(0) = x
0
,
control constraint: u : [t
0
, t
f
] → Ω and u() piecewise continuous.
(9.6)
In (9.6), x ∈ R
n
, u ∈ R
p
, Ω ⊂ R
p
. Φ : R
n
→ R is assumed differentiable and f
0
, f are assumed
to satisfy the conditions stated in VIII.1.1.
As before, for t
0
≤ t ≤ t
f
and x ∈ R
n
, let V (t, x) be the maximum value of the objective
function over the interval [t, t
f
] starting in state x at time t. Then it is easy to see that V must satisfy
V (t, x) = Max¦
t+∆
t
f
0
(τ, x(τ), u(τ))dτ
+V (t + ∆, x(t + ∆))[u : [t, t + ∆] → Ω¦, ∆ ≥ 0 ,
(9.7)
and
V (t
f
, x) = Φ(x) . (9.8)
9.2. CONTINUOUSTIME DP 125
In (9.7), x(τ) is the solution of
˙ x(τ) = f(τ, x(τ), u(τ)) , t ≤ τ ≤ t + ∆ ,
x(t) = x .
Let us suppose that V is differentiable in t and x. Then from (9.7) we get
V (t, x) = Max¦f
0
(t, x, u)∆ +V (t, x) +
∂V
∂x
f(t, x, u)∆
+
∂V
∂t
(t, x)∆ +o(∆)[u ∈ Ω¦, ∆ > 0 .
Dividing by ∆ > 0 and letting ∆ approach zero we get the HamiltonJacobi Bellman partial
differentiable equation for the value function:
∂V
∂t
(t, x) + Max¦f
0
(t, x, u) +
∂V
∂x
(t, x)f(t, x, u)[u ∈ Ω¦ = 0. (9.9)
Theorem 1: Suppose there exists a differentiable function V : [t
0
, t
f
] R
n
→ R which satisﬁes
(9.9) and the boundary condition (9.8). Suppose there exists a function ψ : [t
0
, t
f
] R
n
→ Ω
with ψ piecewise continuous in t and Lipschitz in x, satisfying
f
0
(t, x, ψ(t, x)) +
∂V
∂x
f(t, x, ψ(t, x))
= Max¦f
0
(t, x, u) +
∂V
∂x
f(t, x, u)[u ∈ Ω¦ .
(9.10)
Then ψ is an optimal feedback control for the problem (9.6), and V is the value function.
Proof: Let t ∈ [t
0
, t
f
] and x ∈ R
n
. Let ˆ u : [t, t
f
] → Ω be any piecewise continuous control and
let ˆ x(τ) be the solution of
·
ˆ x (τ) = f(τ, ˆ x(τ), ˆ u(τ)) , t ≤ τ ≤ t
f
,
ˆ x(t) = x .
(9.11)
Let x
∗
(τ) be the solution of
˙ x
∗
(τ) = f(τ, x
∗
(τ), ψ(τ, x
∗
(τ))) , t ≤ τ ≤ t
f
,
x
∗
(τ) = x .
(9.12)
Note that the hypothesis concerning ψ guarantees a solution of (9.12). Let u
∗
(τ) = ψ(τ, x
∗
, (τ)), t ≤
τ ≤ t
f
. To show that ψ is an optimal feedback control we must show that
t
f
t
f
0
(tτ, x
∗
(τ), u
∗
(τ))dτ + Φ(x
∗
(τ))
≤
t
f
t
f
0
(τ, x
∗
(τ), ˆ u(τ))dτ + Φ(ˆ x(t
f
)) .
(9.13)
To this end we note that
V (t
f
, x
∗
(t
f
)) −V (t, x
∗
(t)) =
t
f
f
dV
dτ
(τ, x
∗
(τ))dτ
=∈
t
f
t
¦
∂V
∂τ
(τ, x
∗
(τ) +
∂V
∂x
˙ x
∗
(τ)¦dτ
= −
t
f
t
F −0(τ, x
∗
(τ), u
∗
(τ))dτ ,
(9.14)
126 CHAPTER 9. DYNAMIC PROGRAMING
using (9.9), (9.10). On the other hand,
V (t
f
, ˆ x(t
f
)) −V (t, ˆ x, (t)) =
t
f
t
¦
∂V
∂τ
(τ, ˆ x(τ)) +
∂V
∂x
·
˜ x (τ)¦dτ
≤ −
t
f
t
f
0
(τ, ˆ x(τ), ˆ u
∗
(τ))dτ ,
(9.15)
using (9.9). From (9.14), (9.15), (9.8) and the fact that x
∗
(t) = ˆ x(t) = x we conclude that
V (t, x) = Φ(x
∗
(t
f
)) +
t
f
t
f
0
(τ, x
∗
(τ), u
∗
(τ))
≥ Φ(ˆ x(t
f
)) +
t
f
t
f
0
(τ, ˆ x(τ), ˆ u(τ))dτ
so that (9.13) is proved. It also follows that V is the maximum value function. ♦
• Exercise 1: Obtain the value function and the optimal feedback control for the linear regula
tory problem:
Minimize
1
2
x
(T)P(T)x(t) +
1
2
T
0
¦x
(t)P(t)x(t)
+u
(t)Q(t)u(t)¦dt
subject to
dynamics: ˙ x(t) = A(t)x(t) +B(t)u(t) , 0 ≤ t ≤ T ,
initial condition: x(0) = x
0
,
control constraint: u(t) ∈ R
p
,
where P(t) = P
(t) is positive semideﬁnite, and Q(t) = Q
(t) is positive deﬁnite. [Hint:
Obtain the partial differential equation satisﬁed by V (t, x) and try a solution of the form
V (t, x) = x
R(t)x where R is unknown.]
9.3 Miscellaneous Remarks
There is vast literature dealing with the theory and applications of DP. The most elegant applications
of DP are to various problems in operations research where one can obtain “closedform” analytic
solutions to be recursion equation for the value function. See (Bellman and Dreyfus [1952]) and
(Wagner [1969]). In the case of sequential decisionmaking under uncertainties DP is about the
only available general method. For an excellent introduction to this area of application see (Howard
[1960]). For an important application of DP to computational considerations for optimal control
problems see (Jacobson and Mayne [1970]). Larson [1968] has developed computational tech
niques which greatly increase the range of applicability of DP where closedform solutions are not
available. Finally, the book of Bellman [1957] is still excellent reading. []
Bibliography
[1] J.J. Arrow and L. Hurwicz. Decentralization and Computation in Resource in resource Allo
cation. Essays in Economics and Econometrics. University of North Carolina Press, 1960. in
Pfouts R.W. (ed.).
[2] M. Athans and P.L. Falb. Optimal Control. McGrawHill, 1966.
[3] A.V. Balakrishnan and L.W. Neustadt. Computing Methods in Optimization Problems. Aca
demic Press, 1964.
[4] K. Banerjee. Generalized Lagrange Multipliers in Dynamic Programming. PhD thesis, Col
lege of Engineering, University of California, Berkeley, 1971.
[5] R.E. Bellman. Dynamic Programming. Princeton University Press, 1957.
[6] R.E. Bellman and S.E. Dreyfus. Applied Dynamic Programming. Princeton University Press,
1962.
[7] D. Blackwell and M.A. Girshick. Theory of Games and Statistical Decisions. John Wiley,
1954.
[8] J.E. Bruns. The function of operations research specilists in large urban schools. IEEE Trans.
on Systems Science and Cybernetics, SSC6(4), 1970.
[9] A.E. Bryson and Y.C. Ho. Applied Optimal Control. Blaisdell, 1969.
[10] J.D. Cannon, C.D. Cullum, and E. Polak. Theory of Optimal Control and Mathematical Pro
gramming. McGrawHill, 1970.
[11] G. Dantzig. Linear Programming and Extensions. Princeton University Press, 1963.
[12] C.A. Desoer. Notes for a Second Course on Linear Systems. Van Nostrand Reinhold, 1970.
[13] S.W. Director and R.A. Rohrer. On the design of resistance nport networks by digital com
puter. IEEE Trans. on Circuit Theory, CT16(3), 1969a.
[14] S.W. Director and R.A. Rohrer. The generalized adjoint network and network sensitivities.
IEEE Trans. on Circuit Theory, CT16(3), 1969b.
[15] S.W. Director and R.A. Rohrer. Automated network design–the frequencydomain case. IEEE
Trans. on Circuit Theory, CT16(3), 1969c.
127
128 BIBLIOGRAPHY
[16] R. Dorfman and H.D. Jacoby. A Model of Public Decisions Illustrated by a Water Pollution
Policy Problem. Public Expenditures and Policy Analysis. Markham Publishing Co, 1970. in
Haveman, R.H. and Margolis, J. (eds.).
[17] R. Dorfman, P.A. Samuelson, and R.M.Solow. Linear Programming and Economic Analysis.
McGrawHill, 1958.
[18] R.I. Dowell and R.A. Rohrer. Automated design of biasing circuits. IEEE Trans. on Circuit
Theory, CT18(1), 1971.
[19] Dowles Foundation Monograph. The Economic Theory of Teams. John Wiley, 1971. to appear.
[20] W.H. Fleming. Functions of Several Variables. AddisonWesley, 1965.
[21] C.R. Frank. Production Theory and Indivisible Commodities. Princeton University Press,
1969.
[22] D. Gale. A geometric duality theorem with economic applications. Review of Economic
Studies, XXXIV(1), 1967.
[23] A.M. Geoffrion. Duality in Nonlinear Programming: a Simpliﬁed ApplicationOriented Treat
ment. The Rand Corporation, 1970a. Memo RM6134PR.
[24] A.M. Geoffrion. Primal resource directive approaches for optimizing nonlinear decomposable
programs. Operations Research, 18, 1970b.
[25] F.J. Gould. Extensions of lagrange multipliers in nonlinear programming. SIAM J. Apl. Math,
17, 1969.
[26] J.J. Greenberg and W.P. Pierskalla. Surrogate mathematical programming. Operations Re
search, 18, 1970.
[27] H.J.Kushner. Introduction to Stochastic Control. Holt, Rinehart, and Winston, 1971.
[28] R.A. Howard. Dynamic Programming and Markov Processes. MIT Press, 1960.
[29] R. Isaacs. Differential Games. John Wiley, 1965.
[30] D.H. Jacobson, M.M. Lele, and J.L. Speyes. New necessary condtions of optimality for prob
lems with statevariable inequality constraints. J. Math. Analysis and Applications, 1971. to
appear.
[31] D.H. Jacobson and D.Q. Mayne. Differential Dynamic Programming. American Elsevier
Publishing Co., 1970.
[32] S. Karlin. Mathematical Methods and Theory in Games, Programming, and Economics, vol
ume 1. AddisonWesley, 1959.
[33] H.J. Kelley. Method of Gradients. Optimization Techniques. Academic Press, 1962. in Leit
mann, G.(ed.).
BIBLIOGRAPHY 129
[34] J.H. Kelley, R.E. Kopp, and H.G. Mayer. Singular Extremals. Topics in Optimization. Aca
demic Press, 1970. in Leitman, G. (ed.).
[35] D.A. Kendrick, H.S. Rao, and C.H. Wells. Water quality regulation with multiple polluters. In
Proc. 1971 Jt. Autom. Control Conf., Washington U., St. Louis, August 1113 1971.
[36] T.C. Koopmans. Objectives, constraints, and outcomes in optimal growth models. Economet
rica, 35(1), 1967.
[37] H.W. Kuhn and A.W. Tucker. Nonlinear programming. In Proc. Second Berkeley Symp. on
Math. Statistics and Probability. University of California Press, Berkeley, 1951.
[38] R.E. Larson. State Increment Dynamic Programming. American Elsevier Publishing Co.,
1968.
[39] L.S. Lasdon, S.K. Mitter, and A.D. Waren. The conjugate gradient method for optimal control
problems. IEEE Trans. on Automatic Control, AC12(1), 1967.
[40] E.B. Lee and L. Markus. Foundation of Optimal Control Theory. John Wiley, 1967.
[41] R. Luce and H. Raiffa. Games and Decisions. John Wiley, 1957.
[42] D.G. Luenberger. Quasiconvex programming. Siam J. Applied Math, 16, 1968.
[43] O.L. Mangasarian. Nonlinear Programming. McGrawHill, 1969.
[44] S.R. McReynolds. The successive sweep method and dynamic programming. J. Math. Analy
sis and Applications, 19, 1967.
[45] J.S. Meditch. Stochastic Optimal Linear Estimation and Control. McGrawHill, 1969.
[46] M.D. Mesarovic, D. Macho, and Y. Takahara. Theory of Hierarchical, Multilevel Systems.
Academic Press, 1970.
[47] C.E. Miller. The Simplex Method for Local Separable Programming. Recent Advance Pro
gramming. McGrawHill, 1963. in Graves, R.L. and Wolfe, P. (eds.).
[48] L.W. Neustadt. The existence of optimal controls in the absence of convexity conditions. J.
Math. Analysis and Applications, 7, 1963.
[49] L.W. Neustadt. A general theory of extremals. J. Computer and System Sciences, 3(1), 1969.
[50] H. Nikaido. Convex Structures and Economic Theory. Academic Press, 1968.
[51] G. Owen. Game Theory. W.B. Saunders & Co., 1968.
[52] E. Polak. Computational Methods in Optimization: A Uniﬁed Approach. Academic Press,
1971.
[53] L.S. Pontryagin, R.V. Boltyanski, R.V. Gamkrelidze, and E.F.Mischenko. The Mathematical
Theory of Optimal Processes. Interscience, 1962.
130 BIBLIOGRAPHY
[54] R.T. Rockafeller. Convex Analysis. Princeton University Press, 1970.
[55] M. Sakarovitch. Notes on Linear Programming. Van Nostrand Reinhold, 1971.
[56] L.J. Savage. The Foundation of Statistics. John Wiley, 1954.
[57] K. Shell. Essays in the Theory of Optimal Economic Growth. MIT Press, 1967.
[58] R.M. Solow. The economist’s approach to pollution and its control. Science, 173(3996), 1971.
[59] D.M. Topkis and A. Veinnott Jr. On the convergence of some feasible directions algorithms
for nonlinear programming. SIAM J. on Control, 5(2), 1967.
[60] H.M. Wagner. Principles of Operations Research. PrenticeHall, 1969.
[61] P. Wolfe. The simplex method for quadratic programming. Econometrica, 27, 1959.
[62] W.M. Wonham. On the seperation theorem of optimal control. SIAM J. on Control, 6(2), 1968.
[63] E. Zangwill. Nonlinear Programming: A Uniﬁed Approach. PrenticeHall, 1969.
Index
Active constraint, 50
Adjoint Equation
augmented, 98
continuoustime, 85
Adjoint equation
augmented, 105
continuoustime, 91
discretetime, 80
Adjoint network, 23
Afﬁne function, 54
Basic feasible solution, 39
basic variable, 39
Certaintyequivalence principle, 5
Complementary slackness, 34
Constraint qualiﬁcation
deﬁnition, 53
sufﬁcent conditions, 55
Continuoustime optimal control
necessary condition, 101, 103
problem formulation, 101, 103
sufﬁcient condition, 91, 125
Control of water quality, 67
Convex function
deﬁnition, 37
properties, 37, 54, 55
Convex set, 37
Derivative, 8
Design of resistive network, 15
Discretetime optimal control
necessary condition, 78
problem formulation, 77
sufﬁcient condition, 123
Discretetime optimality control
sufﬁcient condition, 80
Dual problem, 33, 58
Duality theorem, 33, 63
Dynamic programming, DP
optimality conditions, 123, 125
problem formulation, 121, 124
Epigraph, 61
Equilibrium of an economy, 45, 64
Farkas’ Lemma, 32
Feasible direction, 72
algorithm, 71
Feasible solution, 33, 49
Game theory , 5
Gradient, 8
HamiltonJacobiBellman equation, 125
Hamiltonian H,
˜
H, 78, 99
Hamiltonian H
˜
H, 101
Hypograph, 61
Knapsack problem, 124
Lagrange multipliers, 37
Lagrangian function, 35
Langrangian function, 21, 54
Langrangian multipliers, 21
Linear programming, LP
duality theorem, 33, 35
problem formulation, 31
theory of the ﬁrm, 42
Linear programming,LP
optimality condition, 34
Maximum principle
continuoustime, 86, 91, 101, 103
discretetime, 80
Minimum fuel problem, 81
Minimumtime problem, 107
131
132 INDEX
example, 108
Nondegeneracy condition, 39
Nonlinear programming, NP
duality theorem, 63
necessary condition, 50, 53
problem formulation, 49
suﬁcient condition, 54
Optimal decision, 1
Optimal economic growth, 2, 113, 117
Optimal feedback control, 123, 125
Optimization over open set
necessary condition, 11
sufﬁcient condition, 13
Optimization under uncertainty, 4
Optimization with equality constraints
necessary condition, 17
sufﬁcient condition, 21
Optimum tax, 70
Primal problem, 33
Quadratic cost, 81, 112
Quadratic programming, QP
optimality condition, 70
problem formulation, 70
Wolfe algorithm, 71
Recursion equation for dynamic programming,
124
Regulator problem, 81, 112
Resource allocation problem, 65
Separation theorem for convex sets, 73
Separation theorem for stochastic control, 5
Shadow prices, 37, 45, 70
Shadowprices, 39
Simplex algorithm, 37
Phase I, 41
Phase II, 39
Singular case for control, 113
Slack variable, 32
Statespace constraint
continuoustime problem, 117
discretetime problem, 77
Subgradient, 60
Supergradient, 60
Supporting hyperplane, 61, 84
Tangent, 50
Transversality condition
continuoustime problem, 91
discretetime problem, 80
Value function, 123
Variable ﬁnal time, 103
Vertex, 38
Weak duality theorem, 33, 58
Wolfe algorithm, 71
ii
Contents
1 INTRODUCTION 2 OPTIMIZATION OVER AN OPEN SET 3 Optimization with equality constraints 4 Linear Programming 5 Nonlinear Programming 6 Discretetime optimal control 7 Continuoustime linear optimal control 8 Coninuoustime optimal control 9 Dynamic programing 1 7 15 27 49 75 83 95 121
iii
iv CONTENTS .
several people have been using it as a text or as a reference in a course. Turin. The only obstacle was to retype the manuscript in LaTex. They have urged me to republish it.P. Van Nostrand Reinhold was then purchased by a conglomerate which cancelled Notes on System Sciences because it was not sufﬁciently proﬁtable. However. edited by George L. I would appreciate knowing if you ﬁnd any mistakes in the book. 1998 P. Varaiya v . California September. However. The idea of making it freely available over the Web was attractive because it reafﬁrmed the original aim. Notes on Optimization has been out of print for 20 years. Berkeley. I thank Kate Klohe for doing just that. accessible treatments of graduatelevel material in inexpensive books (the price of a book in the series was about ﬁve dollars). Books have since become expensive. The effort was successful for several years. Our aim was to publish short.PREFACE to this edition Notes on Optimization was published in 1971 as part of the Van Nostrand Reinhold Notes on System Sciences. or if you have suggestions for (small) changes that would improve it. the World Wide Web has again made it possible to publish cheaply.
vi CONTENTS .
(deterministic) optimal control. and mathematical economics. matrix inverse). My objective has been to present. an understanding of this material should enable the reader to follow much of the recent technical literature on nonlinear programming. Although the coverage is not encyclopedic. Varaiya vii . Athans. Cohen. and Chapter VII before Chapter VIII. However. adjoint solution) is sufﬁcient for the reader to follow the Notes. I would especially like to acknowledge the help of Professors M. Finally. and Mr. Desoer. I want to thank Professor G. as well as their presentation. Chapter V must be read before Chapter VI. The selection of topics. 1971 P. linear algebra (linear independence. To facilitate the use of these Notes as a textbook. JP. I have incurred the cost of some repetition in order to make almost all chapters selfcontained. basis. Billie Vrtiak for her marvelous typing in spite of starting from a not terribly legible handwritten manuscript. A reasonable knowledge of advanced calculus (up to the Implicit Function Theorem). The examples and exercises given in the text form an integral part of the Notes and most readers will need to attend to them before continuing further. Berkeley. in a compact and uniﬁed manner. Turin for his encouraging and patient editorship. The treatment of the topics presented here is deep. has been inﬂuenced by many of my students and colleagues. who have read and criticized earlier drafts.L. Polak. California November. I also want to thank Mrs. M.A.P.PREFACE These Notes were developed for a tenweek course I have taught for the past three years to ﬁrstyear graduate students of the University of California at Berkeley. and linear differential equations (transition matrix. E. A. Jacob. the main concepts and techniques of mathematical programming and optimal control to students having diverse technical backgrounds. Ripper. C.
viii CONTENTS .
The blend is made up of tobacco and maryjohn leaves. The set of all decisions can be adequately represented as a subset of a vector space with each vector representing a decision. “quality” is 1 . the decisions can be ranked according to whether they incur greater or lesser cost. certain further requirements must be satisﬁed. and what price p should it set? Example 2: Tough University provides “quality” education to undergraduate and graduate students. The cost corresponding to these decisions is given by a realvalued function. Furthermore. In an agreement signed with Tough’s undergraduates and graduates (TUGs). and 2. how much maryjohn and tobacco should it purchase. 1. and brieﬂy introduce two more general models which we cannot discuss further in these Notes. The set of all permissible decisions is known. An optimal decision is then any decision which incurs the least cost among the set of permissible decisions. illustrate decisionmaking situations by a few examples. For legal reasons the fraction α of maryjohn in the mixture must satisfy 0 < α < 1 . we present our model of the optimal decisionmaking problem. The cost of each decision is known. When these conditions are satisﬁed. If Potco wants to maximize its proﬁts. From extensive market research Potco has determined 2 their expected volume of sales as a function of α and the selling price p.Chapter 1 INTRODUCTION In this chapter. Some illustrations will help. whereas the cost of maryjohn is a function of the amount purchased. 1. The phrase complete information is given means that the following requirements are met: 1. tobacco can be purchased at a ﬁxed price. namely. and 2.1 The Optimal Decision Problem These Notes show how to arrive at an optimal decision assuming that complete information is given. Example 1: The Pot Company (Potco) manufacturers a smoking blend called Acapulco Gold. In order to model a decisionmaking situation in mathematical terms.
one of which is a seminar and the rest of which are lecture courses. Shell is the manager of an economy which produces one output. b a . (Obviously. and K are being measured in a common currency.001.1. Suppose that the capital stock decays exponentially with time at a rate δ > 0.2 CHAPTER 1. and the remainder I(t) to investment in capital goods. then the rate of output of wine W (t) at time is given by the production function W (t) = F (K(t). The University has a faculty of 1000. There are two factors of production. C. The labor force is growing at a constant birth rate of β > 0. L(t)) As Manager. The only requirement is that the ﬁnal road should not have a slope exceeding 0. W . Hence. INTRODUCTION deﬁned as follows: every year. Shell allocates some of the output rate W (t) to the consumption rate C(t). 1] is the fraction of output which is saved and invested. Subject to the agreements with the TUGs and WORs how many u’s and g’s should the President admit to maximize his rating? Example 3: (See Figure 1. The Weary Old Radicals (WORs) have a contract with the University which stipulates that every junior faculty member (there are 750 of these) shall be required to teach six lecture courses and two seminars each year. Figure 1. W (t) = C(t) + I(t) = (1 − s(t))W (t) where s(t) = I(t)/W (t) . If it costs $c per cubic foot to excavate or ﬁll the ground. The Regents of Touch rate Tough’s President at α points per u and β points per g “processed” by the University.) An engineer is asked to construct a road (broken line) connection point a to point b.) Thus. how should he design the road to meet the speciﬁcations at minimum cost? Example 4: Mr. The current proﬁle of the ground is given by the solid line. capital and labor. . Mr.1: Admissable set of example. whereas every senior faculty member (there are 250 of these) shall teach three lecture courses and three seminars each year. I.1) = −δK(t) + s(t)F (K(t). each u (undergraduate) must take eight courses. wine. whereas each g (graduate) must take two seminars and ﬁve lecture courses. so that the net rate of growth of capital is given by the following equation: ˙ K(t) = d K(t) dt = −δK(t) + s(t)W (t) (1. A seminar cannot have more than 20 students and a lecture course cannot have more than 40 students. ∈ [0. If K(t) and L(t) respectively are the capital stock used and the labor employed at time t. L(t)).
c = C/L. Shell’s savings policy s(t). be so as to maximize welfare? What savings policy maximizes welfare subject to the additional restriction that the capitaltolabor ratio at time T should be at least kT ? If future consumption is discounted at rate α > 0 and if time horizon ∞ is ∞. what should Mr. the welfare function becomes 0 e− αt c(t)dt. In the second example. constrained by the number of faculty and the agreements with the TUGs and WORs is a permissible decision. (In connection with this example and related models see the critique by Koopmans [1967]. (It is of mathematical but not conceptual interest to note that in this case a decision is represented by a vector in a function space which is inﬁnitedimensional. T ] is identiﬁed with total consumption 0 c(t)dt. 0 ≤ t ≤ T . F (λK. a permissible decision is any realvalued function s(t).1) and (1. (1. whence the consumption per capita of labor becomes c(t) = (l − s(t))f (k(t)). The constraints themselves are presented in terms of functional inequalities or equalities.. Shell starts with capitaltolabor T ratio ko . We must always remember that a mathematical formulation is inevitably an abstraction and the gain in precision may have occurred at a great loss of realism. L) for all λ > 0. Thus.1.1. a permissible decision is any twodimensional vector (α.3). If we deﬁne the relevant variable in terms of per capita of labor.e. 0 ≤ t ≤ T . Suppose there is a planning horizon time T . i. p) satisfying the constraints 0 < α < 1 and 0 < 2 p. constrained by 0 ≤ s(t) ≤ 1. THE OPTIMAL DECISION PROBLEM 3 ˙ L(t) = βL(t). In the last example. and at time 0 Mr. 1) = Lf (k). any vector (u.) In the examples above. then we see that F (K. Using these deﬁnitions and equations (1. in the ﬁrst example.3) where µ = (δ + β). these Notes are concerned with optimizing (i.2) it is easy to see that K(t) satisﬁes the differential equation (1. w = W/L. L)−LF (K/L. g) with u ≥ 0.) More concisely then. The ﬁrst term of the righthand side in (3) is the increase in the capitaltolabor ratio due to investment whereas the second terms is the decrease due to depreciation and increase in the labor force. maximizing or minimizing) a realvalued function over a vector space subject to constraints. and if we let f (k) = F (k. k = K/l.2) Suppose that the production function F exhibits constant returns to scale.e. l). the value of the mathematical exercise is greater the more insensitive are the optimum savings policies with respect to the simplifying assumptions of the mathematical model. Example 2 is caricature (see also a faintly related but more more elaborate formulation in Bruno [1970]). In the latter case. If “welfare” over the planning period [0. ˙ k(t) = s(t)f (k(t)) − µk(t) (1. g ≥ 0. the set of permissible decisions is represented by the set of all points in some vector space which satisfy certain constraints. λL) = λF (K. What is the optimum policy corresponding to this criterion? These examples illustrate the kinds of decisionmaking problems which can be formulated mathematically so as to be amenable to solutions by the theory presented in these Notes. For instance. . whereas Example 4 is lightyears away from reality.
One way of doing this is to assign as cost to each decision the total amount of time taken to make all the trips within this section.2 Some Other Models of Decision Problems Our model of a single decisionmaker with complete information can be generalized along two very important directions. It is customary to model this uncertainty stochastically.25 that the price is unchanged. since the stock which is likely to have higher gains is also likely to involve greater risk. The situation is different from our previous examples in that the outcome (future stock prices) is uncertain. and a decision problem can be formulated as follows. probability 0. Now it may happen that these two objective functions may be inconsistent in the sense that they may give rise to different orderings of the permissible decisions. Since we cannot study these more general models in these Notes. Thus. it may be the case that the optimum decision according to the ﬁrst criterion may be lead to very long waiting times for a few trips. For instance. we need a criterion by which we can compare different decisions. Before we can begin to suggest a design. INTRODUCTION At this point.000 be invested so as to maximize the expected value of the capital gains subject to the constraint that the probability of losing more than $100 is less than 0. In the second place. Indeed. and at the same time minimize the risk of losing his money.000 in the stock market. although convenient for presenting the mathematical theory. 1.” More abstractly. and probability 0. In this way. we need a criterion to determine what is meant by “optimum trafﬁc ﬂow. we merely point out here some situations where such models arise naturally and give some references. In the ﬁrst place. the investor may assign probability 0.1 Optimization under uncertainty. Let us suppose that we know the transportation needs of all the people in this section. A similar model is made for all the other stocks that the investor is willing to consider. consider the design of a controller for a chemical process where the decision variable are temperature. so that this decision is far from optimum according to the second criterion. may be quite artiﬁcial in practice. we can replace the single decisionmaker by a group of two or more agents whose collective decision determines the outcome.4 CHAPTER 1. which in this case are different patterns of trafﬁclight durations.1? As another example. A person wants to invest $1. 1. We can then redeﬁne the problem as minimizing the ﬁrst cost function (total time for trips) subject to the constraint that the waiting time for any trip is less than some reasonable bound (say one minute). the second goal (minimum waiting time) has been modiﬁed and reintroduced as a constraint. How should $1. The two objectives are incompatible. etc. We will see that in most of the results the objective function and the functions describing the constraints are treated in the same manner.2. An alternative and equally plausible goal may be to minimize the maximum waiting time (that is the total time spent at stop lights) in each trip. Usually there are impurities in the chemicals and disturbances in the heating process which may be regarded as additional inputs of a . This interchangeability of goal and constraints also appears at a deeper level in much of the mathematical theory.5 to the event that the price of shares in Glamor company increases by $100. He wants to maximize his capital gains.25 that it drops by $100. it is important to realize that the distinction between the function which is to be optimized and the functions which describe the constraints. the hypothesis of complete information can be relaxed by allowing that decisionmaking occurs in an uncertain environment. suppose we have to choose the durations of various trafﬁc lights in a section of a city so as to achieve optimum trafﬁc ﬂow. input rates of various chemicals.
an optimal decision problem under uncertainty is equivalent to another optimal decision problem under complete information. just as in the case of the portfolioselection problem. these problems represent one of the most important and challenging areas of research in decision theory. The control theorist will probably be most interested in Isaacs [1965]. We also need to worry about ﬁnding equivalent but simpler formulations. et al. Agent Alpha is chasing agent Beta. . known as the CertaintyEquivalence principle in economics has been extended and baptized the Separation Theorem in the control literature. Each agent does not know what the other is planning to do.. The difﬁculty caused by the lack of knowledge of the actions of the other decisionmaking agents arises even if all the agents have the same objective. known as the Theory of Games. We can only refer the reader to the extensive literature on Statistical Decision Theory (Savage [1954]. we can formulate a decision problem in such a way as to take into account these random disturbances. SOME OTHER MODELS OF DECISION PROBLEMS 5 random nature and modeled as stochastic processes. [1970] and Marschak and Radner [1971]. After this. yet the effectiveness of his decision depends crucially upon the other’s decision. it is necessary to give great attention to the various ways in which the uncertainties can be modelled mathematically. 1. (This result. it is of great significance to know that.1. What should Alpha do to get as close to Beta as possible? What should Beta do to stay out of Alpha’s reach? This situation is fundamentally different from those discussed so far. heavy car which does not maneuver easily. the number of results available is pitifully small. To do justice to these decisionmaking situations. Situations such as these have been studied extensively and an elaborate structure. The best single source for Game Theory is still Luce and Raiffa [1957]. whereas the mathematical content of the theory is concisely displayed in Owen [1968]. Although the practical impact of this theory is not great. notably economics and political science. since a particular decision taken by our agent may be better or worse than another decision depending upon the (unknown) decisions taken by the other agents.) In the author’s opinion. then in many cases the techniques presented in these Notes can be usefully applied to the resulting optimal decision problem. See Wonham [1968]. We need a new concept of rational (optimal) decisionmaking. it has proved to be among the most fruitful sources of unifying analytical concepts in the social sciences. Here there are two decisionmakers with opposing objectives. and Blaquiere. whereas Beta is riding a motor scooter. so that optimality cannot be deﬁned as we did earlier. exists which describes and prescribes behavior in these situations. to be able to deal with these models.2.2 The case of more than one decisionmaker.2. For instance. It is of crucial importance to invent schemes to coordinate the actions of the individual decisionmakers in a consistent manner. Alpha is driving a fast. Blackwell and Girshick [1954]) and on Stochastic Optimal Control (Meditch [1969]. [1969]. Kushner [1971]).) Unfortunately.. If the uncertainties are modelled stochastically as in the example above. given appropriate conditions. et al. slow but with good maneuverability. we need a good background in Statistics and Probability Theory besides the material presented in these Notes. The place is a large circular ﬁeld. (See Mesarovic. Although problems involving many decisionmakers are present in any system of large size. however.
6 CHAPTER 1. INTRODUCTION .
.3 Matrices are normally denoted by capital letters. but no confusion will result. n. 0 denotes both the zero vector and the real number zero. . if xi ≥ 0. . yn ) √ then x y = x1 y1 + . If confusion is likely. 2.3 and 2. . . Then we study our example. . the properties of whose solution are stated as a theorem. . . with two consistent exceptions mentioned in 2. .1. . xn ). .1. and x = (x1 .1 All vectors are column vectors. Aj denotes the entry i of A in the ith row and jth column. In particular if x ∈ Rn . Thus if x = (x1 . . i = 1. This will generalize to a canonical problem. I denotes the identity matrix. and different vectors denoted by the same symbol are distinguished by superscripts as in xj and xk . + xn yn as in ordinary n we deﬁne x = + x x. yn ) then x ≥ y means xi ≥ yi . . matrix multiplication. . .1 Notation 2. 2. . and then we also write A = {aij }.2 If x = (x1 . 7 .1. this entry is sometimes also denoted by the lower case letter aij . Some additional properties are mentioned in the last section.5 below and some other minor and convenient exceptions in the text. .1. xn ) and y = (y1 . xn ) and y = (y1 . then x ≥ 0. . . . If A is an m × n matrix. Note that Ai is a row vector. . then Aj denotes the jth column of A. Vectors are normally denoted by lower case letters. We ﬁrst establish some notation which will be in force throughout these Notes. . . i = 1. . . . Prime denotes transpose so that if x ∈ Rn then x is the row vector x = (x1 . xn ) .1. If x ∈ R 2. . n. the ith component of a vector x ∈ Rn is denoted xi . . . we write In to denote the n × n identity matrix. and Ai denotes the ith row of A. . . its size will be clear from the context.Chapter 2 OPTIMIZATION OVER AN OPEN SET In this chapter we study in detail the ﬁrst example of Chapter 1. . . .
(∂f /∂xn )(ˆ)). . y ) is the ndimensional x ˆ row vector fx (ˆ. . x x 2.5 above. . . .1. y ). fm . . y ) = ((∂f /∂y1 )(ˆ.1. .2 Example We consider in detail the ﬁrst example of Chapter 1. For example. in terms of the notation in Section 2. . ∂fm x ∂xn (ˆ) 2. ˆ (ˆ) = fx x = x . = . .1. 2. . . . (∂f /∂ym )(ˆ. v = total amount of mixture produced. ˆ x x This derivative is denoted by (∂f /∂x)(ˆ) or fx (ˆ) or ∂f /∂xx=ˆ or fx x=ˆ . Thus. we can write F : x → Ax to denote the function f : Rn → Rm whose value at a point x ∈ Rn is Ax. .5 If f : Rn → R is a differentiable function. y ) = ((∂f /∂x1 )(ˆ.6 If f : Rn → R is twice differentiable. . . OPTIMIZATION OVER AN OPEN SET 2. Deﬁne the following variables and functions: α = fraction of maryjohn in proposed mixture. y) → f (x. . f (α. . y )).1. then its derivative at x is the m × n matrix ˆ x f1x (ˆ) ∂f . and if the argument x x x ˆ x x is clear from the context it may be dropped. ∂fm (ˆ) x ∂x1 ∂f1 x ∂xn (ˆ) . if A is an m × n matrix. y )). Note that fi : Rn → R. . . y ) = (∂f /∂y)(ˆ. m. . . .4 If f : Rn → Rm is a function.. and similarly x ˆ x ˆ x ˆ x ˆ fy (ˆ. y) is a differentiable function from ˆ Rn × Rm into R. p). In this case we write f : x → f (x).. the derivative of f at x is the row vector ((∂f /∂x1 )(ˆ). p = sale price per pound of mixture. x ∂x1. Finally. ∂x fmx (ˆ) x ∂f1 (ˆ) . p) = expected sales volume (as determined by market research) of mixture as a function of(α. . . y ) = (∂f /∂x)(ˆ. If f : (x. x x i x fxx (ˆ) = (∂/∂x)(fx ) (ˆ). i = 1. The column vector (fx (ˆ)) is also denoted x f (ˆ). Sometimes we describe a function by specifying a rule to calculate f (x) for every x.8 CHAPTER 2. its second derivative at x is the n×n matrix (∂ 2 f /∂x∂x)(ˆ) = ˆ x fxx (ˆ) where (fxx (ˆ))j = (∂ 2 f /∂xj ∂xi )(ˆ). its ith component is written fi . x x and is called the gradient of f at x. y ). (∂f /∂xn )(ˆ. . . . if f : Rn → Rm is x ˆ x ˆ x ˆ x ˆ a differentiable function with components f1 . . the partial derivative of f with respect to x at the point (ˆ.
p) = pf (α. h2 ) in R2 there exists η > 0 (η of course depends on h) such that ((α∗ . p) ∈ Ω whenever (α. i. m = αv. p). p∗ ) ∈ Ω and N (α∗ . where Ω = {(α. p∗ ) < ε (2. subject to (α.1) We are going to establish some properties of (a∗ . p). (α∗ . p∗ ) + δ(h1 .2. 9 Evidently. h2 )) ∈ Ω for 0 ≤ δ ≤ η (2.2) implies that for every vector h = (h1 . EXAMPLE Since it is not proﬁtable to produce more than can be sold we must have: v = f (α. Then the total cost as a function of α. p) − C(α. p∗) is an optimal decision.2) In turn (2. (2. Formally. and t = amount (in pounds) of tobacco purchased. p). p∗ ). Suppose that (α∗ .3) . p) = P1 (αf (α.. p). and t = (l − α)v. so that the net proﬁt is N (α. and P2 = purchase price per pound of tobacco.2. p) − (α∗ .e. p) ∈ Ω. p) for all (α. p)0 < α < 1 . p) ∈ Ω. First of all we note that Ω is an open subset of R2 . 0 < p < ∞}. we 2 have the the following decision problem: Maximize N (α. p) = R(α. Let P1 (m) = purchase price of m pounds of maryjohn. The set of admissible decisions is Ω. p is C(α. p). m = amount (in pounds) of maryjohn purchased. p∗ ) ≥ N (α. p)) + P2 (1 − α)f (α. Hence there exits ε > 0 such that (α. The revenue is R(α.
p ) = 0. p∗ ) + δ(h1 .9) Before evaluating the usefulness of property (2.6) we get 0 ≥ [ ∂N (α∗ .6) Substitution of (2. p∗ )h1 + ∂α +o(δ). and δ is open. p )h2 ] + o(δ). we have concluded that the inequality (2. p ) = 0.8) Thus.7) Letting δ approach zero in (2. p )h2 ]. where oδ δ ∂N ∗ ∗ ∂p (α .9) holds for every vector h ∈ R2 . p∗ ) N (α∗ + δh1 . OPTIMIZATION OVER AN OPEN SET 1 2 (α∗ .3) with (2. (2. p∗ )  h p Figure 2.10 α CHAPTER 2.1) we obtain (2. (2. p∗ + δh2 ) = +δ[ ∂N (δ∗ . ∂N ∗ ∗ ∂p (α . and using (2.4) yields 0 ≥ δ[ ∂N (α∗ . p )h2 ] (2. Clearly this is possible only if ∂N ∗ ∗ ∂α (α . p∗ )h1 + ∂α ∂N ∗ ∗ ∂p (α .4): N (α∗ .1: Admissable set of example. p∗ + δh2 ) for 0 ≤ δ ≤ η Now we assume that the function N is differentiable so that by Taylor’s theorem N (α∗ . p )h2 ] ∂N ∗ ∗ ∂p (α . p∗ )h1 + ∂α ∂N ∗ ∗ ∂p (α .8). let us prove a direct generalization.7). (2. p∗ ) is optimal. Combining (2. h2 ) . (2. Ω δh (a∗ .5) into (2. . + o(δ) δ .5) → 0 as δ → 0. using the facts that N is differentiable. (α∗ .4) (2. p∗ )h1 + ∂α Dividing by δ > 0 gives 0 ≥ [ ∂N (α∗ . p∗ ) ≥ N (α∗ + δh1 .
Since x∗ is optimal.12) implies that for every vector h ∈ Rn there exits η > 0 (η depending on h) such that (x∗ + δh) ∈ Ω whenever 0 ≤ δ ≤ η.12) In turn.14) + o(δ). there exists ε > 0 such that x ∈ Ω whenever x − x∗  < ε. we must then have f (x∗ ) ≥ f (x∗ + δh) whenever 0 ≤ δ ≤ η. we must have 0= and the theorem is proved. THE MAIN RESULT AND ITS CONSEQUENCES 11 2. by Taylor’s theorem we have f (x∗ + δh) = f (x∗ ) + where o(δ) δ ∂f ∗ ∂x (x )δh (2.2.1 Theorem . we see that 0≥ ∂f ∗ ∂x (x )h. (2.18) must hold for every h ∈ Rn . (2.11) Proof: Since x∗ ∈ Ω and Ω is open.16) into account.17) Letting δ approach zero in (2. Let Ω be an open subset of Rn . (2.15) into (2.10) = 0.3 The Main Result and its Consequences 2.3. (2. (2.3.14) yields 0 ≥ δ ∂f (x∗ )h + o(δ) ∂x and dividing by δ > 0 gives 0≥ ∂f ∗ ∂x (x )h + o(δ) δ (2. ∂f ∗ ∂x (x ).17) and taking (2. Let f : Rn → R be a differentiable function.16) Substitution of (2. Then ∂f ∗ ∂x (x ) (2. Since f is differentiable.18) Since the inequality (2. ♦ . Let x∗ be an optimal solution of the following decisionmaking problem: Maximize f (x) subject to x ∈ Ω.15) → 0 as δ → 0 (2.13) (2.
in which case (2. marginal revenue = marginal cost. In each case Ω = (−1.19). p ). OPTIMIZATION OVER AN OPEN SET Table 2.1 A warning. . p ) = ∂C ∗ ∗ ∂α (α . However.12 CHAPTER 2. . .4.11) gives us n equations which must be satisﬁed at any optimal decision x∗ = (x∗ . Suppose that R and C are differentiable. and it does not give us sufﬁcient conditions either.2. Note that in the last three ﬁgures there is no optimal decision since the limit points 1 and +1 are not in the set of permissible decisions Ω = (−1. so that the search for an optimal decision from Ω is reduced to searching among the solutions of (2.1 may x ˜ occur. Let us evaluate the usefulness of (2.11) and its special case (2. every optimal decision must be a solution of these n simultaneous equations of n variables. or. say x∗ More than one point None Exactly one point More than one point Case 1 2 3 4 5 Does there exist an optimal decision for 2. .18). in these Notes we shall not be overly concerned with numerical solution techniques (but see 2.1? Yes Yes No No No Further Consequences x∗ is the unique optimal 2. . The theorem may also have conceptual signiﬁcance. in the language of economic analysis. We have obtained an important economic insight. p∗ ) ∂R ∗ ∗ ∂α (α .2.1 illustrate these cases.4 Remarks and Extensions 2.2 Consequences. ∂f ∗ ∂x2 (x ) = 0.18) implies that at every optimal decision (α∗ .1 At how many points in Ω is 2. More generally. Equation (2. In summary. p ) = ∂C ∗ ∗ ∂p (α . . Equation (2.11) is only a necessary condition for x∗ to be optimal. 1). p ).3.2 satisﬁed? Exactly one point. . We return to the example and recall the N = R − C. The diagrams in Figure 2. the theorem does not give us any clues concerning the existence of an optimal decision. any one of the ﬁve cases in Table 2. ∂f ∗ ∂xn (x ) =0 (2. ∂R ∗ ∗ ∂p (α .19) Thus. In practice this may be a very difﬁcult problem since these may be nonlinear equations and it may be necessary to use a digital computer.6 below). x∗ ) . 2.4. . 1). n 1 These are ∂f ∗ ∂x1 (x ) = 0. . There may exist decisions x ∈ Ω ˜ such that fx (˜) = 0 but x is not optimal.
e. Indeed. Suppose f is twicedifferentiable and let x∗ ∈ Ω be optimal or even locally optimal.2. . and by Taylor’s theorem f (x∗ + δh) = f (x∗ ) + 1 δ2 h fxx (x∗ )h + o(δ2 ). It is easy to see that the theorem holds (i. 2. We say that x∗ ∈ Ω is a locally optimal decision if there exists ε > 0 such that f (x∗ ) ≥ f (x) whenever x ∈ Ω and x∗ − x ≤ ε. Suppose at x∗ ∈ Ω. then +1 is the optimal decision but the derivative is positive at that point. Then fx (x∗ ) = 0. 2 2 (2. if Ω = [−1. we get an additional necessary condition. This means that fxx (x∗ ) is a negative semideﬁnite matrix.4. Thus. and if f is continuous.20) we can conclude that x∗ is a local optimum.5 Sufﬁciency for local optimal..3 Local optimum.4 Secondorder conditions. If the set of permissible decisions Ω is a closed and bounded subset of Rn .4. if we have a twice differentiable objective function.11) for local optima also.4. in the third ﬁgure above. Now for δ > 0 sufﬁciently small f (x∗ + δh) ≤ f (x∗ ). 1]. But if Ω is closed we cannot assert that the derivative of f vanishes at the optimum. 2. so that dividing δ by δ2 > 0 yields 0 ≥ 1 h fxx (x∗ )h + 2 o(δ2 ) δ2 and letting δ approach zero we conclude that h fxx (x∗ )h ≤ 0 for all h ∈ Rn .2: Illustration of 4. 2.4. REMARKS AND EXTENSIONS 13 1 Case 1 1 1 Case 2 1 1 Case 3 1 1 Case 4 1 1 Case 5 1 Figure 2.2 Existence.4. fx (x∗ ) = 0 and fxx is strictly negative deﬁnite. 2. But then from the expansion (2.20) where o(δ2 ) → 0 as δ → 0.1. then it follows by the Weierstrass Theorem that there exists an optimal decision. 2.
If x f (xi ) = 0. one choice is to take di to be an optimal decision for the following problem: Max{f (xi + d x f (xi ))d > 0.11. At any point x ∈ Ω the gradient x f (˜) is a direction along which f (x) increases.6 A numerical procedure. (xi + d x f (xi )) ∈ Ω}.4. fx (x∗ ) = 0.. .14 CHAPTER 2. Another choice is to let di = di−1 if f (xi + di−1 x f (xi )) > f (xi ). i 2. For other numerical procedures the reader is referred to Zangwill [1969] or Polak [1971]. For instance. The step size di can be selected in many ways. We can formalize the scheme as an algorithm. otherwise let di = 1/k di−1 where k is the smallest positive integer such that f (xi + 1/k di−1 x f (xi )) > f (xi ). Step 2. f (xi+1 ) > f (xi ) if xi+1 = xi . Then 1. This observation suggests the following scheme for x x ﬁnding a point x∗ ∈ Ω which satisﬁes 2.e. This requires a onedimensional search. Exercise: Let f be continuous differentiable. To start the process we let d−1 > 0 be arbitrary. Go to Step 2. Pick x0 ∈ Ω. Step 3. stop. if x∗ ∈ Ω is a limit point of the sequence {xi }. Let {di } be produced by either of these choices and let {xi } be the resulting sequence. Step 1. Calculate x f (xi ). Otherwise let xi+1 = xi + di x f (xi ) and go to Step 3. Set i = 0. f (˜ + ε x ˜ x x f (˜)) > f (˜) for all ε > 0 sufﬁciently small. OPTIMIZATION OVER AN OPEN SET 2. Set i = i + 1 and return to Step 2. i.
3.1 (3. Y ∗ ) = 0. Let us suppose y ∗ = 0. (3.3) In particular this implies that y ∗ = g(x∗ ). Additional properties are summarized in Section 3 and a numerical scheme is applied to determine the optimal design of resistive networks.1 Example We want to ﬁnd the rectangle of maximum area inscribed in an ellipse deﬁned by f1 (x. y) = α and (x. if (x∗ . (ii) an open set V containing (x∗ . so that this assertion follows from the Implicit Function Theorem. y) for all (x. x∗ + ε) → V such that f1 (x. y ∗ ) is an optimal decision we cannot assert that f0 (x∗ . y ∗ ). y) = α}.1) The problem can be formalized as follows (see Figure 3. y) = 4xy subject to (x.1): Maximize f0 (x. y)f1 (x. y ∗ ) ≥ f0 (x. y ∗ ) is an optimal decision.Chapter 3 OPTIMIZATION OVER SETS DEFINED BY EQUALITY CONSTRAINTS We ﬁrst study a simple example and examine the properties of an optimal decision.2) The main difference between problem (3. This will generalize to a canonical problem. Hence. y) = x2 a2 y2 b2 + = α. y) ∈ V iff f y = g(x).2) and the decisions studied in the last chapter is that the set of permissible decisions Ω is not an open set. The assertion is false if y ∗ = 0. Since Note that y ∗ = 0 implies f1y (x∗ . Returning to problem (3.2). and that f1 (x. Clearly then either x∗ = 0 or y ∗ = 0. suppose (x∗ . Then from ﬁgure 3. and the properties of its optimal decisions are stated in the form of a theorem. g(x)) = α whenever x − x∗  < ε. (3. In the present case let 0 < ε ≤ a − x∗ and g(x) = +b[α − (x/a)2 ]1/2 . and (iii) a differentiable function g : (x∗ − ε.1 it is evident that there exist (i)ε > 0. y ∗ ). y) ∈ Ω = {(x. 1 15 . y) in an open set containing (x∗ .
2). g(x)) subject to x − x∗  < ε. y ∗ ) + f0y (x∗ . and since f1y (x∗ .6): −1 f0x − f0y f1y f1x = 0 at (x∗ . it follows that x∗ is an optimal solution for (3. y ∗ ). (x∗ . y ∗ ) = (x∗ . f0x (x f0x (x∗ . y∗ = + 1/2 b. y ∗ ) (f1x . y ∗ )gx (x∗ ) = 0 Using the fact that f1 (x.1: Illustration of example.5) and substitute in (3. which we can also express as ˆ so that by Theorem 2. y ∗ ) = 0 we can evaluate gx (x∗ ). y ∗ ) = α and (3. Solving these yields x∗ = + 1/2 a − (α/2) .6).16 CHAPTER 3. (3. −1 gx (x∗ ) = −f1y f1x (x∗ .6) Thus an optimal decision (x∗ . y ∗ ). OPTIMIZATION WITH EQUALITY CONSTRAINTS Tangent plane to Ω at (x∗ . (3.1.4) is an open set (in R1 ) and the objective function f0 is differentiable. y ∗ ) must satisfy the two equations f1 (x∗ . y ∗ ) + f1y (x∗ .4) ˆ But the constraint set in (3.5) to obtain the condition (3. ∗ ) = 0. − (α/2) . (3.4): ˆ Maximize f0 (x) = f0 (x. g(x)) ≡ α for x − x∗  < ε. f1y ) y∗ g(x) V ( x∗ x  ) Ω Figure 3. y ∗ )gx (x∗ ) = 0. we see that f1x (x∗ .3. g(x∗ )) is optimum for (3.
. (3. (3. GENERAL CASE Evidently there are two optimal decisions. m (m < n). m. y ∗ ) = m(α) = 2αab.1 Theorem. Finally we note that λ∗ = where m(α) = maximum area. we can assume that the m × m matrix [(∂fi /∂xj )(x∗ )]. then by relabeling the coordinates of x if necessary. i = 1. are linearly independent. If x∗ (α) is a differentiable function of α then m(α) is a differentiable function of α. . Let x∗ (α) be an optimal decision for (3. αm ) .3. y ∗ )]λ∗ .12): Maximize f0 (x) subject to fi (x) = αi . y ∗ ) = [ f1 (x∗ . b). . + λ∗ fmx (x∗ ) m 1 (3. i = 1. Deﬁne −1 λ∗ = f0y f1y (x∗ .9) which means that at an optimal decision the gradient of the objective function f0 is normal to the plane tangent to the constraint set Ω. . . . Since fix (x∗ ).2 General Case 3. Then there exists a vector λ∗ = (λ∗ .13) Furthermore. i = 0. f1y ) at (x∗ . be continuously differentiable functions and let x∗ be an optimal decision of problem (3. are linearly independent. .7) (3.2. .8) Then (3. . f1 . . and (λ∗ ) = ∂m ∂α (3.9) is equivalent to f0 (x∗ . . i = 1. . . (x∗ . . . m. m. (ii) an .2. . f0y ) = λ∗ (f1x .12) as a function of α = (α1 . . . . . By the Implicit Function Theorem (see Fleming [1965]) it follows that there exist (i) ε > 0.12). λ∗ ) such that m 1 f0x (x∗ ) = λ∗ f1x (x∗ ) + . . Let fi : Rn → R. αm ) be the maximum value of (3.12) Suppose that at x∗ the derivatives fix (x∗ ).8) can be rewritten as (3. j ≤ m.6) and (3. let m(α1 . . . + 1/2 (a. The condition (3. 1.11) 3. . (3. − (α/2) 17 and the maximum area is (3. . .9): (f0x .14) Proof. . y ∗ ). . .6) can be interpreted differently. (3. .10) (3. is nonsingular. y ∗ ) In terms of the gradients of f0 . 1 ≤ i. ∂m ∂α .
. . .12).17): ˆ Maximize f0 (u) = f0 (g(u). . . u) subject to u ∈ U. and substitute in (3. . . u∗ ) = (g(u∗ ). .1 . . Differentiating (3.15) For convenience. In particular this implies that x∗ = gj (x∗ . Since x = (w. . . . where U = [(xm+1 . . u = (xm+1 .19) and (3.20) can be written as (3. gu (u∗ ) = −[fw (x∗)]−1 fu (x∗ ). . xn ) and f = (f1 . xn )] xm+ − x∗  < ε. we see that fw (x∗ )gu (u∗ ) + fu (x∗ ) = 0. . . xn ) = αi . let us deﬁne w = (x1 . (3. so that by Theorem 2. and since the m × m matrix fw (x∗ ) is nonsingular we can evaluate gu (u∗ ). . . . + λ∗ fmx (x∗ ). . . . . . .3. . xn ) ∈ V iff xj = gj (xm+1 . xm+1 . . such that m+ fi (x1 . . 1 ≤ j ≤ m. . . this is the same as f0x (x∗ ) = (λ∗ ) fx (x∗ ) = λ∗ f1x (x∗ ) + .18) to obtain the condition −1 −f0w fw fu + f0u = 0 at x∗ = (w∗ .19) Next. x∗ ). . fu (x∗ )). .2). u∗ ). which we can also express using the chain rule for derivatives as ˆ f0u (u∗ ) = f0w (x∗ )gu (u∗ ) + f0u (x∗ ) = 0. . .21): (f0w (x∗ ). . m. . f0u (u∗ ) = 0. OPTIMIZATION WITH EQUALITY CONSTRAINTS open set V in Rn containing x∗ . .18) (3. . . . xn ).21) . xn ) ∈ U (see Figure 3. xm ) . m 1 (3. 1 ≤ i ≤ m. . f0u (x∗ )) = (λ∗ ) (fw (x∗ ). (3. it follows that u∗ is an optimal decision for (3. . . xn ) . . .16) with respect to u = (xm+1 . and (iii) a differentiable function g : U → Rm . and n m+1 j fi (g(xm+1 . = 1. . . fm ) . Then. xn ) = αi . u).18 CHAPTER 3. . and (x1 .16) (3. deﬁne the mdimensional column vector λ∗ by −1 (λ∗ ) = f0w fw x∗ .17) ˆ But U is an open subset of Rn−m and f0 is a differentiable function on U (since f0 and g are ˆ differentiable). .20) Then (3. . . 1 ≤ j ≤ m. . . i = 1. . . and (xm+1 . n − m]. u∗ ) is optimal for (3. (3. since x∗ = (w∗ . . xn ). . (3. .
2. −1 −f0w fw fu + f0u = 0 at (w∗ (α). .22) with respect to α gives ∗ fw wα + fu u∗ = I. .x . α so that ∗ −1 −1 wα + fw fu u∗ = fw . x∗ (α)) and u∗ (α) = (x∗ (α). x∗ α)) . . m . . . . ∗ V xm+1 U xn 2 Figure 3. say α. (xm+1 . . . By hypothesis.14). . .22) (3. x∗ ) n m+1 (3. . . . Also. . . .2: Illustration of theorem. . . u∗ (α)). say N . . fw is nonsingular at m 1 m+1 ( ∗ (α). To prove (3. m(α) = f0 (x∗ (α)).13). xn ) (x∗ . .23) for α ∈ N . x∗ ) m 1 Ω= {xf i (x) = αi } i = 1. xn ) (x∗ . . which is equation (3. . it follows that f is x w nonsingular at x∗ (α) in a neighborhood of α. . . . α . so that ∗ mα = f0w wα + f0u u∗ α (3. we vary α in a neighborhood of a ﬁxed value. . GENERAL CASE x1 . . . u∗ (α)) = α. We deﬁne w∗ (α) = (x∗ (α). We have the equation f (w∗ (α). . . .3. Since f (x) and x∗ (α) are continuously differentiable by hypothesis. . x m 19 g(xm+1 . . .24) Differentiating (3.
the m equations f (w. we obtain (3.3 Algebraic interpretation. x x x where −1 ˜ λ = f0w fw˜ . together with (3.20) and (3.12).28) in the last section. . Then the Implicit Function ˜ ˜ ˜ Theorem enables us to solve. . this equation can be rewritten as ∗ −1 f0w wα + f0u u∗ = f0w fw . u). . xn ) . + λm x fm (x∗ )λi ∈ R. . . i = 1. Suppose that fw (˜) is nonsinx gular at some point x = (w.27) ˜ and (3. α Using (3. . if we substitute from (3. . As u varies.2 Geometric interpretation. α (3.13).26) f1 (x∗ ) + . . m}. The ˆ at u is derivative of f0 ˜ ˆ u ˜ f0u (˜) = f0w gu + f0u˜ = −λ fu (˜) + f0u (˜). .25) ♦ In (3. 3. the direction of steepest increase of f0 at u is ˜ u u f0 (˜) ˆ = −fu (˜)λ + fOu (˜) . We shall use (3.27) is equation (3. .2. the gradient of the objective function x f0 (x∗ ) is normal to the tangent surface (3. The equality constraints of the problem in 3. .20 CHAPTER 3. u f0 (˜) = 0 which. u) = α.2.27) ˆ Therefore. Let us again deﬁne w = (x1 . .25). OPTIMIZATION WITH EQUALITY CONSTRAINTS and multiplying on the left by f0w gives ∗ −1 −1 f0w wα + f0w fw fu u∗ = f0w fw .28) ˆ u and if u is optimal. i = 1. . The hypothesis of linear independence of {fix (x∗ )1 ≤ i ≤ m} guarantees that the tangent plane through Ω at x∗ is described by {hfix (x∗ )h = 0 .23). in a neighborhood of x. and the objective function changes according to f0 (u) = f0 (g(u). .13) is therefore equivalent to saying that at an optimal decision x∗ . i = 1. m}. . Condition (3. . u) in Ω which is not necessarily optimal.12 deﬁne a n − m dimensional surface Ω = {xfi (x) = αi . .24). m}. w must change according to w = g(u) (in order to ˜ ˆ maintain f (w. xm ) and u = (xm+1 . . u) = α). so that the set of (column vectors orthogonal to this tangent surface is {λ1 x (3. . x (3. . u can then vary ˜ arbitrarily in a neighborhood of u. x ˜ x (3. . 3. . .14) and the theorem is proved.
2. and a necessary condition for x∗ to be optimal for (3. the following exercise expresses if this matrix is negative deﬁnite then x ˆ f f0uu (u∗ ) in terms of derivatives of the functions fi . REMARKS AND EXTENSIONS 21 3. are linearly independent. all ˆ . ˆ Since we can convert the problem (3.3. deﬁne the Lagrangian function L : Rn+m → R by L : (x. However. λ∗ ) = 0. are not linearly independent.4 will apply to the function f0 remarks in terms of the original function f0 and f ..16) in a neighborhood of x∗ .2 An alternative condition. Furthermore. 3. 0 ≤ i ≤ m. This can be checked in the following example Minimize subject to sin(x2 + x2 ) 1 2 π 2 2 2 (x1 + x2 ) = 1...12).1.3 Remarks and Extensions 3. This is possible because the function g is uniquely speciﬁed by (3. 1 ≤ i ≤ m. (3. i. Let x∗ be optimal for (3. λ∗ ) is a stationary point of L. It follows that if the functions fi . i .1 The condition of linear independence. so is g (see Fleming [1965]).3. .12) into a problem of maximizing f0 over an open set. and suppose that fix (x∗ ). The following is a reformulation of 3.3 Secondorder conditions. Lx (x∗ .I] where m Lww Lwu Luw Luu gu .3.12. The necessary condition (3. are twice continuously ˆ differentiable.e. u∗ ) gu (u∗ ) = −[fw (x∗ )]−1 fu (x∗ ).12) and (3.29) 3. λ∗ ) = 0 and Lλ (x∗ . it is useful to translate these the comments of Section 2. Keeping the notation of Theorem 3. Furthermore. L(x) = f0 (x) − i=1 λ∗ fi (x). ∗ is a local optimum. 1 ≤ i ≤ m. ˆ f0uu (u∗ ) = [gu .13) need not hold if the derivatives fix (x∗ ).3. and its proof is left as i=1 an exercise. I (w∗ .3. λ) → f0 (x) − m λi fi (x). then so is f0 . if f is twice differentiable. Then there exists λ∗ ∈ Rm such that (x∗ .13) and ˆ the condition that the (n − m) × (n − m) matrix f0uu (u∗ ) is negative semideﬁnite. Exercise: Show that .
vb ) and j = (j1 . (3. 1 ≤ k ≤ b. 1 ≤ i ≤ m. Hence. Find x0 arbitrary so that fi (x0 ) = αi . .31) where vrk is the voltage across the resistor. 2 (3. Step 1.6. .31) is replaced by (3.3.22 CHAPTER 3. u) so that fw (xk ) is nonsingular.32) Although (3. 1 ≤ i ≤ m.30) implies that the current (jk −j s k) through the kth resistor depends only on the voltage vrk = (vk −v sk ) across itself.5 Design of resistive networks. ˆk Step 3. Then the Kirchhoff current and voltage laws respectively yield the equations Aj = 0 and A e = v (3. If f0 (uk ) = 0. Set uk = uk + dk f0 (uk ). . Orient the network graph and let v = (v1 .31) we shall assume that gk is a function of vr . . no essential simpliﬁcation is achieved. . p). . . Consider a network N with n + 1 nodes and b branches. and f0 (uk ) = −fu (xk )λk + f0u (xk ).4.31): j − js = g(v − vs ) = g(vr ). . stop. x ˜ ˜ Remarks.32): j − jx = g(vr . Set ˜ ˜ ˜ ˜ k+1 = (wk . uk ). As before. . 3. . and g = (g1 .30) Next we suppose that each branch k contains a (possibly nonlinear)resistive element with the form shown in Figure 3. vs ∈ Rb for the sources. uk ) = α. thus enabling us to calculate λk . Using the obvious vector notation js ∈ Rb . OPTIMIZATION WITH EQUALITY CONSTRAINTS 3. Then the following algorithm is a straightforward adaptation of the procedure in Section 2.3. . 1 ≤ i ≤ m. and gk is the characteristic of the resistor. . p ) which are under our control. Step 2. Set k = 0 and go to Step 2. let us suppose that there are design parameters p = (p1 . denote the vectors of branch voltages and branch currents. Find a partition x = (w. . Find wk such that fi (wk .4 A numerical procedure. The practical applicability of the algorithm depends upon two crucial factors: the ease with which we can ﬁnd a partition x = (w. we can rewrite (3. the step sizes dk > 0 can be selected various ways. vr ∈ Rb for the resistor voltages. jb ) respectively. . In the next section we apply this algorithm to a practical ˜ ˜ ˜ problem where these two steps can be carried out without too much difﬁculty. . .3. We assume that the derivatives fix (x). so that jk − jsk = gk (vrk ) = gk (vk − vsk ). in (3.33) This is just a notational convenience. p) = g(v−v s . uk ) = 0. . and the ease with which we can ﬁnd wk so that f (wk . vsk are the source current and voltage in the kth branch. . This allows us to include coupled resistors and voltagecontrolled current sources. u)2 of the variables such that fw (xk ) is nonsingular. We choose one of the nodes as datum and denote by e = (e1 . Furthermore. Otherwise go to Step 3. Let A be the n × b reduced incidence matrix of the network graph. and return to Step 2. . (3. en ) the vector of nodetodatum voltages. Here jsk . so that (3. gb ) . set k = k + 1.30) as (3. The w variable may consist of any m components of x. . are linearly independent for all x. Calculate λk −1 ˆk ˆk by (λk ) = f0w fw(xk). .
3: The kth branch. p) − is = 0. To do this we make the following assumption. is the transpose of the incremental branch admittance matrix (evaluated at (˜r . vs . vs . p. is ) satisfying (3. p) = (∂g/∂vr )(˜r . If we combine (3. Then the corresponding λ = λ ˜ λ = f0w (˜)fw (˜) = f0e (˜)fe (˜). If we compare (3. is ) there is a unique e = E(p. The network design problem can then be stated as ﬁnding p. is ) = Ag(A e−v s . is ) subject to Ag(A e − vs . Formally.33) ˜ with (3. In terms of the notation of 3. Furthermore.3. ˜s ) be a ˜ e ˜ ˜ i ˜ is given by (see (3. is ). G (˜r . For this reason. To this end let x = (˜. N (˜r .32) we obtain (3. vs . moreover. p).33): Ag(A e − vs . p)A is nonsingular for all v ∈ Rb . (c) The network N described by (3.36) we see immediately that λ is the nodetodatum response voltages of a linear network N (˜r . vs . Assumption: (a) f0 is differentiable.34) where we have deﬁned is = Ajs .36) has the following extremely interesting physical interpretation. x −1 x x −1 x From the deﬁnition of f we have fe (˜) = AG(˜r .36) Now (3. v ˜ ˜ x (3. this network has the same graph as v ˜ x the original network (since they have the same incidence matrix).3. for every value of (p. is ). x v ˜ ˜ where vr = A e − vs . p) is called the adjoint network (of N ) at (˜r . vs . is so as to minimize some speciﬁed function f0 (e.34): Minimize f0 (e. p. vs . Now the crucial part in the algorithm is to obtain λk at some point xk . vs . p).e. p) = is .4. p).4 to this problem. is ). vs .37) (3. (b) g is differentiable and the n×n matrix A(∂g/∂v)(v.29) and (3. Therefore. vs . (3. p ∈ R .3.27)) ﬁxed point. (3..33) is determinate i. Also let f (x) = f (e. REMARKS AND EXTENSIONS jsk jk − jsk jk o 23 + + vsk  vrk +  o vk  Figure 3. p. p. then assumption (b) allows us to identify w = e. λ is the solution (unique by ˜ ˜ ˜ v ˜ v ˜ assumption (b)) of the following linear equation: AG (˜r . p)A λ = f0e (˜). and u = (p.33).35) We shall apply the algorithm 3. p)) v ˜ v ˜ of the original network N . p) driven by the current sources f0e (˜). p) − is . p)A . if we let x = (e. p. we have the optimization problem (3. its branch admittance matrix. and G(˜r . v ˜ v ˜ .3.
33) to obtain ek+1 = (Epk+1 . pk ) driven by the current source f (xk ). Their derivation of the adjoint network does not appear as transparent as the one given here. one can use a more general cutset matrix. Similarly. Suppose that is is ﬁxed. This latter step may be very complex. In every case the “adjoint” network arises from a network interpretation of (3. which is equivalent to . and vsj for j ∈ J are ﬁxed. Solve (3. In practice we can control only some of the components of vs and is . OPTIMIZATION WITH EQUALITY CONSTRAINTS u u f0 (˜) ˆ using (3. j ∈ J.27). Elementary calculations yield ˜ Once we have obtained λ we can obtain (3. If we choose nonnegative numbers αt . calculate f0e (xk ).24 CHAPTER 3. (3. Remark 2. Calculate vr = A ek − vs . and one nonlinear network analysis step (the computation of ek+1 in step 3). step size. more general representations of the resistive elements may be employed. [fw (˜)] λ = f0w (˜). The interpretation of λ as the response of the adjoint network has been exploited for particular function f0 in a series of papers (director and Rohrer [1969a]. Calculate ˆ k the adjoint network N (vr u f0 (u ) from 0e (3. vs . Calculate the nodetodatum response λk of k .33) to obtain e0 = E(p0 . vsj for j ∈ J are variable. Set k = k + 1 and return to Step 2. [1969b]. x ˜ x with the transpose of the matrix giving rise to the adjective “adjoint.28). ik+1 ) = uk − dk u f0 (uk ). where as vsj = vsj − dk (∂ f0 /∂vsj )(uk ) and ˆ ik+1 = ik − dk (∂ f0 /∂ism )(uk ) with j and m ranging only over the controllable components and sm sm the rest of the components equal to their speciﬁed values.38) 0 0 Step 1. k k Step 2. stop. Remark 3. Let uk+1 = (pk+1 . p)] A ˆ ˜ ˆ u G (˜r . v k+1 . ik+1 ). [1969c]).] Let N be a transistor circuit. 3 u ˆ f0 (uk ). Let k = 0 and s s go to Step 2. t ∈ T . and let (3. with relative magnitudes reﬂecting the importance of the different transistors then we can formulate the criterion Note the minus sign in the expression uk − dk maximizing (−f0 ). vs . Some of the components vt .” Exercise: [DC biasing of transistor circuits (see Dowell and Rohrer [1971]). Otherwise go to Step 3. Each iteration from uk to uk+1 requires one linear network analysis step (the computation of λk in Step 2). s s Remark 1.33) model the dc behavior of this circuit. vs . and we wish to choose vsj . i0 ). Remember we are minimizing f0 . the rest being ﬁxed. where dk > 0 is a predetermined s 3 Solve (3. Although we have used the incidence matrix A to obtain our network equation (3. will correspond to bias voltages for the transistors in the network. i0 ) arbitrary. so that vt is as d close as possible to a desired bias voltage vt . p)A λ + f0v (˜) x u v ˜ u f0 (˜) = f0vs (˜) = s ˆ u f0is (˜) x −I f0is (˜) We can now state the algorithm. j ∈ J. k+1 ˆ Step 3. Select u0 = (p0 .37). t ∈ T . For each choice of vsj . If this gradient is zero. we obtain the vector e and / hence the branch voltage vector v = A e. The only change this requires in the algorithm is that in Step 3 we set k+1 k ˆ ˆ pk+1 = pk − dk f0p (uk ) just as before.33).37): ∂g ˆ u f0p (˜) x f0p (˜) v ˜ [ ∂p (˜r .
3. REMARKS AND EXTENSIONS f0 (e) = t∈T 25 αt vt −v d 2 . t (i) Specialize the algorithm above for this particular case.3. (ii) How do the formulas change if the network equations are written using an arbitrary cutset matrix instead of the incidence matrix? .
OPTIMIZATION WITH EQUALITY CONSTRAINTS .26 CHAPTER 3.
Then the number of seminars demanded per year is 2g+u . For a detailed and readily accessible treatment of the material presented in this chapter see the companion volume in this Series (Sakarovitch [1971]). 27 .1 The Linear Programming Problem 4.1. 000 and 5g+7u 40 ≤ 5250 or 5g + 7u ≤ 210. the faculty can 40 offer 2(750) + 3(250) = 2250 seminars and 6(750) + 3(250) = 5250 lecture courses. 000 . Let g and u respectively be the number of graduate and undergraduate students admitted. 4. In section 4 we apply the results of Sections 2 and 3 to study the linear programming theory of competitive economy. In Section 3 we present the Simplex algorithm which is the main procedure used to solve linear programming problems. Because of his contractual agreements. On the supply side of our accounting. and the number of 20 lecture courses demanded per year is 5g+7u . Recall Example 2 of Chapter I. Additional miscellaneous comments are collected in the last section.Chapter 4 OPTIMIZATION OVER SETS DEFINED BY INEQUALITY CONSTRAINTS: LINEAR PROGRAMMING In the ﬁrst section we study in detail Example 2 of Chapter I. the President must satisfy 2g+u 20 ≤ 2250 or 2g + u ≤ 45. In the second section we present the duality theory for linear programming and use it to obtain some sensitivity results.1 Example. and then we deﬁne the general linear programming problem.
and futhermore at x∗ the direction c points away from Ω. the President receives the payoff c x. Formally then the President faces the following decision problem: Maximize αg + βu subject to 2g + u ≤ 45.2. and (ii) x∗ yields a higher payoff than all points in the cone K ∗ consisting of all rays starting at x∗ and passing through Ω.2) Let Ai . since K ∗ lies “below” π ∗ .2) is given by Ω = {xAi x ≤ bi .2)1 Maximize c x subject to Ax ≤ b . there are also the constraints g ≥ 0. so that K ∗ is given by K ∗ = {x∗ + hA1 h ≤ 0 .1) can be rewritten as (4. Now x∗ satisﬁes Ax x∗ = b1 . Then the set Ω of all vectors x which satisfy the constraints in (4.1) It is convenient to use a more general notation. the surface of constant payoff k say. A2 x∗ = b2 . The ﬁrst conclusion is the foundation of the powerful Simplex algorithm which we present in Section 3.) Evidently an optimal decision is any point x∗ ∈ Ω which lies on a hyperplane π(k) which is farthest along the direction c. Therefore. A4 x∗ < b4 .3) as an exercise.1. (4.2. β) . For the situation depicted in Figure 4. Here we pursue consequences of the second conclusion. LINEAR PROGRAMMING Since negative g or u is meaningless. b = (45000. denote the rows of A. A2 h ≤ 0 . is the hyperplane π(k) = {xc x = k}. (4. A2 h ≤ 0} . A= −1 0 0 −1 Then (4. so that x ≤ y means xi ≤ yi for all i. For each choice x. (Obviously we are assuming in this discussion that c = 0. 0) and let A be the 4×2 matrix 2 1 5 7 . 1 (4. Furthermore. 1 ≤ i ≤ 4. We can rephrase this by saying that x∗ ∈ Ω is an optimal decision if and only if the plane π ∗ through x∗ does not intersect the interior of Ω.1. From this condition we can immediately draw two very important conclusions: (i) at least one of the vertices of Ω is an optimal decision. So let x = (g. Since c x∗ ≥ c y for all y ∈ K ∗ we conclude that c h ≤ 0 for all h such that A1 h ≤ 0. as k increases π(k) moves in the direction c. and A3 x∗ < b3 . 1 ≤ i ≤ 4} and is the polygon OP QR in Figure 4. 0. These hyperplanes for different values of k are parallel to one another since they have the same normal c. 210000. .28 CHAPTER 4. 000 5g + 7u ≤ 210. 000 g ≥ 0. We pause to formulate the generalization of (4. c = (α. u) .3) Recall the notation introduced in 1. u ≥ 0 . u ≥ 0.1 we can see that x∗ = Q is an optimal decision and the cone K ∗ is shown in Figure 4.
i ∈ I(x∗ ).1. be ndimensional row vectors. For any x satisfying the constraints. the optimal decision will change. Suppose x∗ satisﬁes the constraints. 1 ≤ i ≤ k. .3) is satisﬁed as long as c lies between A1 and A2 . let I(x) ⊂ {1. its generalization to n dimensions is a deep theorem known as Farkas’ lemma (see Section 2).1): Although this statement is intuitively obvious. We can see from our analysis that the situation is as follows (see Figure 4. Ai x < bi . i ∈ I(x). Show that x∗ is optimal if an only if / c h ≤ 0 for all h such that Ai h ≤ 0 . Mathematically this means that (4. . Let c ∈ Rn .1: Ω = OP QR.4. 1 ≤ i ≤ k . Consider the problem Maximize c x subject to Ai x ≤ bi . Exercise 1: Let Ai . λ∗ ≥ 0 such that 1 2 c = λ∗ . Returning to our problem. i ∈ I(x).4)   {xA2 x = b2 }  x1 2 As c varies. be real numbers. THE LINEAR PROGRAMMING PROBLEM x2 29  π∗  π(k) = {xc x = k} direction of increasing payoff k  P  Q = x∗  c ⊥ π∗ A2 ⊥ P Q A1 ⊥ QR A3 . n} be such that Ai (x) = bi . O A4 R {xA1 x = b1 } Figure 4. and let bi . 2 . it is clear that (4. . A1 + λ∗ A2 . . 1 2 (4.3) is satisﬁed if and only if there exist λ∗ ≥ 0. 1 ≤ i ≤ k.
(b) if Ai x∗ < bi then λ∗ = 0 . 2.6) . 2. x∗ = Q is optimal iff c lies between A1 and A2 iff c = λ∗ A1 + λ∗ A2 for some λ∗ ≥ 0. λ∗ ≥ 2 3 2 3 0. Exercise 2: Show that (4. 1 ≤ i ≤ 4. 2. 2 2 3. λ∗ ≥ 1 2 1 2 0. such that i 4 (a) c = i=1 λ∗ ai . ai2 ). (Here Ai = (ai1 . These statements can be made in a more elegant way as follows: x∗ ∈ Ω is optimal iff there exists λ∗ ≥ 0 . i i (4. 2 1i i (4. 2.6). etc. below.5) For purposes of application it is useful to separate those constraints which are of the form xi ≥ 0. 1.5) is equivalent to (4.2: K ∗ is the cone generated by Ω at x∗ . 1 2 (b) if aj1 x∗ + aj2 x∗ < bj then x∗ = 0. j = 1. x∗ ∈ QP is optimal iff c lies along A2 iff c = λ∗ A2 for some λ∗ ≥ 0. x∗ = P is optimal iff c lies between A3 and A2 iff c = λ∗ A2 + λ∗ A3 for some λ∗ ≥ 0. and to reformulate (4. from the rest.5) accordingly We leave this as an exercise. i = 1. i = 1.30 CHAPTER 4. λ∗ ≥ 0 such that 1 2 (a) ci ≤ λ∗ a1i + λ∗ a2i .) x∗ ∈ Ω is optimal iff there exist λ∗ ≥ 0 . 1 2 j (c) if ci < λ∗ + λ∗ a2i then x∗ = 0. LINEAR PROGRAMMING P x∗ = Q K∗ A2 c A1 A3 O R π∗ A4 Figure 4.
9) xj ≥ 0 Although (4. where the cj . j=1 subject to 1≤i≤m.8) and (4. . . + ain xn ≤ bi . aij . A linear programming problem (or LP in brief) is any decision problem of the form 4. 1≤j≤p.7) can be transformed into an equivalent LP of the form (4. Step 3: Replace each variable xj which is constrained xj ≤ 0 by a variable yj = −xj constrained yj ≥ 0 and then replace aij xj by (−aij )yj for every i and cj xj by (−cj )yj . p + 1 ≤ j ≤ q. . .2 Problem formulation.8): n (4. . . . THE LINEAR PROGRAMMING PROBLEM 31 4. xj arbitary .7. + 1 ≤ i ≤ m . . . (4.7) is of the form (4. .8). j=1 subject to 1≤i≤m. xj ≥ 0 . (−aij )xj ≤ (−bi ). + cn xn subject to ail x1 + ai2 x2 + . q + 1 ≤ j ≤ n . Maximize c1 x1 + c2 x2 + . .1. Step 1: Replace each inequality constraint aij xj ≥ bi by (−aij )xj ≤ (−bi ). . k + 1 ≤ i ≤ . . and xj ≥ 0 . Proof.4.7) appears to be more general than (4. Step 2: Replace each equality constraint aij xj = bi by two inequality constraints: aij xj ≤ bi .7) is of the form (4. . 1≤j≤n. bi are ﬁxed real numbers. . ail x1 + . . .1. such is not the case. . There are two important special cases: Case I: (4.7) Maximize j=1 n cj xj aij xj ≤ bi . + ain xn ≥ bi . .9). . Proposition: Every LP of the form (4.8) xj ≥ 0 Case II: (4. 1≤j≤n (4.9): n Maximize j=1 n cj xj aij xj = bi . . . + ain xn = bi . l ≤ i ≤ k . ail x1 + . .
1). λk ≥ 0 such that c = i=1 λi Ai .) Step 3. Step 4: Repeat these steps from the previous proposition. Ax ≤ 0 implies c x ≤ 0. Exercise 1: With the same hypothesis and notation of Exercise 1 in 4. The following statements are equivalent. Consider the pair of LPs (4. Step 2: Replace each inequality constraint aij xj ≥ bi by the equality constraint aij xj − yi = bi where yi is an additional variable constrained by yi ≥ 0.9) Proof. (i) for all x ∈ Rn .2. whereas x ∈ Rn and λ ∈ Rm will be variable. ♦ Proposition: Every LP of the form (4.11) . (The new variables added in these steps are called slack variables. i i Exercise 2: Let x∗ satisfy the constraints for problem (4. Let c ∈ Rn be a column vector. b ∈n are ﬁxed vectors.1. and A = {aij } is a ﬁxed m × n matrix. Let Ai . Farkas’ Lemma (algebraic version). . λ ∈ Rk . Use the previous exercise to show that x∗ is optimal iff there exist λ∗ ≥ 0. .17).9) and is equivalent to the original one. Let A be a k × n matrix. We leave this development as two exercises and follow a more elegant but less intuitive approach. . i j In the remaining discussion. be ndimensional row vectors. . Farkas’ Lemma. 1 ≤ j ≤ m. 1 ≤ j ≤ n i m i=1 (b) if j=1 aij x∗ < bi then λ∗ = 0 .10) and (4. zj ≥ 0 and then replace aij xj by aij yj + (−aij )zj for every i and cj xj by cj yj + (−cj )zj . (ii) there exists λ ≥ 0.32 CHAPTER 4. The following statements are equivalent: (i) for all x ∈ Rn . Evidently the new LP has the form (4. use the ﬁrst version of Farkas lemma to show that there exist λ∗ ≥ 0 for i ∈ I(x∗ ) such that λ∗ Ai = c . λ∗ ≥ 0 such that m 1 m i∈I(x∗ ) (a) cj ≤ n i=1 λ∗ aij . .7) can be transformed into an equivalent LP of the from (4. such that A λ = c. Evidently the resulting LP has the form (4.2 Qualitative Theory of Linear Programming 4. LINEAR PROGRAMMING Step 4: Replace each variable xj which is not constrained in sign by a pair of variables yj −z j = xj constrained yj ≥ 0. 1 ≤ i ≤ k. Using this result it is possible to derive the main results following the intuitive reasoning of (4. . k (ii) there exists λ1 ≥ 0. c ∈ Rn . Step 1: Replace each inequality constraint aij xj ≤ bi by the equality constraint aij xj + yi = bi where yi is an additional variable constrained yi ≥ 0. Let c ∈ Rn . Ai x ≤ 0 for 1 ≤ i ≤ k implies c x ≤ 0. . We begin by quoting a fundamental result.8) and is equivalent to the original one. ♦ 4. . 1 ≤ i ≤ m (c) if j i λ∗ aij > cj then x∗ = 0 . For a proof the reader is referred to (Mangasarian [1969]). An algebraic version of this result is sometimes more convenient.1 Main results.
x ≥ 0} be the set of all points satisfying the constraints of the primal problem. Then there exists x∗ which is optimum for (4. c x∗ = (λ∗ ) b. Lemma 1: (Weak duality) Let x ∈ Ωp . this is equivalent to the existence of x ≥ 0. . (4.11) Deﬁnition: Let Ωp = {x ∈ Rn Ax ≤ b. (4. Furthermore. Theorem 1: (Strong duality) Suppose Ωp = φ and Ωd = φ. −w ≤ 0 . λ ≥ 0}. Maximize c1 x1 + . The next result is trivial. i. + λm amj ≥ cj .10) Maximize λ1 b1 + . this is possible only if A ξ − cθ ≤ 0 . QUALITATIVE THEORY OF LINEAR PROGRAMMING below. Then c x ≤ λ Ax ≤ λ b.11). + ain xn ≤ bi . + cn xn subject to ai1 x1 + . . Similarly let Ωd = {λ ∈ Rm λ A ≥ c .12) Proof: x ≥ 0 and λ A − c ≥ 0 implies (λ A−c )x ≥ 0 giving the ﬁrst inequality.10) and λ∗ which is optimum for (4. r ∈ R. A point x ∈ Ωp (λ ∈ Ωd ) is said to be a feasible solution or feasible decision for the primal (dual). (4. .13) . 1 ≤ i ≤ m xj ≥ 0 .10) and λ∗ is optimal for (4. ξ ≤ 0 .11). then x∗ is optimal for (4. . Proof: Because of the Corollary 1 it is enough to prove the last statement. θ≤0 implies b ξ + c w ≤ 0. + λm bm subject to λ1 a1j + . ♦ Corollary 1: If x∗ ∈ Ω and λ∗ ∈ Ωd such that c x∗ = (λ∗ ) b. 1 ≤ j ≤ n λi ≥ 0 . we must show that there exist x ≥ 0. (4. . . such that Ax ≤ b. y ≥ 0.. A λ ≥ c and b λ−c x ≤ 0. b−Ax ≥ 0 and λ ≥ 0 implies λ (b−Ax) ≥ 0 giving the second inequality.10) is called the primal problem and (4. r ≤ 0 such that A −c Im A b −I n 1 x y λ µ r b = c 0 By the algebraic version of Farkas’ Lemma. . 33 (4. Aw = bθ ≤ 0 .e.14) (4. µ ∈ Rm . 1 ≤ j ≤ n . λ ≥ 0. . 1 ≤ i ≤ m . λ ∈ Ωd . µ ≤ 0. λ ≥ 0.4.2. By introducing slack variables y ∈ Rm .11) is called the dual problem.
−b ξ = b (−ξ) ≥ (Ax) (−ξ) = x (−A ξ) ≥ 0.15) λ∗ aij < cj implies x∗ = 0 . Ωp = φ. The sufﬁciency part of (i) follows from Theorem 1. Suppose (4. so that only the necessity remains. so that by Lemma 1 c w/(−θ) ≤ b ξ/θ. which is equivalent to (4. µ ≤ 0 such that λ  A −In − − − = c  µ By Farkas’ Lemma there exists w ∈ Rn such that Aw ≤ 0. (4. w ≥ 0. Proof Because of the symmetry of the primal and dual it is enough to prove only (i).15) is known as the condition of complementary slackness. Theorem 2: (i) Suppose Ωp = φ. Theorem 3: (Optimality condition) x∗ ∈ Ωp is optimal if and only if there exists λ∗ ∈ Ωd such that m j=1 aij x∗ < bi implies λ∗ = 0 . Ωd = φ. Suppose. that Ωd = φ. and c w > 0. i j and m i=1 ((4. θ) satisﬁes (4. Then there exists an optimum decision for the primal LP iff Ωd = φ.15) is equivalent to (4. ♦ . λ ∈ Ωd . The ﬁrst equality in (4. there exist x ∈ Ωp . the hypothesis that Ωp = φ is essential.16) yields (λ∗ ) b = (λ∗ ) Ax∗ = (A λ∗ ) x∗ . A(x + θw) ≤ b. Sufﬁciency.) Proof: First of all we note that for x∗ ∈ Ωp . ♦ Remark: In Theorem 2(i). and (A λ∗ − c) x∗ = 0 . Suppose x∗ ∈ Ωp is optimal. ξ. so that by Theorem 1 there exists λ∗ ∈ Ωd such that c x∗ = (λ∗ ) b. Then there exists an optimum decision for the dual LP iff Ωp = φ. −ξ ≥ 0. We will show that sup {c xx ∈ Ωp } = +∞. By Lemma 1 we always have c x∗ ≤ (λ∗ ) Ax∗ ≤ (λ∗ ) b so that we must have c x∗ = (λ∗ ) Ax∗ = (λ∗ ) b.14) since θ < 0. there does not exist λ ≥ 0. but then for any θ > 0. By Corollary 1. Hence. so that (x + θw) ∈ Ωp . (w/−θ) ∈ Ωp . so that −A ξ ≥ 0. in contradiction.16) Necessity. ♦ The existence part of the above result can be strengthened. (x + θw) ≥ 0. Now. Then (ξ/θ) ∈ Ωd . Also. Equivalently. (4. x∗ is optimal. so there exists x ≥ 0 such that Ax ≤ b. Aw ≤ 0. λ∗ ∈ Ωd . But (4.16): (λ∗ ) (Ax∗ − b) = 0.16) holds for some x∗ ∈ Ωp . and c w ≤ (A λ) w = λ (Aw) ≤ 0. −w ≤ 0. λ∗ ∈ Ωd . sup {c xx ∈ Ωp } = +∞ so that there is no optimal decision for the primal. Then from Theorem 2. By hypothesis. By hypothesis. Ωd = φ means there does not exist λ ≥ 0 such that A λ ≥ c.13) and θ = 0. So that b ξ + c w ≤ 0. Evidently then.34 CHAPTER 4.16) is just an equivalent rearrangement of these two equalities. Case (ii): Suppose (w. θ) satisﬁes (4. ξ. Exercise 3: Exhibit a pair of primal and dual problems such that neither has a feasible solution. j i (4. Consider the following exercise. (ii) Suppose Ωd = φ. c (x + θw) = c x + θc w. so that c x∗ = (λ∗ ) b. while the second yields (A λ∗ ) x∗ = c x∗ . LINEAR PROGRAMMING Case (i): Suppose (w.13) and θ < 0.
We begin with a pair of LPs: Maximize c1 x1 + . (−A)x ≤ (−b). whose proof is left as an exercise.2. (4. x ≥ 0} × {λλ ∈ Rm .15) as in the following result. + cn xn subject to ail x1 + .2. λ ∈ Rm . + λm amj ≥ cj (4. . (Note that. .19) .19) is called the primal and (4. We state these results as exercises. λ∗ ≥ 0 provided we strengthen (4.19) by the equivalent LP: maximize c x. The function L is called the Lagrangian.7). . 1 ≤ i ≤ m . subject to Ax ≤ b. . . and obtain the result analogous to Exercise 5. QUALITATIVE THEORY OF LINEAR PROGRAMMING 35 The conditions x∗ ∈ Ωp .) Exercise 6: Show that x∗ ∈ Ωp is optimal iff there exists λ∗ ∈ Ωd such that m x∗ j > 0 implies i=1 λ∗ aij = cj . + λm bm subject to λ1 a1j + . + ain xn = bi . Theorem 4: (Saddle point) x∗ ≥ 0 is optimal for the primal if and only if there exists λ∗ ≥ 0 such that L(x. λ∗ ) ≤ L(x∗ . λ ≥ 0}.20) respectively. λ∗ ) ≤ L(x∗ .17) is said to form a saddlepoint of L over the set {xx ∈ Rn . x ≥ 0. λ is not restricted in sign. xj ≥ 0 . Ωd denote the set of all x. Exercise 5: Prove Theorems 1 and 2 with Ωp and Ωd interpreted as above. (Hint. x∗ ∈ Ωd in Theorem 3 can be replaced by the weaker x∗ ≥ 0. indicating how to use the results already obtained. λ∗ ) ≤ L(x∗ . .9). where L: Rn xRm → R is deﬁned by L(x. .2 Results for problem (4.18).20) the dual.9). (4. λ) for all x ≥ 0. λ) = c x − λ (Ax − b) (4.19). This is now of the form (4.) Exercise 8: Formulate a dual for (4. λ∗ ) satisfying (4. Again (4. i Exercise 7: x∗ ≥ 0 is optimal iff there exists λ∗ ∈ Rm such that L(x.17) Exercise 4: Prove Theorem 4. A pair (x∗ . Apply Theorems 1 and 2. 4.20) the λi are unrestricted in sign.20) Note that in (4. where L is deﬁned in (4. λ∗ ) ≤ L(x∗ . and allλ ≥ 0. unlike (4. . 1≤j≤n . λ) for all x ≥ 0. We let Ωp . Remark. 1 ≤ j ≤ n .10).18) (4. Replace (4. It is possible to derive analogous results for LPs of the form (4. Minimize λ1 b1 + .4.17). λ satisfying the constraints of (4. .
22) .36 CHAPTER 4. ˆ as follows: ∂M + ˆ ˆ ∂bi (b. ˆ Let B . ε). c2 . b. b2 . ˆ ∂M + ˆ ˆ ∂cj (b. c) ∈ B × C deﬁne M (b. The matrix A will remain ﬁxed. LINEAR PROGRAMMING 4. ˆ ∂M − ˆ ˆ ∂cj (b.20). .19) and (4. if b. cj+1 . . Let Ωp and Ωd be the sets of feasible solutions for the pair (4. Let B = {b ∈ Rm Ωp (b) = φ} and C = {c ∈ Rn Ωd (c) = φ}. . . c) .21) = lim ε→0 ε>0 1 ˆ ˆ ε {M (b(i. C respectively. c) = max {c xx ∈ Ωp (b)} = min {λ bλ ∈ Ωd (c)} . then ˆ b). b. c(j. . c)} . . We write Ωp (b) and Ωd (c) to denote the explicit dependence on b and c respectively. c − M (ˆ c(j. . bi−1 .10) and (4.10) or (4. For 1 ≤ i ≤ m. 1≤i≤m. c) (4. ◦ ◦ Theorem 5: At each (ˆ c) ∈B × C . ε ∈ R. b. ε ∈ R. ˆ x ∈ Ωp (ˆ λ ∈ Ωd (ˆ) are optimal.19) changes as the vectors b and c change. . b ∈ Rm denote b(i. ˆ ∂M − ˆ ˆ ∂bi (b. . . c ∈ Rn denote c(j. c) ◦ ◦ ˆ ≤ λi ≤ ∂M − ˆ ˆ ∂bi (b. Furthermore. (4. ε) = (c1 . . bi + ε. C denote the interiors of B. bm ) . cj + ε.11) or for the pair (4.3 Sensitivity analysis. and for (b. and for 1 ≤ j ≤ n. . b(i. ε) = (b1 . cn ) . ˆ c ∂M + ˆ ˆ ∂bi (b. −ε))} . . bi+1 . cj−1 .2. the partial derivatives given above exist. . c ) − M (ˆ c)} . . . ε)) − M (ˆ c} . c) = lim ε→0 ε>0 1 ˆ ˆ ε {M (b. c) − M (ˆ −ε). c) = lim ε→0 ε>0 1 ˆ ˆ ε {M (b. We deﬁne in the usual way the right and left hand partial derivatives of M at a point (ˆ c) ∈ B × C b. c) = lim ε→0 ε>0 1 ˆ ˆ ε {M (b. We investigate how the maximum value of (4.
(b) Show that for ﬁxed c ∈ C. ˆ ˆˆ b.22). (i) f is said to be convex if X is convex. −ε))} ≤ 1 {ˆ − c(j. Deﬁnition: X ⊂ Rn is said to be convex if x. ε) − b}λi . 1 ≤ i ≤ m xj ≥ 0 . c ) − M (b.4. . M (ˆ c) = c x. c) and M (b. 0 ≤ θ ≤ 1 implies f (θx + (1 − θ)y) ≥ θf (x) + (1 − θ)f (y). ˆ b(i. and M (ˆ c(j. and f : X → R be convex. ·) : C → R is convex. {M (b. ˆ c ˆ 1 1 ˆ ˆ ˆ ˆ c ˆ ˆ ˆ ε {M (b. −ε)} x = xj . 0 ≤ θ ≤ 1 implies f (θx + (1 − θ)y) ≤ θf (x) + (1 − θ)f (y). ♦ We recall some fundamental deﬁnitions from convex analysis. M (·. C ⊂ Rn deﬁned above are convex sets. and x. ˆ b(i. Remark 3: The variables of the dual problem are called Lagrange variables or dual variables or shadowprices. This is useful in some computational problems. then we have equality in (4. . ε > 0. M (b.23) assuming that the partial derivatives exist. By strong duality ˆ b.3 The Simplex Algorithm 4. i.e. 9 below. linear plus constant) functions on B and C respectively. and the sets B ⊂ Rm . ε1 )) − f (ˆ))} ≥ x x x x (1/δ1 ){f (ˆ(i. ˆ ≤ ≥ 1ˆ ˆ ˆ ˆ ε λ {b(i. (ii) f is said to be concave if −f is convex. ε)) x. Ωd . δ2 )) − f (ˆ)}. x. gives (4.3. + cn xn subject to ail x1 + .14). ε)) ≥ (ˆ(j. for ε > 0. so that b. c) ≤ λ ˆ ε). On the other hand. ε2 )) − f (ˆ)} ≥ (1/ε1 ){f (ˆ(i. c)} b(i. M (ˆ c) = λ ˆ and by weak duality M (ˆ ε). Finally.3. y ∈ X and 0 ≤ θ ≤ 1 implies (θx + (1 − θ)y) ∈ X. y ∈ X.22). 4. (4.24) . (Hint: First show that for ε2 > ε1 > 0 > δ1 > δ2 . (4. ˆ b. c) . Then the result follows immediately. ˆ 1 ˆ ˆ ˆ ˆ ε {M (b(i. THE SIMPLEX ALGORITHM 37 ∂M + ˆ ˆ ∂cj (b. + ain xn = bi .23) Proof: We ﬁrst show (4. Exercise 8: (a) Show that Ωp . ε). c(j. c) : B → R is concave and for ﬁxed b ∈ B. ·) are piecewise linear (more accurately. ε)) − M (b. > 0.(1/ε2 ){f (ˆ(i. c) ≥ xj ≥ ˆ ∂M − ˆ ˆ ∂cj (b.) x x x x Remark 1: Clearly if (∂M/∂bi )(ˆ exists. 1≤j≤n.23) as ε → 0. y ∈ X. c)} 1 ˆ c) − M (ˆ −ε). 1 ≤ j ≤ n . (4. . Deﬁnition: Let X ⊂ Rn and f : X → R. Taking limits as ε → 0. δ1 )) − f (ˆ))} ≥ (1/δ2 ){f (ˆ(i. for ε 1ˆ ˆ ˆ −ε)} = λi . The reason behind the last name will be clear in Section 4.22). the existence of the right and left partial derivatives follows from Exercises 8. Exercise 9: Let X ⊂ Rn . c)} ≥ ε {ˆ(j. ε) − c} x = xj . ε > 0. so that b.1 Preliminaries We now present the celebrated Simplex algorithm for ﬁnding an optimum solution to any LP of the form (4.. . for ε > 0. Remark 2: We can also show without difﬁculty that M (·. the left and right hand partial derivatives of f exist.24): Maximize c1 x1 + . ˆ ε λ {b − b(i. ˆ c ˆ ˆ ˆ ε ε which give (4. ˆ ε {M (b. for ε > 0. 1 ˆ c) − M (ˆ c(j. and then this result b) compares with 3. Show that at each point x in the interior of ˆ X.
z in Ωp and 0 < λ < 1. (n − j)! Lemma 2: Let x∗ be an optimal decision of (4. Hence z(θ) ∈ Ωp whenever θ ≤ θ ∗ . Lemma 1: Let x ∈ Ωp . and then c x∗ = c z(θ). such that γj Aj = 0 . .24).. Then there is a vertex z ∗ of Ωp which is optimal. implies x = y = z. with y. it follows that z(θ) ≥ 0 when j θ ≤ min x∗ j γj  j ∈ I(x∗ ) = θ ∗ say . Hence suppose {Aj j ∈ I(x∗ )} is linearly dependent so that there exist γj . amj ) . we only have to investigate a ﬁnite set. Since x∗ is optimal we must have c x∗ ≥ c z(θ) = c x∗ + θ j∈I(x∗ ) cj yj for −∗ θ ≤ θ ≤ θ ∗ .38 CHAPTER 4. Exercise 1: Prove Lemma 1. let I(x) = {jxj > 0}.e.1 the algorithm rests upon the observations that if an optimal exists. Aj = (a1j . Then x is a vertex of Ωp iff {Aj j ∈ I(x)} is a linearly independent set. In the following we let Aj denote the jth column of A. But from the deﬁnition of z(θ) it is easy to see that we can pick θ0 with θ0  = θ ∗ such that zj (θ0 ) = x∗ +θ0 γj = 0 j for at least one j = j0 in I(x∗ ). not all zero. let z ∗ = x∗ and we are done. Since x∗ > 0 for j ∈ I(x∗ ). . I(z(θ0 )) ⊂ I(x∗ ) − {j0 } . cj γj = J∈I(x∗ ) Since θ can take on positive and negative values. then at least one vertex of the feasible set Ωp is an optimal solution. LINEAR PROGRAMMING As mentioned in 4. the inequality above can hold on if 0. m n! Corollary 1: Ωp has at most vertices. i. j ∈ I(x∗ ) . j x∗ Aj + θ j j∈I(x∗ ) j∈I(x∗ ) Az(θ) = j∈I(x∗ ) zj (θ)Aj = γj Aj =b+θ·0=b. Deﬁnition: For x ∈ Ωp . j∈I(x∗ ) j=1 For θ ∈ R deﬁne z(θ) ∈ Rn by zj (θ) = x∗ = θγj . Proof: If {Aj j ∈ I(x∗ )} is linearly independent. Deﬁnition: x ∈ Ωp is said to be a vertex of Ωp if x = λy + (1 − λ)z. . The practicability of this investigation depends on the ease with which we can characterize the vertices of Ωp . j ∈ I(x∗ ) j x∗ = 0 . . so that z(θ) is also an optimal solution for θ ≤ θ ∗ . We begin with a precise deﬁnition of a vertex. Since Ωp has only ﬁnitely many vertices (see Corollary 1 below). This is done in Lemma 1. Then. .
2 The Simplex Algorithm.e. . and if not obtains another basic feasible solution with a higher value. . We shall discuss Phase II ﬁrst.2. . Proof: By Exercise 6 of Section 2. j2 . . .. (4. in a ﬁnite number of steps. Iterating on this procedure. . Step 2.3. ♦ At this point we abandon the geometric term “vertex” and how to established LP terminology. . for . are called the basic variables at z.25) (4. z is optimal iff there exists λ such that λ Aj = cj . Assumption of nondegeneracy. Every basic feasible solution is nondegenerate. jm ]. Evidently 0 < θ < ∞. λ Aj ≥ cj . 4. either we obtain an optimum solution or we discover that no optimum exists. because z k is optimal ˆ by Lemma 3. We will comment on it later. The set I(z) is then called the basis at z. Otherwise go to Step 4. j ∈ I(z). If γ k ≤ 0. . Let I(z k ) consist of j1 < j2 < . THE SIMPLEX ALGORITHM 39 Again. j j Step 3. Deﬁne z k+1 by . and let j1 < j2 < . . for . Otherwise pick any ˆ ∈ I(z k ) such that cˆ − λ (z k )Aj > 0 and go to Step 3. j ∈ I(z) . Lemma 3: Let z be a nondegenerate basic feasible solution. j ∈ I(z) are called the nonbasic variables at z. Phase II starts with a basic feasible solution and determines if it is optimal or not. < jm .26) holds iff λ = λ(z) and then (4. . (4. and {Aj j ∈ I(z)} is linearly independent. there is no ﬁnite optimum.c(z k ). and xj . xj . . We call λ(z) the shadowprice vector at z. Deﬁnition: (i) z is said to be a basic feasible solution if z ∈ Ωp . in a ﬁnite number of steps we will ﬁnd an optimal decision z ∗ which is also vertex.25). sup {c xx ∈ Ωp } = +∞. The algorithm is divided into two parts: In Phase I we determine if Ωp is empty or not. k k k Step 4.A constitute I(z). We make the following simplifying assumption. Compute θ = min {(zj γj )j ∈ i(z). If all these numbers are ≤ 0.27) is the same as (4. and if not. . cjm ) and deﬁne λ(z) by λ (z) = c (z)[D(z)]−1 .26) (4. Compute the vector ˆ k k γ k = (γj1 . Then z is optimal if and only if λ (z)A ≥ cj . Calculate [D(z k )]−1 . i. let c(z) denote the mdimensional column vector c(z) = (cj1 . j ∈ I(z) .3. . because by Lemma 4 below. Let z 0 be a basic feasible solution obtained from Phase I or by any other means. for all . if {Aj j ∈ I(z(θ0 ))} is linearly independent. Deﬁnition: A basic feasible solution z is said to be nondegenerate if I(z) has m elements. . stop. and the shadowprice vector λ (z k ) = c (z k )[D(z k )]−1 . . Let D(z) denote the m × m nonsingular matrix D(z) = [Aj1 . For each j ∈ I(z k ) calculate cj − λ (z k )Aj . Set k = 0 and go to Step 2. Clearly. stop.A . γj > 0}. . Otherwise we repeat the procedure above with z(θ0 ). < jm . Phase II: Step 1. Notation: Let z be a nondegenerate basic feasible solution. we obtain a basic feasible solution. .4. . then we let z ∗ = z(θ0 ) and we are done. γjm ) = [D(z k )]−1 Aj .27) ♦ But since z is nondegenerate. . j ∈ I(z) .
Set k = k + 1 and return to Step 2. so that c z(θ) → ∞ as θ → ∞. since γ k ≤ 0 it follows that z(θ) ≥ 0 for θ ≥ 0.30) = c z + θ{cˆ − λ j ˆ (z k )Aj }i . ♦ ˆ Corollary 2: In a ﬁnite number of steps Phase II will obtain an optimal solution or will determine that sup{c xx ∈ Ωp } = ∞. we must have γ˜ j from (4. Lemma 4: If γ k ≤ 0. j=ˆ j = k zj = 0 . z(θ) ∈ Ωp for θ ≥ 0.28) and (4. Finally if we compare (4. Next. j=ˆ and j ∈ I(z) . LINEAR PROGRAMMING k k zj − θγj . j which is positive from Step 2. so that in (4.28) By Lemma 5 below. zj = 0 j First of all. k+1 k k k Proof: Let ˜ ∈ I(z k ) be such that γ˜ > 0 and z˜ = θγ˜ .28) we see that z˜ = 0. j k+1 zj (4.40 CHAPTER 4. j =ˆ θ j . j ∈ I(z) θ . j = ˜ But if this is not the j}. ♦ ˆ But from step 2 {cˆ − λ (z k )Aj } > 0. Exercise 2: Prove Corollaries 2 and 3. Finally.29). sup {c xx ∈ Ωp } = ∞. Hence. giving a contradiction.30) that c zk+1 − c zk = θ{cˆ − γ (z k )Aj } .31) we must have equality. I(z k+1 ) has m elements. j ∈ I(z) zj (θ) = . j ∈ I(z) and j = ˆ . j Lemma 5: z k+1 is a basic feasible solution and c z k+1 > c z k .29) k γj Aj + θA = Az by deﬁnition of ˆ j γk. j} (4. Corollary 3: Suppose Phase II terminates at an optimal basic feasible solution z ∗ .24). Proof: Deﬁne z(θ) by k zj − θγj . k = 0. Then γ(z ∗ ) is an optimal solution of the dual of (4. We see then that D(z k+1 ) is obtained from D(z k ) by replacing the column Aj by . k k c z(θ) = c z − θc (z )γ + θcˆ j ˆ = c z + θ{cˆ − c (z k )[D(z k )]−1 Aj } j (4.31) ˜ so that it is enough to prove that Aj is independent of {Aj j ∈ I(z). j j j j j hence I(z k+1 ) ⊂ (I(z) − {˜ j}) {ˆ . Az(θ) = Az − θ j∈I(z) (4. z k+1 is a basic feasible solution with c z k+1 > c z k . Then from (4. Remark 1: By the nondegeneracy assumption. we see case.
But then in Step 4 it may turn out that θ = 0 so that k+1 = z k . 1 ↑ ith column Then [D(z k+1 )]−1 = P M [D(z k )]−1 . jm .A . ˆ. jm ] and if . . 1 ≤ i ≤ m .A . + ain xn + yi = bi . . jk+1 . . . ˜. For details see (Canon. In this way the nondegeneracy alternatives for I(z assumption can be eliminated. .A . . . . . .4 is striking.24) we can guarantee that the matrix A ¯ has rank n. Phase I: Step I. . . jk . . . . ˆ. THE SIMPLEX ALGORITHM 41 . ji−1 . . Next. . by multiplying some of the equality constraints in (4. . This net increase is due to the direct increase cj minus the indirect decrease λ (z k )Aj due to the compensating changes in the basic variables necessary to maintain feasibility. Replace the LP (4. . so that these inverses can be easily computed. The reason for this is that I(z k ) is not unique. . . y i ≥ 0 . .A . Let E be the . we can assume that b ≥ 0. .A ]. . . . . . We now describe how to obtain an initial basic feasible solution. ˆ the column Aj . . The basic variables at z k correspond to the variables wk and nonbasic variables correspond to uk . j . . [1970]). it is easy to check that E −1 = M [D(z k )]−1 where 1 1 . . j. .32) involving the variables x and y: m Maximize − i=1 yi (4. . jm ]. xj ≥ 0 . .3..A . The analogous quantity in 3. matrix E = [Aj1 . . Remark 3: By eliminating any dependent equations in (4. 1 ≤ i ≤ m . Hence at any degenerate basic feasible solution z k we can always ﬁnd I(z k ) ⊃ I(z k ) ¯ k ) has m elements and {Aj j ∈ I(z k )} is a linearly independent set. . ji−1 . . . ji+1 . et al. ji+1 . .A . . For each j ∈ I(z k ) we can interpret the number cj − λ (z k )Aj to be the net increase in the objective value per unit increase in the jth component of z k .24) by the LP (4. . . . . Remark 2: The similarity between Step 2 of Phase II and Step 2 of the algorithm in 3. .A .32) subject to ail x1 + . . . . . .A .A .A . jk < ˆ < jk+1 then D(z k+1 ) = [Aj1 . if ˆ Aj m = =1 γj Aj .A j .. .3.A . .A . More precisely if D(z k ) = [Aj1 .4 is (∂f0 /∂uj )(xk ) − (λk ) (∂f /∂uj )(xk ). . j . . so that we have to try various ¯ z ¯ k ) until we ﬁnd one for which θ > 0.4. Then [D(z k+1 )]−1 = P E −1 where the matrix P permutes the columns of D(z k+1 ) such that E = D(z k+1 )P . ji−1 .24) by −1 if necessary. 1 ≤ j ≤ n . . ji+1 . . We can apply ¯ such that I(z ¯ Phase II using I(z k ) instead of I(z k ). 1 M = −γ j1 γ˜ j 1 γ˜ j −γ jm γ˜ j 1 .3.
capital goods 3 such as machines of various kinds. We now make three basic assumptions about the ﬁrm. Labor is not usually considered an output since slavery is not practiced. . The ﬁrm’s outputs themselves may be raw materials (if it is a mining company) or intermediate products (if it is a steel mill) or capital goods (if it manufactures lathes) or ﬁnished goods (if it makes shirts or bakes cookies) which go directly to the consumer. . We think of a ﬁrm as a system which transforms input into outputs.. . different combinations of inputs can be used to produce the same combination of outputs. . .e. . y∗ x∗ y∗ 4. amj xj ) = xj Aj into the output vector (b1j xj . . . chemicals. n ] and B be the k × n matrix B = [B 1 . y 0 ) = (0. Inputs are usually classiﬁed into raw materials such as iron ore. (i) The transformation of inputs into outputs is organized into a ﬁnite number. i. yk ) with y ≥ 0. Precisely. . amj ) and B j = (bij .24) has no feasible solution.4. . (See the von Neumann model in (Nikaido [1968]). (ii) Each activity combines the k inputs in ﬁxed proportions into the m outputs in ﬁxed proportions. the jth activity is characterized completely by two vectors Aj = (a1j . Phase II must terminate in an optimum based feasible solution (x∗ . or textiles. . . CHAPTER 4. . b) is a basic feasible solution of (4. y ∗ ) m since the value of the objective function in (4. . We formalize it by specifying which transformation possibilities are available to the ﬁrm. Let . There are m kinds of inputs and k kinds of outputs.32). Apply phase II to (4.” dynamic Malthusian framework where the increase in labor is a function of the output.B It is more accurate to think of the services of capital goods rather than these goods themselves as inputs. rm ) with r ≥ 0. etc. If = 0. however. It is these services which are consumed in the transformation into outputs. . is a basic feasible solution for (4. then it combines (transforms) the input vector (a1j xj . . Go to Step 3. . . . . by Exercise 3 below. . . this transformation can be conducted in different ways. . ﬁnally various kinds of labor services. 3 . . LINEAR PROGRAMMING Step 2. . . intermediate products such as steel. By an input vector we mean any mdimensional vector r = (r1 . A be the m × n matrix [A1 . . of processes or activities. . or factory buildings.4 LP Theory of a Firm in a Competitive Economy 4.42 Go to step 2. ofﬁce equipment. . . If = 0.24) has a feasible solution iff y ∗ = 0.24). Note that (x0 . . . each activity can be conducted at any nonnegative intensity or level. bkj ) so that if it is conducted at a level xj ≥ 0.A . . . Exercise 3: Show that (4. . and by an output vector we mean any kdimensional vector y = (y1 . or computers. 141. . This substitutability among inputs is a fundamental concept in economics. crude oil. say n. Furthermore. n ].1 Activity analysis of the ﬁrm. or raw cotton.32) starting with this solution.32) lies between − i=1 bi and 0. a2j . bkj xj ) = xj B j . . it may be considered an output in a “closed. . since human labor can do the same job as some machines and machines can replace other kinds of machines.) Within the ﬁrm. (4. Step 3. p.
. .35): Maximize c x − (q ∗ ) ∆ subject to Ax ≤ r ∗ + ∆ . . LP THEORY OF A FIRM IN A COMPETITIVE ECONOMY 43 (iii) If the ﬁrm conducts all the activities simultaneously with the jth activity at level xj ≥ 0. We say that the prices (p∗ .4. (4. . faces the following decision problem: m Maximize p y − j= +1 q j rj (4. if he is maximizing the ﬁrm’s proﬁts. . . .34): Minimize (r ∗ ) q subject to A q ≥ B p∗ q≥0. + ain xn ≤ ri .2 Shortterm behavior.33) subject to y = Bx. + 1 ≤ i ≤ m . . . . the ri are the ﬁxed shortterm supplies. . . . Let us suppose that these ∗ inputs are 1. . then it transforms the input vector x1 A1 + . rm . Theorem 1: p∗ . ∗ ai1 x1 + . . In the shortterm. . + 1 ≤ i ≤ m . . and in turn this depends upon the prices p. By deﬁnition. ai1 x1 + . . (4. . + xn An into the output vector x1 B 1 + . Which of these possible transformations will actually take place depends upon their relative proﬁtability and availability of inputs. rm .33) has an optimal solution. pk ) of the outputs. 1 ≤ j ≤ n. . . and they are available in the amounts r1 . With these assumptions we know all the transformations technically possible as soon as we specify the matrices A and B. q . which the ﬁrm ac∗ cepts as given. . . . . r ∗ are in equilibrium iff for all ﬁxed ∆ ∈ Rm . .3 Longterm equilibrium behavior. x∗ . . . q ∗ ). 4. qm ) of the inputs is ﬁxed. . . q ∗ . q ∗ ) and a set of input supplies ∗ ∗ r ∗ = (r1 . . n 1 4. In the long run the supplies of the ﬁrst inputs are also variable and the ﬁrm can change these ∗ supplies from r1 . xn . + xn B n . The decision variables are the activity levels x1 . . We study this next. + ain xn ≤ ri . .35) . r ∗ .34) Proof: Let c = B p∗ . . r ∗ are in equilibrium if and only if q ∗ is an optimal solution of (4. whereas the pi . and q = (q1 . . 1 ≤ i ≤ .4. and perhaps some raw materials. x≥0. ∗ The coefﬁcients of B and A are the ﬁxed technical coefﬁcients of the ﬁrm. xj ≥ 0. Whether the ﬁrm will actually change these inputs will depend upon whether it is proﬁtable to do so. whereas the supply of the remaining inputs can be varied. the ﬁrm cannot change the amount available to it of some of the inputs such as capital equipment. . . ri ≥ 0 . . . . . qj are prices determined by the whole economy. . We assume that the ﬁrm is operating in a competitive economy which means that the unit prices p = (p1 . Then the manager of the ﬁrm. . rm ) are in (longterm) equilibrium if the ﬁrm has no proﬁt incentive to change r ∗ under the prices (p∗ . x∗ . q ∗ . 2. Under realistic conditions (4. . . q. . . p∗ . certain kinds of labor. 1 ≤ j ≤ n. .4. . M (∆) ≤ M (0) where M (∆) is the maximum value of the LP (4.4. . say. r ∗ by buying or selling these inputs at the market price q1 . r ∗+1 . . and the shortterm input supplies r +1 . .
15) we see that if x= astj > 0 then m cj = i=1 ∗ qi aij . On the other hand if the jth activity is operated at level xj = 1.38) This relation between p∗ . for q = q ∗ . if x∗ is the optimum activity levels for (4. (4.. again from (4.34). Thus. Now bij is the amount of the ith output produced by operating the 1 2 k jth activity at a unit level xj = 1. r ∗ are in equilibrium iff c x − (q ∗ ) ∆ ≤ M (0) = (r ∗ ) q ∗ . (q ∗ ) (Ax∗ − r ∗ ) = 0 so that c x∗ = (q ∗ ) r ∗ . i.e.38): Maximize c x subject to Ax ≤ r ∗ x≥0.35). But from (4.35) so that by the strong duality theorem.34) becomes the dual of (4.36) whenever x is feasible for (4. i.. i. M (0) = (r ∗ ) q ∗ . Recall that c = B p∗ .34)) of (4. Also if m cj < i=1 ∗ qi aij .38) then the output revenue is c x∗ and the input cost is (q ∗ ) Ax∗ .. the revenue of an activity operated at a positive level = input cost of that activity. in particular. q ∗ . q ∗ . From (4.35) and q is feasible for (4. In fact. . in equilibrium. Hence p∗ . then the input cost of operating at xj = 1. x∗ j . If the ith input is valued at m a∗ .39) i. By weak duality if x is feasible for (4. we can say even more. at the optimum activity levels. total revenues = total cost of input supplies. Finally. LINEAR PROGRAMMING For ∆ = 0. Hence. q ∗ . c x − (q ∗ ) ∆ ≤ q (r ∗ = ∆) − (q ∗ ) ∆ . is i i=1 qi aij . if the revenue of an activity is less than its input cost.e. and..e. (4. if an equilibrium the optimum ith input supply ri is greater than the optimum demand for the ith input. at the optimum. c x − (q ∗ ) ∆ ≤ (q ∗ ) (r ∗ + ∆) − (q ∗ ) ∆ = (q ∗ ) r ∗ ♦ (4.44 CHAPTER 4.37) Remark 1: We have shown that (p∗ . . r ∗ has a very nice economic interpretation. + p∗ bkj . cj is the revenue per unit level operation of the jth activity so that c x is the revenue when the n activities are operated at levels x. then at the optimum it is ∗ operated at zero level. cj = p∗ b1j + p∗ b2j + .e. (4. it uses an amount aij of the ith input.15).16). r ∗ are in longterm equilibrium iff q ∗ is an optimum solution to the dual (namely (4. then = 0. so that the input cost of operating the n activities at levels x is (A q ∗ ) = (q ∗ ) Ax. (4.
. B) characterizing a ﬁrm we can suppose that all the h commodities are possible inputs and all the h commodities are possible outputs. The design procedure is to ﬁx these parameters at some nominal value b∗ .e. . j ∗ then qi = 0. Then if (∂M/∂∆i )∆=0 exists. . λh ) ≥ 0. λ∗+1 . We are including in ω(j) the amount of his labor services which consumer j is willing to sell. Now suppose that at time T the prevailing prices of the h commodities are λ = (λ1 . labor. . λ∗ . . ∆ ) the optimum value of ∗ (4. Suppose that the market m 1 prices of inputs 1. . r(i) ∈ Rh .33) when the amounts of the inputs in ﬁxed supply are r1 + ∆1 . . the sole purpose for making this change is that we no longer need to distinguish between prices of inputs and prices of outputs. capitalist economy. Let us suppose that there are a total of h commodities in the economy including raw materials. . the jth consumer owns the vector of commodiJ ties ω(j) ≥ 0. Often engineering design problems can be formulated as LPs of the form (4. . which means that the ownership of ω is divided among the various consumers j = 1. ω is that portion of the outputs produced prior to T which have not been consumed up to T . i 4. ωh ) ≥ 0. More precisely. . then we see (assuming i differentiability) that it is worth increasing b∗ if the unit cost of increasing this parameter is less i than λ∗ . from our previous analysis we know that the manager of the ith ﬁrm will plan to buy input supplies r(i) ≥ 0. . and conversely it is proﬁtable to sell some i of the ith input at price qi if λ∗ < qi . . are q1 . Let us denote by M (∆1 . . such . and j=1 ω(j) = ω. We think of the economy as a feedback process involving ﬁrms and consumers. . The proﬁtmaximizing behavior of the ﬁrm presented above is one of the two fundamental building blocks in the equilibrium theory of a competitive. . q . we can see from (4. Suppose the resulting optimal dual variables are λ∗ . λ∗ ) is an optimum solution of the dual of (4. the equilibrium price of an input which is in excess supply must be zero. Thus λ∗ can be interpreted as the ﬁrm’s internal valuation of i i the ith input or the ﬁrm’s imputed or shadow price of the ith input. Of course. . . . Then. LP THEORY OF A FIRM IN A COMPETITIVE ECONOMY n ∗ ri > j=1 45 aij x∗ . suppose that (λ∗ .. . intermediate and capital goods.4. We are assuming that this is a capitalist economy. J. which we mention brieﬂy. and it is worth decreasing this parameter if the reduction in total cost per unit decrease is i greater than λ∗ .4. . Unfortunately we cannot present the details here. suppose that the managers of the various ﬁrms assume that the prices λ are not going to change for a long period of time. .10) or (4. .22) that it is always proﬁtable to increase the ith input by buying some additional amount at price qi if λ∗ > qi . for an individual ﬁrm most of the inputs and most of the outputs will be zero. We observe the economy starting at time T .33). By adding zero rows to the matrices (A. . r ∗ + ∆ . . . . .19). Remark 2: Returning to the shortterm decision problem (4. Next.33). . and carry out the i optimization problem. capitalist economy. . in other words it must be a free good. . where some of the coefﬁcients bi are design parameters. . . i. . .4 Longterm equilibrium of a competitive. This interpretation has wide applicability. and ﬁnished products.4. We shall limit ourselves to a rough sketch. . . At this time there exists within the economy an inventory of the various commodities which we can represent by a vector ω = (ω1 .
say y(i). and he will plan to produce an optimum amount. LINEAR PROGRAMMING that (λ.42) that is the value of the supply offered to consumers is equal to the value of the commodities (and labor) which they own. I. the net supply offered for sale to consumers is S(λ). The value of the jth consumer’s possessions is λ ω(j). The theory assumes that he will plan to buy a set of commodities d(j) = (d1 (j). 2. Chapter V). 1 ≤ i ≤ I . λ) . (4. 4. . and Solow [1958]. .41) We note two important facts. λ). λ) − i=1 r(i. so that we explicitly write r(i. r(i)) is in long term equilibrium. λ). We know that r(i) and y(i) depend on λ. Here i = 1. Chapter 13). where J I i S(λ) = j=1 ω(j) + i=1 y(i.40). Unfortunately we must stop at this point since we cannot proceed further without introducing some more convex analysis and the ﬁxed point theorem. (4. (4.46 CHAPTER 4.40) Now the ith manager can buy r(i) from only two sources: outputs from other ﬁrms.e.44) The most basic question of equilibrium theory is to determine conditions under which there exists a price vector λE such that the economy is in equilibrium. For a much more general mathematical treatment see (Nikaido [1968]. The second point is that there is no reason to expect that S(λ) ≥ 0. then at that price the production plans of all the ﬁrms and the buying plan of all the consumers can be realized. Here also d(j) will depend on λ. and the consumers who collectively own ω. where I is the total number of ﬁrms. First of all. (4. λ). If we add up the buying plans of all the consumers we obtain the total demand J D(λ) = j=1 d(j. because if such an equilibrium price λE exists. Samuelson. We also recall that (see (4. . S(λE ) = D(λE ). . . (4. λ) ≥ 0 . dh (j)) ≥ 0 so as to maximize his satisfaction subject to the constraint λ d(j) = λ ω(j). y(i. For a simple treatment the reader is referred to (Dorfman. i. . λ) = λ y(i.41) we immediately conclude that J λ S(λ) = j=1 λ ω(j) .38)) λ r(i. from (4. Now we come to the second building block of equilibrium theory. (4. λ) . so we write d(j. ..5 Miscellaneous Comments . . Thus.43) which also satisﬁes J λ D(λ) = j=1 λ ω(j) . Similarly. the ith manager can sell his planned output y(i) either as input supplies to other ﬁrms or to the consumers.
and such that there is no equivalent LP. To obtain a feeling for the scope of LP we refer the reader to the book by one of the originators of LP (Dantzig [1963]). See (Miller [1963]). However. x ≤ 0 y ≤ (ci ) x . .47).5. However. (4. the following exercise shows how to transform (4. It turns out however. (c2 ) x.45) into an equivalent LP.46) below. where ci : R → R are concave.47): n (4. x ≥ 0 .3. where the ci are piecewise linear (but not concave). and. given the capabilities of modern computers. . x ≥ 0. piecewiselinear functions of the kind shown in Figure 4. Exercise 3: Construct an example of the kind (4. MISCELLANEOUS COMMENTS 47 4. an elementary modiﬁcation of the Simplex algorithm can be given to obtain a “local” optimal decision. It is often the case in practical decision problems that the objective is not welldeﬁned. that even if the ci are not concave. 1 ≤ i ≤ k .47) subject to Ax ≤ b. x ≤ 0 . . In our LP framework this situation can be formulated as follows.45) This is not a LP since f0 is not linear. It is reasonable then to deﬁne a single objective function f0 (x) by f0 (x) = minimum {(c1 ) x. This is because many decision problems can be adequately formulated as LPs. Exercise 2: Obtain an equivalent LP for (4.1 Some mathematical tricks. .45) is equivalent to (4. There may be a number of plausible objective functions. the Simplex method (together with its variants) is an extremely powerful technique for solving LPs involving thousands of variables. .5. The constraints are given as usual by Ax ≤ b. . The abovegiven assumption of the concavity of the ci is crucial.45) iff (x∗ . Maximize f0 (x) subject to Ax ≤ b. .4.2 Scope of linear programming. . LP is today the single most important optimization technique. the interpretation of “equivalent” is purposely left ambiguous. In the next exercise. k objective functions (c1 ) x.46). in the sense that x∗ is optimal for (4. . f0 (x∗ )) is optimal for (4. Exercise 1 will also indicate how to do Exercise 2. y ∗ ) = (x∗ .46) Maximize j=1 ci (xi ) (4. say. there are. 4. Maximize y subject to Ax ≤ b. so that we have the decision problem. Exercise 1: Show that (4.5. (ck ) x}. (ck ) x.
. . LINEAR PROGRAMMING ci (xi ) .48 CHAPTER 4. xi Figure 4.3: A function of the form used in Exercise 2. .
1. .1 Qualitative Theory of Nonlinear Programming 5. m. . . are differentiable functions.Chapter 5 OPTIMIZATION OVER SETS DEFINED BY INEQUALITY CONSTRAINTS: NONLINEAR PROGRAMMING In many decisionmaking situations the assumption of linearity of the constraint inequalities in LP is quite restrictive. fi : Rn → R. and Ω ⊂ Rn is the subset of all feasible solutions. The general NP is a decision problem of the form: Maximize f0 (x) subject to (x) ≤ 0 . (5.2 it is clear that equality constraints and sign constraints on some of the components of x can all be transformed into the form (5. Two applications are given.1). i = 0. x ∈ Rn . is equivalent to (5.1) where x ∈ Rn . . The linearity of the objective function is not restrictive as shown in the ﬁrst exercise below. As in Chapter 4. The next exercise shows that we could restrict ourselves to objective functions which are linear. . . x∗ ∈ Ω is said to be an optimal decision or optimal solution if f0 (x∗ ) ≥ f0 (x) for x ∈ Ω.1 The problem and elementary results. i = 1. x ∈ Rn is said to be a feasible solution if it satisﬁes the constraints of (5.1. we will not do this. however.1). . The last section is devoted to computational considerations.1): 49 .1. .2). Section 2 deals with Duality theory for the case where appropriate convexity conditions are satisﬁed. with variables y ∈ R. Section 3 is devoted to the important special case of quadratic programming. m. From the discussion in 4. 5. Exercise 1: Show that (5. In Section 1 we present the general nonlinear programming problem (NP) and prove the KuhnTucker theorem.
Then f0x (x∗ )h ≤ 0 for all h ∈ C(Ω.1). x)}. Deﬁnition: Let x be a feasible solution.50 CHAPTER 5. . x) is a closed subset of Rn . .3) . . . Exercise 2: (i) Show that C(Ω. x∗ ) . and y − f0 (x) ≤ 0 . in Ω and a sequence of numbers εk . NONLINEAR PROGRAMMING Maximize y subject to fi (x) ≤ 0. i. we see that 0 ∈ C(Ω. . x) we made no use of the particular functional description of Ω.1). (Hint for (ii): For m = 1. k = 1. 2. εk > 0. k = 1. fi (x) < 0 for i ∈ I(x). The argument parallels very closely that developed in Exercise 1 of 4.) In the deﬁnition of C(Ω.1 and 4. Show that there exist subsequences {xmkm . (ii) Let C(Ω. εmkm }∞ such that m=1 xmkm → x and (1/εmkm )(xmkm − x) → h as m → ∞. we are interested in obtaining conditions which any optimal decision must satisfy. x) = {x + hh ∈ C(Ω. k→∞ k→∞ lim 1 (xk εk − x) = h . A vector h ∈ Rn is said to be an admissible direction for Ω at x if there exists a sequence xk . Suppose k=1 that hm → h as m → ∞. Let K(Ω. x) so that the tangent cone is always nonempty. (See Figures 5. m} be such that fi (x) = 0 for ı ∈ I(x).2) Returning to problem (5. k = 1. be such that (5. if h ∈ C(Ω.18) in Chapter 2 and Exercise 1 of 4.e. 2. . .) If we take xk = x and εk = 1 for all k. Proof: Let xk ∈ Ω. x). . x) is a cone. (5. Two more properties are stated below. εmk > 0}∞ be such that xmk → x and (1/εmk )(xmk − x) → hm as k → ∞. . (ii) Show that C(Ω. . (The set I(x) is called the set of active constraints at x. x) and θ ≥ 0. 2. The basic idea is to linearize the functions fi in a neighborhood of an optimal decision x∗ . C(Ω. . . .1.1 and 5. .2.. x) = {hh is an admissible direction for Ω at x}.2 and compare them with Figures 4. 3.2. . . .) Deﬁnition: (i) Let x ∈ Ω. let hm and {xmk . with εk > 0 for all k such that k lim x = x . . The following elementary result is more interesting in this light and should be compared with (2. . x) is called the tangent cone of Ω at x. . Lemma 1: Suppose x∗ ∈ Ω is an optimum decision for (5. 1 ≤ i ≤ m.1 and Exercise 1 of 4. then θh ∈ C(Ω. and let I(x) ⊂ {1.
4) implies lim 1 εk xk − x∗  = h . Taking limits as k → ∞. using (5. 1 (xk εk k→∞ k→∞ lim − x∗ ) = h . ♦ k→∞ lim o(xk −x∗ ) xk −x∗  k→∞ lim xk −x∗  εk . (5. so that 0 ≥ f0x (x∗ ) (x k −x∗ ) (5.6) εk + o(xk −x∗ ) εk . Since xk ∈ Ω. and x∗ is optimal.4) Note that in particular (5.5) k→∞ Since f0 is differentiable. by Taylor’s theorem we have f0 (xk ) = f0 (x∗ + (xk − x∗ )) = f0 (x∗ ) + f0x (x∗ )(xk − x∗ ) + o(xk − x∗ ) . QUALITATIVE THEORY OF NONLINEAR PROGRAMMING direction of increasing payoff π(k) = {xf0 (x) = k} Q x∗ 51 {xf3 (x) = 0} P {xf2 (x) = 0} Ω R .1.5).4) and (5. {xf1 (x) = 0} Figure 5. (5.5.1: Ω = P QR k ∗ lim x = x . we have f0 (xk ) ≤ f0 (x∗ ). we can see that 0≥ = k→∞ lim f0x (x∗ ) (xk −x∗ ) εk + f0x (x∗ )h.
Then we can apply Farkas’ Lemma just as in Exercise 1 of 4. x∗ ) is the tangent cone of Ω at x∗ . 0). NONLINEAR PROGRAMMING x∗ K(Ω. . fi (xk ) ≤ 0. and if i ∈ I(x∗ ). f1 (x1 . x∗ ) ⊂ {hfix (x∗ )h ≤ 0 for all i ∈ I(x∗ )} . The main reason for this is that the set {fix (x∗ )i ∈ I(x∗ )} is not in general linearly independent. Then I(x∗ ) = {1. Let (x∗ . . in general the inclusion sign in (5. εk > 0.7) Proof: Let h ∈ Rn and xk ∈ Ω. .4). Exercise 3: Let x ∈ R2 . 2. x2 ) = −x2 . by Taylor’s theorem we have fi (xk ) = fi (x∗ ) + fix (x∗ )(xk − x∗ ) + o(xk − x∗ ) . x∗ ) Figure 5. x∗ ) in terms of the derivatives of the functions fi .2: C(Ω. satisfy (5. 2}. Show that 1 2    . ♦ Lemma 2 gives us a partial characterization of C(Ω. .7) cannot be reversed. Following the proof of Lemma 1 we can conclude that 0 ≥ fix (x∗ )h. k = 1. Then C(Ω. x∗ ). x2 ) = (x1 − 1)3 + x2 . so that fi (xk ) ≤ fi (x∗ ). fi (x∗ ) = 0. x∗ )     0  C(Ω. Since xk ∈ Ω. (5. Unfortunately. Since fi is differentiable. Lemma 2: Let x∗ ∈ Ω.2. The basic problem that remains is to characterize the set C(Ω.52 CHAPTER 5. x∗ ) = (1. and f2 (x1 .
5.1. QUALITATIVE THEORY OF NONLINEAR PROGRAMMING
C(Ω, x∗ ) = {hfix (x∗ )h ≤ 0 , i = 1, 2, }. (Note that {f1x (x∗ ), f2x (x∗ )} is not a linearly independent set; see Lemma 4 below.)
53
5.1.2 KuhnTucker Theorem.
Deﬁnition: Let x∗ ∈ Ω. We say that the constraint qualiﬁcation (CQ) is satisﬁed at x∗ if C(Ω, x) = {hfix (x∗ )h ≤ 0 for all i ∈ I(x∗ )}, and we say that CQ is satisﬁed if CQ is satisﬁed at all x ∈ Ω. (Note that by Lemma 2 C(Ω, x) is always a subset of the righthand side.) Compare the next result with Exercise 2 of 4.2. Theorem 1: (Kuhn and Tucker [1951]) Let x∗ be an optimum solution of (5.1), and suppose that CQ is satisﬁed at x∗ . Then there exist λ∗ ≥ 0, for i ∈ I(x∗ ), such that i f0x (x∗ ) =
i∈I(x∗ )
λ∗ fix (x∗ ) i
(5.8)
Proof: By Lemma 1 and the deﬁnition of CQ it follows that f0x (x∗ )h ≤ 0 whenever fix (x∗ )h ≤ 0 for all i ∈ I(x∗ ). By the Farkas’ Lemma of 4.2.1 it follows that there exist λ∗ ≥ 0 for i ∈ I(x∗ ) i such that (5.8) holds. ♦ In the original formulation of the decision problem we often have equality constraints of the form rj (x) = 0, which get replaced by rj (x) ≤ 0, −rj (x) ≤ 0 to give the form (5.1). It is convenient in application to separate the equality constraints from the rest. Theorem 1 can then be expressed as Theorem 2.
Theorem 2: Consider the problem (5.9). Maximize f0 (x) subject to fi (x) ≤ 0 , i = 1, . . . , m, rj (x) = 0 , j = 1, . . . , k .
(5.9)
Let x∗ be an optimum decision and suppose that CQ is satisﬁed at x∗ . Then there exist λ∗ ≥ 0, i = i 1, . . . , m, and µ∗ , j = 1, . . . , k such that j
m
f0x (x∗ ) =
i=1
λ∗ fix (x∗ ) + i
k j=1
µ∗ rjx (x∗ ) , j
(5.10)
and λ∗ = 0 whenever fi (x∗ ) < 0 . i Exercise 4: Prove Theorem 2. (5.11)
54
CHAPTER 5. NONLINEAR PROGRAMMING
An alternative form of Theorem 1 will prove useful for computational purposes (see Section 4). Theorem 3: Consider (5.9), and suppose that CQ is satisﬁed at an optimal solution x∗ . Deﬁne ψ : Rn → R by ψ(h) = max {−f0x (x∗ )h, f1 (x∗ ) + f1x (x∗ )h, . . . , fm (x∗ ) + fmx (x∗ )h} , and consider the decision problem Minimize ψ(h) subject to −ψ(h) − f0x (x∗ )h ≤ 0, −ψ(h) + fi (x∗ ) + fix (x∗ )h ≤ 0 , 1 ≤ i ≤ m −1 ≤ hi ≤ 1 , i = 1, . . . , n . Then h = 0 is an optimal solution of (5.12). Exercise 5: Prove Theorem 3. (Note that by Exercise 1 of 4.5, (5.12) can be transformed into a LP.) Remark: For problem (5.9) deﬁne the Lagrangian function L:
m k
(5.12)
(x1 , . . . , xn ; λ1 , . . . , λm ; µ1 , . . . , µk ) → f0 (x) −
i=1
λi fi (x) −
j=1
µj rj (x).
Then Theorem 2 is equivalent to the following statement: if CQ is satisﬁed and x∗ is optimal, then there exist λ∗ ≥ 0 and µ∗ such that Lx (x∗ , λ∗ , µ∗ ) = 0 and L(x∗ , λ∗ , µ∗ ) ≤ L(x∗ , λ, µ) for all λ ≥ 0, µ. There is a very important special case when the necessary conditions of Theorem 1 are also sufﬁcient. But ﬁrst we need some elementary properties of convex functions which are stated as an exercise. Some additional properties which we will use later are also collected here. Recall the deﬁnition of convex and concave functions in 4.2.3. Exercise 6: Let X ⊂ Rn be convex. Let h : X → R be a differentiable function. Then (i) h is convex iff h(y) ≥ h(x) + hx (x)(y − x) for all x, y, in X, (ii) h is concave iff h(y) ≤ h(x) + hx (x)(y − x) for all x, y in X, (iii) h is concave and convex iff h is afﬁne, i.e. h(x) ≡ α + b x for some ﬁxed α ∈ R, b ∈ Rn . Suppose that h is twice differentiable. Then (iv) h is convex iff hxx (x) is positive semideﬁnite for all x in X, (v) h is concave iff hxx (x) is negative semideﬁnite for all x in X, (vi) h is convex and concave iff hxx (x) ≡ 0. Theorem 4: (Sufﬁcient condition) In (5.1) suppose that f0 is concave and fi is convex for i = 1, . . . , m. Then (i) Ω is a convex subset of Rn , and (ii) if there exist x∗ ∈ Ω, λ∗ ≥ 0, i ∈ I(x∗ ), satisfying (5.8), then x∗ is an optimal solution of i (5.1). Proof: (i) Let y, z be in Ω so that fi (y) ≤ 0, fi (z) ≤ 0 for i = 1, . . . , m. Let 0 ≤ θ ≤ 1. Since fi is convex we have
5.1. QUALITATIVE THEORY OF NONLINEAR PROGRAMMING
fi (θy + (1 − θ)z) ≤ θfi (y) + (1 − θ)fi (z) ≤ 0 , 1 ≤ i ≤ m, so that (θy + (1 − θ)z) ∈ Ω, hence Ω is convex. (ii) Let x ∈ Ω be arbitrary. Since f0 is concave, by Exercise 6 we have f0 (x) ≤ f0 (x∗ ) + f0x (x∗ )(x − x∗ ) , so that by (5.8) f0 (x) ≤ f0 (x∗ ) +
i∈I(x∗ )
55
λ∗ fix (x∗ )(x − x∗ ) . i
(5.13)
Next, fi is convex so that again by Exercise 6, fi (x) ≥ fi (x∗ ) + fix (x∗ )(x − x∗ ) ; but fi (x) ≤ 0, and fi (x∗ ) = 0 for i ∈ I(x∗ ), so that fix (x∗ )(x − x∗ ) ≤ 0 for i ∈ I(x∗ ) . (5.14) Combining (5.14) with the fact that λ∗ ≥ 0, we conclude from (5.13) that f0 (x) ≤ f0 (x∗ ), so that i x∗ is optimal. ♦ Exercise 7: Under the hypothesis of Theorem 4, show that the subset Ω∗ of Ω, consisting of all the optimal solutions of (5.1), is a convex set. Exercise 8: A function h : X → R deﬁned on a convex set X ⊂ Rn is said to be strictly convex if h(θy + (1 − θ)z) < θh(y) + (1 − θ)h(z) whenever 0 < θ < 1 and y, z are in X with y = z. h is said to be strictly concave if −h is strictly convex. Under the hypothesis of Theorem 4, show that an optimal solution to (5.1) is unique (if it exists) if either f0 is strictly concave or if the fi , 1 ≤ i ≤ m, are strictly convex. (Hint: Show that in (5.13) we have strict inequality if x = x∗ .)
5.1.3 Sufﬁcient conditions for CQ.
As stated, it is usually impractical to verify if CQ is satisﬁed for a particular problem. In this subsection we give two conditions which guarantee CQ. These conditions can often be veriﬁed in practice. Recall that a function g : Rn → R is said to be afﬁne if g(x) ≡ α + b x for some ﬁxed α ∈ R and b ∈ Rn . We adopt the formulation (5.1) so that Ω = {x ∈ Rn fi (x) ≤ 0 , 1 ≤ i ≤ m} . Lemma 3: Suppose x∗ ∈ Ω and suppose there exists h∗ ∈ Rn such that for each i ∈ I(x∗ ), either fix (x∗ )h∗ < 0, or fix (x∗ )h∗ = 0 and fi is afﬁne. Then CQ is satisﬁed at x∗ . Proof: Let h ∈ Rn be such that fix (x∗ )h ≤ 0 for i ∈ I(x∗ ). Let δ > 0. We will ﬁrst show that (h + δh∗ ) ∈ C(Ω, x∗ ). To this end let εk > 0, k = 1, 2, . . . , be a sequence converging to 0 and set xk = x∗ + εk (h + δh∗ ). Clearly xk converges to x∗ , and (1/εk )(xk − x∗ ) converges to (h + δh∗ ). Also for i ∈ I(x∗ ), if fix (x∗ )h < 0, then fi (xk ) = fi (x∗ ) + εk fix (x∗ )(h + δh∗ ) + o(εk h + δh∗ ) ≤ δεk fix (x∗ )h∗ + o(εk h + δh∗ ) < 0 for sufﬁciently large k , whereas for i ∈ I(x∗ ), if fi is afﬁne, then
56 CHAPTER 5. . for i ∈ Jδ . k = 1.15) and (5. consist of p elements. η ∗ ) corresponding to the partition of x = (w. u) = 0. wk = g(uk ). ♦ Exercise 9: Suppose x∗ ∈ Ω and suppose there exists x ∈ Rn such that for each i ∈ I(x∗ ). h∗ = (ξ ∗ . Since g is differentiable. an open set V ⊂ Rn containing x∗ = (w∗ . and so by deﬁnition (h + δh∗ ) ∈ C(Ω. fi x(x∗ )h∗ = 0}. Finally. such that fi (w. h∗ as h = (ξ. uk ) = 0. and w = g(u) . x∗ ). We will show that (h + δh∗ ) ∈ C(Ω.15) Also. εk (η + δη ∗ )). and hence 0 = fiw (x∗ )gu (u∗ )(η + δη ∗ ) + fiu (x∗ )(η + δη ∗ ) . u∗ )i ∈ Jδ } is linearly independent. Let δ > 0. so that {fix (x∗ . u) ∈ V iff u ∈ U. where U = {u ∈ Rn−p u − u∗  < ρ}. 0 = fi (g(u). there exist ρ > 0. . and ﬁnally xk = (sk . fix (x∗ )(h + δh∗ ) = 0}. First of all. Next we partition h. i ∈ I(x∗ ). x∗ ) is a closed set by Exercise 2. fix (x∗ )h∗ = 0} is a linearly independent set. and a differentiable function g : U → Rp . NONLINEAR PROGRAMMING fi (xk ) = fi (x∗ ) + εk fix (x∗ )(h + δh∗ ) ≤ 0 for all k . fi (xk ) = fi (x∗ ) + fix (x∗ )(xk − x∗ ) + o(xk − x∗ ) fi (x∗ ) + εk fix (x∗ )(h + δh∗ ) + o(εk ) + o(xk − x∗ ). But for i ∈ Jδ we have 0 = fix (x∗ )(h + δh∗ ) = fiw (x∗ )(ξ + δξ ∗ ) + fiu (x∗ )(η + δη ∗ ) . and set uk = u∗ + εk (η + δη ∗ ). so wk = g(uk ) converges to w∗ = g(u∗ ). x∗ ). Clearly Jδ ⊂ J = {ii ∈ I(x∗ ). and {fix (x∗ )i ∈ I(x∗ ). 2 .16) If we compare (5. xk converges to x∗ . η + δη ∗ ). Let Jδ = {ii ∈ I(x∗ ). u) for u ∈ U so that 0 = fiw (x∗ )gu (u∗ ) + fiu (x∗ ). either ˆ ∗ ) < 0 and f is convex. and (w.16) and recall that {fiw (x∗ )i ∈ Jδ } is a basis in Rp we can conclude that (ξ + δξ ∗ ) = gu (u∗ )(η + δη ∗ ) so that (1/εk )(xk − x∗ ) converges to (h + hδh∗ ). Now (1/εk )(xk − x∗ ) = (1/εk )(wk − w∗ . and since C(Ω. Since δ > 0 can be arbitrarily small. Proof: Let h ∈ Rn be such that fix (x∗ )h ≤ 0 for all i ∈ I(x∗ ). (5. u∗ ). fi (xk ) = fi (g(uk ). Then CQ is satisﬁed at x∗ .) ˆ Lemma 4: Suppose x∗ ∈ Ω and suppose there exists h∗ ∈ Rn such that fix (x∗ )h∗ ≤ 0 for i ∈ I(x∗ ). so that fi (xk ) < 0 for sufﬁciently large k. i ∈ Jδ . or f (ˆ) ≤ 0 and f is afﬁne. u). . Let εk > 0. Then CQ is satisﬁed at x∗ . η). be any sequence converging to 0. for i ∈ I(x∗ ) we have fi (x∗ ) < 0. for i ∈ Jδ . Thus we have also shown that xk ∈ Ω for sufﬁciently large k. uk − u∗ ) = (1/εk )(g(uk ) − g(u∗ ). (Hint: Show fi (x i i x i that h∗ = x − x∗ satisﬁes the hypothesis of Lemma 3. it follows that h ∈ C(Ω. . whereas for i ∈ Jδ . (5. Thus. We note that uk converges to u∗ . it follows that (1/εk )(xk − x∗ ) converges to (gu (u∗ )(η + δη ∗ ). It remains to show that xk ∈ Ω for sufﬁciently large k. By the Implicit Function Theorem. uk ). x∗ ).
However. so that h ∈ C(Ω. Exercise 10: Prove Lemma 5 5. We can only present some of the basic results here. x∗ ). . We wish to examine the behavior of the maximum value of (5. are given convex functions. x∗ ).3 we give some application of duality theory and in 2.9). x∗ ) is closed by Exercise 2. . and C(Ω. 5. In 2. fi : Rn → R. fm ) : Rn → Rm .2. and it has provided many unifying conceptual insights into economics and management science. (5. 1 ≤ i ≤ m. Lemma 5: Suppose x∗ is feasible for (5. . k} is linearly independent.2 we refer to some of the important generalizations.17) as ˆ varies. X is a given convex subset of Rn and ˆ = (ˆ1 . we will give some geometric insight. b b b For convenience. Let λ ∈ Rm . fix (x∗ )h∗ = 0} {rjx (x∗ )j = 1. rjx (x∗ )h∗ = 0 for 1 ≤ j ≤ k. . Its proof is left as an exercise since it is very similar to the proof of Lemma 4. f (x) ≤ b} = sup{f0 (x)x ∈ Ω(b)} . The results in 2.17) which we call the primal problem: Maximize f0 (x) subject to fi (x) ≤ ˆi . We need to consider b) x the following problem also. ˆm ) is a given vector. It has resulted in many applications within nonlinear programming. So we deﬁne b Ω(b) = {xx ∈ X. λ ≥ 0.2. so that in particular if x∗ is an optimal solution of (5. . in terms of suggesting important computational algorithms. let f = (f1 .2 Duality Theory Duality theory is perhaps the most beautiful part of nonlinear programming. and fix (x∗ )h∗ ≤ 0 for i ∈ I(x∗ ). .1 should be compared with Theorems 1 and 4 of 4. Hence. To ﬁnish the proof we note that δ > 0 can be made arbitrarily small. we can conclude that fi (xk ) < 0 for sufﬁciently large k. and M : B→R {+∞} by M (b) = sup{f0 (x)x ∈ X. (h + δh∗ ) ∈ C(Ω. 1 ≤ i ≤ m b x∈X .2. .1 and the results in 4. and even so some of the proofs are relegated to the Appendix at the end of this Chapter since they depend on advanced material.5.9) and suppose there exists h∗ ∈ Rn such that the set {fix (x∗ )i ∈ I(x∗ ). f0 : Rn → R is a given concave function. Thus. f (x) ≤ b}. B = {bΩ(b) = φ}.17) where x ∈ Rn . Maximize f0 (x) − λ (f (x) − ˆ b) subject to x ∈ X . (5. . . .2. Consider problem (5.17) then M (ˆ = f0 (ˆ). ♦ The next lemma applies to the formulation (5. DUALITY THEORY 57 and since fi (x∗ ) = 0 whereas fix (x∗ )(h + δh∗ ) < 0.1 Basic results. Then CQ is satisﬁed at x∗ . xk ∈ Ω for sufﬁciently large k. It may be useful to note in the following discussion that most of the results do not require differentiability of the various functions. be ﬁxed.3.18) . . .
b). and since M (ˆ is independent of λ. b) b). if we take the inﬁmum with respect to λ ≥ 0 in the righthand b) inequality we get (5.) Exercise 1: Prove Lemma 1.20) .e. and λ ≤ 0 is said to satisfy the optimality conditions if x ˆ ˆ (5. b b) f0 (x) ≤ f0 (x) − λ (f (x) − ˆ for x ∈ Ω(ˆ λ ≥ 0 . there is no reason to separate it out.19) is called the dual problem: Minimize m(λ) subject to λ ≥ 0 . f. and λ ≥ 0.20).2 below.3. We ﬁrst give a simple sufﬁciency condition. (5. ♦ ˆ = m∗ in The basic problem of Duality Theory is to determine conditions under which M (b) (5. For example see the problems discussed in Sections 2. n n Lemma 1: m : R+ → R {+∞} is a convex function. it is sometimes possible to include some of the constraints in X in such a way that the calculation of m(λ) by (5. Remark 2: It is sometimes useful to know that Lemmas 1 and 2 below hold without any convexity conditions on f0 . ˆ Deﬁnition: A pair (ˆ..19) Let m∗ = inf {m(λ)λ ≥ 0}. Remark 1: The set X in (5. Lemma 1 shows that the cost function of the dual problem is convex which is useful information since there are computation techniques which apply to convex cost functions but not to arbitrary nonlinear cost functions. So.20).19) become simple. f0 (x) ≤ M (ˆ ≤ m∗ ≤ m(λ) . i.17) is usually equal to Rn and then. However. λ) with x ∈ X. b)x Problem (5. NONLINEAR PROGRAMMING m(λ) = sub{f0 (x) − λ (f (x) − ˆ ∈ X} .1 and 2. Lemma 2 shows that the optimum value of the dual problem is always an upper bound for the optimum value of the primal.58 and deﬁne CHAPTER 5.17). we have λ (f (x) − ˆ ≤ 0. b) Proof: Since f (x) − ˆ ≤ 0. X. b). Lemma 2: (Weak duality) If x is feasible for (5.18) and the solution of the dual problem (5. then b). Hence f0 (x) ≤ sup {f0 (x)x ∈ Ω(ˆ = M (ˆ b)} b) ˆ ∈ Ω(ˆ and since Ω(ˆ ⊂ X. of course. x ∈ Ω(ˆ and if λ ≥ 0. ≤ sup {f0 (x) − λ (f (x) − b)x b)} b) ≤ sup {f0 (x) − λ (f (x) − ˆ ∈ X} = m(λ) .3. b)x Thus. we have f0 (x) ≤ M (ˆ ≤ m(λ) for x ∈ Ω(ˆ λ ≥ 0 . (Here R+ = {λ ∈ Rn λ ≥ 0}.
x since X is convex. x hence (θx + (1 − θ)˜) ∈ Ω(θb + (1 − θ)˜ x b) and therefore. so that b fi (θx + (1 − θ)˜) ≤ θb + (1 − θ)˜ . λ) satisfy the optimality ˆ x ˆ ˆ by virtue of (5. condition. Note that in this case x ∈ Ω(b) ˆ The next result is equivalent to Theorem 4(ii) of Section 1 if X = Rn . are differentiable.21) x x b) = f0 (ˆ) by (5. λ is an optimal solution to the dual. f0 (ˆ) = M (b) x ˆ so that from Weak Duality λ is optimal for the dual. since f0 is concave. . B is convex. x b x b) (5. Theorem 1: (Sufﬁciency) If (ˆ.24) . λ) satisfy the optimality conditions. Also.23) x so that x is optimal for the primal. b) ˆ f0 (x) ≤ f0 (x) − λ (f (x) − ˆ b) ˆ (f (x) − ˆ ∈ X} ≤ sup{f0 (x) − λ b)x ˆ = f0 (ˆ) − λ (f (ˆ) − ˆ by (5. Proof: Let b.23) ˆ λ ≥ 0 is said to be an optimal price vector if there is x ∈ X such that (ˆ. and fi (θx + (1 − θ)˜) ≤ θfi (x) + (1 − θ)fi (˜) x x since fi is convex.21) x is feasible for (5. m .22). i. and M : B → R {+∞} is a concave function. ˆ x b (5. . . DUALITY THEORY ˆ x is optimal solution of (5. ♦ We now proceed to a much more detailed investigation. let x ∈ Ω(b).e. and M (ˆ = m∗ . equivalently.17).5. f0 (θx + (1 − θ)˜) ≥ θf0 (x) + (1 − θ)f0 (˜) . fi (ˆ) ≤ ˆi for i = 1. ˆ ˆ m(λ) = f0 (ˆ) − λ (f (ˆ) − ˆ x x b) ˆ . then x is an optimal solution to x ˆ ˆ ˆ the primal. b) ˆ Proof: Let x ∈ Ω(ˆ so that λ (f (x) − ˆ ≤ 0. . and hence by deﬁnition f0 (ˆ) = M (ˆ Also ˆ x b). and fi . Lemma 3: B is a convex subset of Rm . λ (f (ˆ) − ˆ = 0. x x (5.22) ˆ ˆ λi = 0 when fi (ˆ) < ˆi .18) with λ = λ.2. Then b).. Then (θx + (1 − θ)˜) ∈ X b ˜ b). 0 ≤ i ≤ m. ˜ belong to B. ˆ 59 (5. x ∈ Ω(˜ let 0 ≤ θ ≤ 1.
Deﬁnition: The function M : B → R number K such that {∞} is said to be stable at ˆ ∈ B if there exists a real b M (b) ≤ M (ˆ + Kb − ˆ for b ∈ B .∈M (ˆ b B . x ∈ Ω(˜ b) x ˜ b)} ≥ sup{f0 (x)x ∈ Ω(b)} + (1 − θ) sup {f0 (˜)˜ ∈ Ω(˜ x x b)} ˜ = θM (b) + (1 − θ)M (b). Figure 5. M is stable at ˆ if M does not increase inﬁnitely steeply in a neighborhood of ˆ See b b. λ is a supergradient at ˆ b Figure 5. ♦ Deﬁnition: Let X ⊂ Rn and let g : X → R {∞. NONLINEAR PROGRAMMING Since (5. b) b (In words. Deﬁne the subset A ⊂ R1+m by . A vector λ ∈ Rn is said to be a supergradient (subgradient) of g at x ∈ X if ˆ g(x) ≤ g(ˆ) + λ (x − x) for x ∈ X. b ˆ b M is not stable at ˆ b .) A more geometric way of thinking about subgradients is the following.3: Illustration of supergradient of stability. x ˆ (g(x) ≥ g(ˆ) + λ (x − x) for x ∈ X. −∞}. b M (b)  ˆ b M is stable at ˆ b M (ˆ + λ (b − ˆ b) b)  .) M (b) M (b) M (ˆ b) b) .3.  ˆ b   b .) x ˆ (See Figure 53.24) holds for all x ∈ Ω(b) and x ∈ Ω(˜ it follows that ˜ b) M (θb + (1 − θ)ˆ ≥ sup {f0 (θx + (1 − θ)˜)x ∈ Ω(b).60 CHAPTER 5.
hence again by Exercise 3 ˆ M (ˆ − λ ˆ ≥ f0 (x) − λ f (x) . b) ˆ b ˆ Since f0 (ˆ) = M (ˆ and λ (f (ˆ) − ˆ = 0. ˆ x b. b) b ♦ The next two results give important alternative interpretations of supergradients. A lies below the hyperplane π = {(r.17). f (ˆ) = M (ˆ x ∈ X. b) ˆ b b) ˆ x ˆ ˆ b so that λ (f (ˆ) − ˆ ≥ 0.) The supporting (In words. λm ) is said to be the normal to a hyperplane supporting A at a point(ˆ. M (b) ≤ M (ˆ + λ (b − ˆ b) b) ≤ M (ˆ + λb − ˆ . 1 . M (˜ and (˜. . . (λm /λ0 )) is a supergradient of M at ˆ b. . b) ((λ1 /λ0 . See Figure 5. λ) x ˆ Proof: By hypothesis. Then x b) x (f0 (x).25) λi bi }. b b). or see the Appendix at the end of this Chapter. Lemma 4: (Gale [1967]) M is stable at ˆ iff M has a supergradient at ˆ Proof: (Sufﬁciency only) b b. We will prove only one part of the next crucial result. and r ≤ M (b)} . and f (ˆ) ≤ ˆ Let λ be a supergradient of M at ˆ x b). . . . .4.23). . ˜ ≥ ˆ and r ≤ M (ˆ then ˜ ∈ B. . λi ≥ 0 for 1 ≤ i ≤ m. . ˆ ˆ Lemma 5: Suppose that x is optimal for (5. λ1 . . . λ ≥ 0 and ˆ By Exercise 2. and then (ˆ. b)λ0 r+ λi bi = λ0 r + ˆ ˆ hyperplane is said to be nonvertical if λ0 = 0. (M (b). −λ1 . we can rewrite the inequality above as x b). −λm ) deﬁnes a nonvertical hyperplane b supporting A at (M (ˆ ˆ (ii) if (λ0 . . The reader who is familiar with the Separation Theorem of convex sets should be able to construct a proof for the second part based on Figure 5. ˆ if r b) m m λ0 r + ˆ i=1 b λiˆi ≥ λ0 r + λi bi for all (r.4. b). . . ˜ ∈ A. Next let x ∈ X. Deﬁnition: A vector (λ0 . −λm ) deﬁnes a hyperplane supporting A at b). But then λ (ˆ − f (ˆ)) = 0.2. . and M (ˆ < ∞. DUALITY THEORY A = {(r. . ˆ f (ˆ)) ∈ A and by Exercise 3. b. and (1. Then λ is a supergradient of M at ˆ iff λ is an ˆ b ˆ satisfy the optimality conditions. futhermore. (M (ˆ ˆ then λ0 ≥ 0. Exercise 2: Show that if ˆ ∈ B.5. b b b. ˜ b). This neologism contrasts with the epigraph of a function which is the set lying above the graph of the function. if the hyperplane is nonvertical then b). optimal price vector. . b)b ∈ B. x M (ˆ − λ ˆ ≥ M (ˆ − λ f (ˆ) . 61 Thus A is the set lying ”below” the graph of M . Show that (i) if λ = (λ1 . We call A the hypograph1 of M . ˆ then Let λ be a supergradient at b. λm ) is a Exercise 3: Assume that b b) supergradient of M at ˆ then λ ≥ 0. −λ1 . . i=1 (5. giving (5. Since M is concave it follows immediately that A is convex (in fact these are equivalent statements). r b) ˆ ∈ B. . x b) From the Greek “hypo” meaning below or under. f (x)) ∈ A. b) ∈ A .
♦ ˆ ˆ Lemma 6: Suppose that ˆ ∈ B. x x b) b) so that (5. λ) satisfy the optimality conditions. ˆ be a supergradient of M at ˆ Let x ∈ X. b) ˆ b) Hence M (b) = sup{f0 (x)x ∈ Ω(b)} ≤ M (ˆ + λ (b − ˆ . . suppose x ∈ X.23). . λ ≥ 0 satisfy (5.21). (5. A ˆ b b π is a nonvertical hyperplane supporting A at (M (ˆ ˆ ˆ b). ˆ ˆ f0 (ˆ) + λ (f (ˆ) − ˆ ≥ f0 (x) − λ (f (x) − ˆ .4: Hypograph and supporting hyperplane. Let x ∈ Ω(b). and M (ˆ < ∞.23) .e.. x ˆ ˆ Conversely.21) by (5. Then λ is a supergradient of M at ˆ iff λ is an b b) b ˆ optimal solution of the dual (5. f (x) ≤ b.19) and m(λ) = M (ˆ b).21) holds. It follows that (ˆ. . Then λ ˆ f0 (x) ≤ f0 (x) λ (f (x) − b) ˆ = f0 (x) − λ (f (x) − ˆ + λ (b − ˆ b) ˆ b) ˆ (f (ˆ) − ˆ + λ (b − ˆ ˆ ≤ f0 (ˆ) − λ x x b) b) ˆ (b − ˆ = f0 (ˆ) + λ x b) = M (ˆ + λ (b − ˆ .62 M (b) CHAPTER 5. and (5. A b ˆ b No nonvertical hyperplane supporting A at (M (ˆ ˆ b). b) Figure 5. . b) M (b) π ˆ M (ˆ b) (λ0 . i. by (5.22). By Exercises 2 and 3 Proof: Let λ b. ˆ ˆ (f (x) − b) ≤ 0 so that x ∈ X. b) ˆ b) ˆ so that λ is a supergradient of M at ˆ b. λm ) . NONLINEAR PROGRAMMING M (ˆ b) .
and (5. then λ (f (x) − b) ≤ 0. b) ˆ ≥ 0. if λ is an optimal solution to the dual then (∂M + /∂bi )(ˆ ≤ λi ≤ (∂M − /∂bi )(ˆ b) ˆ b). ˆ ˆ (ii) λ is optimal for the dual iff λ is a supergradient of M at ˆ b. It is easy to see using convexity properties that. M (ˆ < ∞. The “if” part of (iii) follows from Theorem 1. (ii) is implied by Lemma 6. b) b) so that ˆ ˆ M (ˆ ≥ sup{f0 (x) − λ (f (x) − ˆ ∈ X} = m(λ) .2. b) ˆ b) 63 ˆ so that λ is a supergradient.3. b However. the various conditions of Section 1.3 imply stability.21).23) are equivalent to the KuhnTucker condition (5.21).5. ˆ b. and (5.2 Interpretation and extensions. by Lemmas 4.22).2. if X = Rn and fi . ˆ ˆ (i) there exists an optimal solution λ for the dual. and m(λ) = M (ˆ b).) 5. then M is stable at ˆ b. so that ˆ M (ˆ ≥ f0 (x) − λ (f (x) − ˆ + λ (f (x) − b) b) b) ˆ ˆ ˆb = f0 (x) − λ b + λ ˆ for x ∈ Ω(b) . 6 stability is equivalent to the existence of optimal dual variables. Theorem 2: (Duality) Suppose ˆ ∈ B.8). (5. n and the f are differentiable. whereas CQ is only a sufﬁcient condition.2. then the optimality conditions (5. λ) satisfy the (iii) if λ ˆ x ˆ optimality conditions of (5. and M is stable at ˆ Then b b) b. ♦ ˆ Corollary 1: Under the hypothesis of Theorem 2. DUALITY THEORY ˆ M (ˆ − λ ˆ ≥ f0 (x) − λ f (x) b) ˆ b or ˆ M (ˆ ≥ f0 (x) − λ (f (x) − ˆ . M (b) = sup{f0 (x)x ∈ Ω(b)} ≤ M (ˆ + λ (b − ˆ . Proof: (i) follows from Lemmas 4.6. (Hint: See Theorem 5 of 4. whereas the “only if” part of (iii) follows from Lemma 5. 0 ≤ i ≤ m. Thus the condition of stability of M at ˆ plays a similar role to the constraint qualiﬁcation. ˆ M (ˆ ≥ f0 (x) − λ (f (x) − ˆ .23). and m(λ) = M (ˆ Then for any x ∈ X ˆ Conversely suppose λ b). Here if X = R i we give one sufﬁcient condition which implies stability for the general case. ♦ We can now summarize our results as follows. are differentiable. in particular if there exists x ∈ X such that fi (x) < ˆi for b b 1 ≤ i ≤ m.22). ˆ is any optimal solution for the dual. Lemma 7: If ˆ is in the interior of B. In other words if CQ holds at x then M is stable at ˆ In particular. b) b)x ˆ ˆ By weak duality (Lemma 2) it follows that M (ˆ = m(λ) and λ is optimal for (5. Hence. (5. .19). b) b) ˆ and if moreover f (x) ≤ b. Exercise 4: Prove Corollary 1. then x is optimal for the primal iff (ˆ.
18) we would then have more general problem of the form (5. (5. b). . b as the vector of current resource supplies. see Figure 5. we can think of x as the vector of n activity levels. A ˆ b b Figure 5.3 or Figure 5.64 CHAPTER 5. Thus. if its hypograph A is not convex.5: If M is not concave there may be no supporting hyperplane at (M (ˆ ˆ b). Referring to Figure 5. b). NONLINEAR PROGRAMMING The proof rests on the Separation Theorem for convex sets. 1 ≤ i ≤ m. A much more promising development has recently taken place. M (ˆ < ∞ we can see from Theorem 2 and its Corollary 1 that there exists b b) an equilibrium price system iff (∂M + /∂bi )(ˆ < ∞. and comparing with Figure 5. The various convexity conditions are generalizations of the economic hypothesis of nonincreasing returnstoscale. For details see the b) b Appendix. The basic idea involved is to consider supporting A at (M (ˆ ˆ by (nonvertical) surfaces π more genb). Much of duality theory can be given an economic interpretation similar to that in Section 4. λm ) .18).5 it is evident that if M is not concave or.2. .26): Maximize f0 (x) − F (f (x) − ˆ b) subject to x ∈ X . if the current ˆ resources can be bought or sold at prices λ = (λ1 . .4. (5. It is possible to obtain the duality theorem under conditions slightly weaker than convexity but since these conditions are not easily veriﬁable we do not pursue this direction any further (see Luenberger [1968]). if we interpret (∂M + /∂bi )(ˆ b) b) as the marginal revenue of the ith resource. Assuming the realistic condition ˆ ∈ B.17) also is an optimal solution for (5. and ﬁnally f (x) the amount of these resources used up at activity levels x. M (b) M (ˆ b) . there may be no hyperplane supporting A at (M (ˆ ˆ This is the reason why duality theory requires the often restrictive convexity hypothb). Next. we can say that equilibrium prices exist iff marginal productivities of every (variable) resource is ﬁnite.17) is the shortterm decision problem faced by the ﬁrm. the ﬁrm faces the decision problem ˆ an optimal solution of (5.6. and only depends on the fact that M is concave. These ideas are developed in (Gale [1967]).4. ˆ then we can interpret λ as a system of equilibrium prices just as in 4. X as constraints due to physical or longterm limitations. . esis on X and fi . f0 (x) the corresponding revenue. equivalently. M (ˆ < ∞ without loss of generality. b) ˆ eral than hyperplanes. and ˆ is the interior of B. If for a price system λ. The primal problem (5.18).26) . Instead of (5.
The economic interpretation of (5. ·) then the resources f (x) − ˆ can be bought (or sold) for the amount φ(µ. M (b) π ˆ (5.6: The surface π supports A at (M (ˆ ˆ ˆ b).1 see (Geoffrion [1970a]) and for a mathematically more elegant treatment see (Rockafellar [1970]). but quite interesting development along these lines is b. of course. Decentralized resource allocation.27) would be that if the prevailing (nonuniform) price system is φ(µ. f (x) − ˆ b) subject to x ∈ X .27) M (ˆ b) . DUALITY THEORY 65 where F : Rm → R is chosen so that π (in Figure 5.5. . f (x) − ˆ ∈ X} . µk ) ≥ 0. For noneconomic applications. (Banerjee [1971]). . The following references are pertinent: (Gould [1969]).6) is the graph of the function b → M (ˆ − ˆ b) F (b − ˆ Usually F is chosen from a class of functions φ parameterized by µ = (µ1 . in analogy with (5. (Greenberg and Pierskalla [1970]). b)x then the dual problem is Minimize ψ(µ) subject to µ ≥ 0 . If we let ψ(µ) =sup{f0 (x) − φ(µ. Also see (Arrow and Hurwicz [1960]). presented in (Frank [1969]). In particular ˆ from Theorem 2 (iii). .2. b) ≥ φ(µ. f (x) − ˆ For b b). b) ≥ 0 for b ≥ 0.27) instead of (5. For more details concerning the topics of 2. and φ(µ.19). b). such an interpretation to make sense we should have φ(µ. A ˆ b b Figure 5. . Then for each ﬁxed µ ≥ 0 we have (5.18) for λ = λ . if we have an optimal dual solution λ then the optimal primal solutions are ˆ which also satisfy the feasibility condition (5. no such limitation on φ is necessary.2. ˜ b) whenever b ≥ ˜ A relatively unnoticed. 5.26): Maximize f0 (x) − φ(µ.22) and those optimal solutions of (5. b). Parts (i) and (iii) of Theorem 2 make duality theory attractive for computation purposes.3 Applications.
f i (xi ) ≤ ˆ . and m(λ) = i=1 (5. but perhaps more importantly the decision problems in (5. Consider a decision problem in a large system (e. are strictly concave.28) because. subject to x i k i If we let mi (λ) = sup{f0 (xi ) − λ f i (xi )xi ∈ X i }. 1 ≤ i ≤ k. The subsystem has individual constraints of the form xi ∈ X i where xi is a convex set. 1 ≤ i ≤ k . . in fact. the problem corresponding to (5. .g. for each λ ≥ 0 there is a unique optimal solution of (5. NONLINEAR PROGRAMMING the “complementary slackness” condition (5.29) may be much easier to solve than (5. if k is very large it may be practically impossible to solve (5.29) are decentralized whereas in (5. which decomposes into k separate problems: i Maximize f0 (xi ) − λ f i (xi ) i ∈X . a multidivisional ﬁrm). The system is made up of k subsystems (divisions).19) is k i Maximize f0 (xi ) − λ f i (xi ) − λ ( i=1 f i (xi ) − ˆ b) subject to xi ∈ X i .28): k Maximize i=1 k i=1 i f0 (xi ) subject to xi ∈ X i . and also suppose that (5. This is useful because generally speaking (5.. Then by Exercise 8 of Section 1.17) since (5. we need to ﬁnd an optimal dual solution so that we can use Theorem 2(iii).28) all the decision variables x1 . b (5. . + f0 (xk ) where f0 : Rni → R are concave functions.28) has an optimal solution and the stability condition is satisﬁed. Assuming that (5. Consider the following algorithm.29) may be trivial if the dimensions of xi are small.28) For λ ∈ Rm . λ ≥ 0. (5. say xi (λ).66 CHAPTER 5. .29) has an optimal solution for every λ ≥ 0. . .28) whereas (5.e. (5. Suppose that the objective function of the large system 1 k i is additive. 1 ≤ i ≤ k. . the subsystems share some resources in common and this limitation is expressed as f 1 (x1 ) + . .29) b. and the decision variable of the ith subsystem is a vector xi ∈ Rni . . xk are coupled together. For simplicity suppose that the i f0 .18) has fewer constraints. + f k (xk ) ≤ ˆ where f i : Rni → Rm are convex functions and ˆ ∈ Rm b b is the vector of available common resources. Furthermore. subject to λ ≥ 0 .30) Note that (5. i. Thus we have the decision problem (5.18) is easier to solve than (5.23). 1 ≤ i ≤ k. 1≤i≤k . it is the form f0 (x1 ) + . mi (λ) + λ ˆ then the dual problem is Minimize m(λ) .29) involves fewer constraints. ﬁrst of all.29).
xp is feasible for (5.31) and (5. et al. See (Dorfman and Jacoby [1970]. [1971]). Control of water quality in a stream.32): zi (t + 1) − zi (t) = −αi zi (t) + ψi−1 zi−1 (t) vi − ψi zi (t) vi + si (t)mi (t) vi . (5. Select λ0 ≥ 0 arbitrary. . (5. DUALITY THEORY 67 Step 1. Set p = 0. Step 3. xk (λp )).) Figure 5.2. Solve (5. N. .. . si (t) = concentration of BOD of efﬂuent discharge in mg/liter. but for simplicity of exposition we assume that their impact on the quality of the stream is measured in terms of a single quantity.28) see (Geoffrion [1970b]). t = 1. For more detail see (Arrow and Hurwicz [1960]). . it is enough to study the problem over a 24hour period. namely the biochemical oxygen demand (BOD) which they place on the dissolved oxygen (DO) in the stream. t = 1. k Compute ep = i=1 f i (xi (λp )) − ˆ If ep ≥ 0. The pollutants consist of various materials. . optimal. the quality of the stream improves with the amount of DO and decreases with increasing BOD. it is important to treat the efﬂuents before they enter the stream in order to reduce the BOD to concentration levels which can be safely absorbed by the DO in the stream. Step 2. Hence. . T . For an informal discussion of schemes of pollution control which derive their effectiveness from duality theory see (Solow [1971]).31) s qi (t + 1) − qi (t) = βi (qi − qi (t)) + ψi−1 qi−1 (t) − ψi qi (t) vi vi +αi zi (t) − ηi vi .28) and can easily be seen to be b. indeed. . The discussion in this section is mainly based on (Kendrick. and mi (t) = amount of efﬂuent discharge in liters.7 is a schematic diagram of a part of a stream into which n sources (industries and municipalities) discharge polluting efﬂuents. We ﬁrst derive the equations which govern the evolution in time of BOD and DO in the n areas of the streams. . . . . qi (t) = concentration of DO measured in mg/liter. . Set p = p + 1 and return to Step 3. T and i = 1. Set λp=1 according to λp+1 = i if ep ≥ 0 λp i i p p ep if ep < 0 λi − d i i where dp > 0 is chosen a priori. During interval t and in area i let zi (t) = concentration of BOD measured in mg/liter. . It can be shown that if the step sizes dp are chosen properly. xp will converge to the optimum solution of (5. . .29) for λ = λp and obtain the optimal solution xp = (x1 (λp ).5. and for other decentralization schemes for solving (5.” Therefore. The ﬂuctuations of BOD and DO will be cyclical with a period of 24 hours. It is a welladvertized fact that if the DO drops below a certain concentration.28). Since the DO in the stream is used to breakdown chemically the pollutants into harmless substances. We divide this period into T intervals. then life in the stream is seriously threatened. the stream can “die.32) . . The principle of conservation of mass gives us equations (5. and go to Step 2. In this example we are concerned with ﬁnding the optimal balance between costs of waste treatment and costs of high BOD in the stream.
q s are parameters of the stream and are assumed known. They may vary with the time interval t. zi−1 qi−1 .. . (5. qi (1).68 CHAPTER 5. The costs associated with increased amounts of BOD and reduced amounts of DO are much more difﬁcult to quantify since the stream is used by many institutions for a variety of purposes (e. Finally. q0 (t) which are the concentrations immediately upstream from area 1 are assumed known. vi = volume of water in area i measured in liters. instead of attempting to quantify these costs let us suppose that some minimum water quality standards are set. We further assume that f is convex. Hence. Then we face the ¯ . αi is the rate of decay of BOD per interval. NONLINEAR PROGRAMMING direction of ﬂow 0 z0 q0 (1 − π1 )si given 1 z1 q1 1 − πi−1 i−1 i zi qi i+1 zi+1 qi+1 N N +1 . Here. Finally. Also z0 (t). Therefore. Now suppose that the waste treatment facility in area i removes in interval t a fraction πi (t) of the concentration si (t) of BOD.31) is replaced by zi (t + 1) − zi (t) = −αi zi (t) + ψi zi−1 vi − ψi zi (t) vi + (1−πi (t))si (t)mi (t) vi . mi (t)) where the function must be monotonically increasing in all of its arguments. αi . . recreational). .7: Schematic of stream with efﬂuent discharges.33) We now turn to the costs associated with waste treatment and pollution. industrial. .g.. In period t the ith facility treats mi (t) liters of efﬂuent with a BOD concentration si (t) mg/liter of which the facility removes a fraction πi (t). The vi . and the disutility caused by a decrease in the water quality varies with the user. ψi = volume of water which ﬂows from area i to are i + 1 in each period measured in liters. The cost of waste treatment can be readily identiﬁed. . zN qN (1 (1 − πi )si − πi+1 )si+1 (1 − πN )sN si si−a si si+1 sN Figure 5. ηi is the DO requirement in the bottom sludge. municipal... Then (5. βi is the rate of generation of DO. N are assumed known. The increase in DO is due to various natural oxygenproducing biochemical reactions in the stream and the increase is proportional to (q s − qi ) where q s is the saturation level of DO in the stream. i = 1. the initial concentrations zi (1). This decay occurs by combination of BOD and DO.. .. the cost in period t will be fi (πi (t). ψi . agricultural. ηi . si (t). Let q be the minimum acceptable DO concentration and let z be the maximum permissible BOD concentration.
2. . . . N . N. t = 1. and an optimal solution By the duality theorem there exists a 2N T dimensional vector λ ∗ πi (t). . The question we now pose is whether there exist tax rates such that if each individual polluter minimizes its own total cost (i. . w(t)). i = 1. .36) and. ¯ 0 ≤ πi (t) ≤ 1 . . and we write the components of p∗ as p∗ (t) to match i . si (t). furthermore. On the other hand. ¯ ¯ ∗ ≥ 0. t = 1. . .32). (5. then the individual polluters may not (and usually do not) have any incentive to cooperate among themselves to achieve these standards. Suppose that all the treatment facilities are in the control of a single public agency. . the resulting amount of waste treatment is carried out at the minimum expenditure of resources (i.36) and (5. and r is the NTdimensional vector with components (1 − πi (t))si (t)mi (t). . zi (t) ≤ z .. . si (t). (5. Then assuming that the agency is required to maintain the standards (q.33). . If we let p∗ = A λ∗ ≥ 0. .34) subject to (5. T. But if ¯ there is no such centralized agency. t = 1. . . Let the minimum cost be m(q. i = 1. . it does not make sense to enforce legally a minimum standard qi (t) ≥ q. . Furthermore.. . DUALITY THEORY following NP: N T 69 Maximize − i=1 t=1 fi (πi (t). N .37) are equal. i = 1. ¯ 0 ≤ πi (t) ≤ 1 . . t = 1. wN (t)). and let w = (w(1). mi (t)) − λ∗ (b + Ar − w) subject to 0 ≤ πi (t) ≤ 1. . . where the 2N T dimensional vector w has its components equal to −q or z in the obvious manner. si (t). cost of waste treatment + tax on remaining pollutants). . . .34) and arrive at an optimal solution. it may be economically and politically acceptable to tax individual polluters in proportion to the amount of pollutants discharged by the individual.34) as follows: Maximize − i t fi (πi (t). .32) and (5. . . t = 1. . i = 1. To see this let wi (t) = (zi (t). . . .35) we can rewrite (5. z ) and it does this at a minimum cost it will ¯ solve the NP (5. . mi (t)) (5. let w(t) = (w1 (t). . . . . .34)). t = 1. . T . T. . T . . . T. . Note that the coefﬁcients of the matrix must be nonnegative because an increase in any component of r cannot decrease the BOD levels and cannot increase the DO levels. furthermore. −qi (t)) . (5. . N . .36) subject to b + Ar ≤ w . . of the problem: Maximize − i t fi (πi (t). then the resulting water quality will be acceptable and. . Then we can solve (5. i = 1.e. . . . the optimal values of (5. It should be clear from the duality theory that the answer is in the afﬁrmative.e.35) where the matrix A and the vector b depend upon the known parameters and initial conditions. . T. N . zi (t) ≤ z on every polluter since the ¯ pollution levels in the ith area depend upon the pollution levels on all the other areas lying upstream.5. and −qi (t) ≤ −q . Using (5.37) ∗ such that {πi (t)} is also an optimal solution of (5. . . z ). mi (t)) (5. N .33) for w and obtain w = b + Ar . . will be an optimal solution of (5. . i = 1. .
43) Suppose we try to solve (5. .3. hence the necessity of these conditions follows from Theorem 2 of 1. i Before we leave this example let us note that the optimum dual variable or shadow price λ∗ plays an important role in a larger framework. (5.3 Quadratic Programming An important special case of NP is the quadratic programming (QP) problem: Maximize c x − 1 x P x 2 subject to Ax ≤ b. z ) ¯ was somewhat arbitrary.41) and (5.2. ♦ From (5. µ ≥ 0 .2. µ∗ ) to (5. A is a ﬁxed m × n matrix and P = P is a ﬁxed positive semideﬁnite matrix. This estimate can now serve as a basis in making a beneﬁts/cost i i analysis of the proposed new standard.2). NONLINEAR PROGRAMMING with the components (1 − πi (t))si (t)mi (t) of r we can see that (5. si (t). t = 1. If the corresponding components of λ∗ are λq∗ (t) and λz∗ (t).43): Ax + Im Y = b −P x − A λ + In µ = −c . λ ≥ 0.3. (5. since P is positive semideﬁnite it follows from Exercise 6 of Section 1. . . (λ∗ ) (Ax∗ − b) = 0 . Then we must apply Phase II to the LP: m n Maximize − i=1 zi − j=1 ξj . Theorem 1: A vector x∗ ∈ Rn is optimal for (5. and (5.39) iff there is a solution (x∗ . . µ∗ ∈ Rn .38) Thus. CQ is satisﬁed.70 CHAPTER 5. b ∈ Rm are ﬁxed. We noted earlier that the quality standard (q. (µ∗ ) x∗ = 0 .41) (5. (5. such that Ax∗ ≤ b. Now suppose it is proposed to change the standard in the ith area during period t to q + ∆qi (t) and z + ∆zi (t).42) by Phase I of the Simplex algorithm (see 4. λ∗ ≥ 0.37) is equivalent to the set of N T problems: Maximize − fi (πi (t). 5. .39) where x ∈ Rn is the decision variable . . so that the sufﬁciency of these conditions follows from Theorem 4 of 1.40) Proof: By Lemma 3 of 1.42) (5. µ x = 0 .42). µ∗ ≥ 0 . x∗ ≥ 0 c − P x∗ = A λ∗ − µ∗ . λ∗ . x ≥ 0 y ≥ 0. x ≥ 0 . mi (t)) − p∗ (t)(1 − πi (t))si (t)mi (t) i 0 ≤ πi (t) ≤ 1 . y ∗ . i = 1. ¯ i i then the change in the minimum cost necessary to achieve the new standard will be approximately λq∗ (t)∆qi (t) + λz∗ (t)∆zi (t). .2 that f0 : x → c x − 1/2 x P x is a concave function. T . c ∈ Rn . p∗ (t) is optimum tax per mg of BOD in area i during period t. On the other hand.41).λ y = 0 . (5. N . (5.40) we can see that x∗ is optimal for (5. .39) iff there exist λ∗ ∈ Rm .
41). (5. Step 2 of the Simplex algorithm must be modiﬁed as follows to satisfy (5. We have the following result.43) have a solution. λ ≥ 0. and Polak [1970].42). and (5.41). . and (5. λ.43).44). (5. Furthermore. y . then there is an optimal basic feasible solution of (5. Cullum.42). ξ = 0 is ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ an optimal solution of (5.4. If z = 0 or ξ ˆ ˆ solution to (5.42). ♦ This lemma suggests that we can apply the Simplex algorithm of 4.4 Computational Method We return to the general NP (5. stop.39).43). ˆ Proof: Let x.45).43) and x is an optimal solution of (5. µ. y . are differentiable. Maximize f0 (x) subject to fi (x) ≤ 0. For x ∈ Ω deﬁne the function ψ(ˆ) : Rn → R by ˆ x ψ(ˆ)(h) = max{−f0x (ˆ)h. z = 0.41). p. ξ ≥ 0. from (5.44) which is also a solution f (5. µ) are nonzero.41) and (5.43).42) have a solution then the maximum value in (5. λ. . . x x −ψ(ˆ)(h) + fi (ˆ)fix h ≤ 0 .46) . COMPUTATIONAL METHOD subject to Ax + Im y +z =b −P x − A λ + In µ + ξ = −c x ≥ 0. λ.1 will x ˆ ˆ ˆ also prove this lemma. However. and there is no feasible solution of (5. and (5. 5.42). µ ≥ 0. do not consider µj as a candidate for entry into the basis. z . and (5.41). 159 ff).44). ξ) of (5.3.42). and (5. (We have assumed. that b ≥ 0 and −c ≥ 0. λ.2 to solve (5. y . (5. without loss of generality.3. (5. y .45) where x ∈ Rn . z ≥ 0.41). . starting with the basic feasible solution z = b. (5. in order to obtain a solution of (5.41). 71 (5. x x 1 ≤ i ≤ m . f1 (ˆ) + f1x (ˆ)h.42) and (5. If z = 0 and ξ = 0 then (ˆ.43) we see that at most (n + m) components of (ˆ.44) starting with a basic feasible solution z = b.39). . µ be a solution of (5. ξ = −c. The algorithm will stop in a ﬁnite number of steps at an ˆ optimal basic feasible solution (ˆ. For a proof of this result as well as for a generalization of the algorithm which permits positive semideﬁnite P see (Cannon. y . do not consider λi as a candidate for entry into the basis. Let Ω ⊂ Rn denote the set of feasible solutions. But then a repetition of the proof of Lemma 1 of 4. ξ = −c. µ. then there is no (5. (5. fi : Rn → R. (5. µ) solve x ˆ ˆ ˆ ˆ ˆ ˆ x ˆ ˆ ˆ ˆ = 0.43). λ. The above algorithm is due to Wolfe [1959]. y ≥ 0. . Theorem 2: Suppose P is positive deﬁnite. −1 ≤ hj ≤ 1 . The behavior of the algorithm is summarized below. if a variable yi is currently in the basis.43): If a variable xj is currently in the basis. (5.44) is 0. (5. 0 ≤ i ≤ m. m .) If (5. x x x x x x Consider the problem: Minimize ψ(ˆ)(h) x subject to − ψ(ˆ)(h) − f0x (ˆ)h ≤ 0 . Lemma 1: If (5.5. Then x.42). . . If it not possible to remove the zi and ξj from the basis. i = 1. 1 ≤ j ≤ n . fm (ˆ) + fmx (ˆ)h}.44).
f0 (x) ≥ f0 (x0 )} is compact. and has a nonempty interior.8. Theorem 1: Suppose that the set Ω(x0 ) = {xx ∈ Ω. . Let x∗ be any limit point of the sequence x0 . Compute an optimum solution µ(xk ) to the onedimensional problem. generated by the algorithm. set k = k + 1 and return to Step 2. Step 2. h(xk ). If h0 (xk ) = 0. and go to Step 2. x1 . (Note that by Exercise 1 of 4. Find x0 ∈ Ω. xk .46) can be solved as an LP. Maximize f0 (xk + µh(xk )) . . which is dense in Ω(x0 ). otherwise go to Step ˆ 3. stop. subject to (xk + µh(xk )) ∈ Ω.1 (5. 1 ≤ i ≤ m.72 CHAPTER 5. For this reason h(xk ) is called a (desirable) feasible direction.5. µ ≥ 0 .) . .46) and let h0 (ˆ) = ψ(ˆ)(h(ˆ)) be the minimum value atx x x x tained. f3 (xk ) xk h(xk ) f2 (xk ) f1 = 0 Ω f3 = 0 Figure 5. Solve (5. and go to Step 4. Remark: If h0 (xk ) < 0 in Step 2. NONLINEAR PROGRAMMING f0 (xk )f0 (x) = F0 (x∗ ) > f0 (xk ) f1 (xk ) f0 (x) = f0 (xk ) f2 = 0 . Step 3. Then the KuhnTucker conditions are satisﬁed at x∗ . For a proof of this result and for more efﬁcient algorithms the reader is referred to (Polak [1971]). Call h(ˆ) an optimum solution of (5.) The following algorithm is due to Topkis and Veinott [1967].46) for x = xk and obtain h0 (xk ).8: h(xk ) is a feasible direction. Step 1. . . Set xk+1 = xk + µ(xk )h(xk ). . then the direction h(xk ) satisﬁes f0x (xk )h(xk ) > 0. and fi (xk )+ fix (xK )h(xk ) < 0. The performance of the algorithm is summarized below. . set k = 0. . (See Figure 5. Step 4.
and θ ∈ R such that λ g ≤ θ for all g ∈ G λ f ≥ θ for all f ∈ F . . b) (5. b) ∈ F .48) m b λiˆi ≤ θ. b)b ∈ B. . . . and (5. λ0 ♦ Proof of Lemma 7: Since ˆ is in the interior of B. it can be veriﬁed that (5. Also from (5. there exists ε > 0 such that b b ∈ B whenever b − ˆ < ε . .49) λ0 r + i=1 b λiˆi ≥ θ . Hence.5. But then from (5.48) λi bi ≥ θ for (r. λm ) = 0.5 Appendix The proofs of Lemmas 4. . . .5. APPENDIX 73 5. b) ∈ G . b)} It is easy to check that F.47) implies that F ∩ G = φ. and θ such that m (5. .48) we get m 1 λ0 [θ m M (b) − M (ˆ ≤ b) − i=1 λ i bi ] = i=1 (− λi )(bi − ˆ b). b)b ∈ B. Then there exists λ ∈ Rn . r ≤ M (b) − M (ˆ . λm ) = 0.49) we can see that m i=1 b λiˆi ≥ θ.49) can hold m only if λ0 > 0.50) . λm ) = 0. Let F. so that there exist (λ0 . for r > M (ˆ . G are disjoint. and the fact that (λ0 . and θ such that m (5. r > Kb − ˆ . there exist (λ0 . i=1 so that i=1 b λiˆi = θ. F. .47) λ0 r + i=1 m λi bi ≤ θ for (r. G are convex. λ = 0. Evidently. (5. . b} G = {(r. whereas from (5.7 of Section 2 are based on the following extremely important theorem (see Rockafeller [1970]). b)b ∈ Rm . G be convex subsets of Rn such that the relative interiors of F. r ≤ M (b)} . Proof of Lemma 4: Since M is stable at ˆ there exists K such that b M (b) − M (ˆ ≤ Kb − ˆ for all b ∈ B . G are convex and F ∩ G = φ. ˆ > M (ˆ b)r b} G = {(r. Separation theorem for convex sets. b) b In R1+m consider the sets F = {(r. b In R1+m consider the sets F = {(r. . i=1 λ0 r + From the deﬁnition of F .
b) ∈ G .74 m CHAPTER 5. (5.51) we get λ0 M (ˆ + b) so that (5. b λ0 ♦ . From (5. and the fact that (λ0 . λm ) = 0 we can see that (5.50).51) From (5. . .50) and (5. for (r. (− i=1 λi )(bi − ˆi ) .51) imply λ0 > 0.52) implies M (b) ≤ (ˆ + b) m m i=1 b λiˆi = θ . .49). NONLINEAR PROGRAMMING λ0 r + i=1 b λiˆi ≤ θ . .(5.
and thrust force CT x3 . At a prescribed ﬁnal time tf .1 Examples The trajectory of a vertical sounding rocket is controlled by adjusting the rate of fuel ejection which generates the thrust force. The “dot” denotes differentiation with respect to t. drag ¨ ˙ force = CD ρ(x1 )x2 where CD is constant. x3 (t) is the weight of the rocket (= weight of remaining fuel) at time t. Speciﬁcally suppose that the equations of motion are given by (6. 6. gravitational force = gx3 with g assumed constant. See Figure 6. it is desired that the rocket be at a position as high above the ground as possible. The decision variable at ˙ time t is u(t). the 75 . x1 (t) = x2 (t) ˙ x2 (t) = − xCD ρ(x1 (t))x2 (t) − g + ˙ 2 3 (t) x3 (t) = −u(t) . Thus. x2 (0). 0. At time 0 we assume that (x1 (0). x3 (0)) = (0. that is. the rate of fuel ejection. M ). ρ(x1 ) is a friction coefﬁcient depending on atmospheric 2 density which is a function of x1 . with initial fuel of weight M . ˙ CT x3 (t) u(t) (6. and in Section 2 we derive the main result.1. at rest. namely: inertia = x3 x1 = x3 x2 .1). assumed proportional to rate of fuel ejection. x2 (t) is the (vertical) speed at time t. In the ﬁrst section we give two examples.1) where x1 (t) is the height of the rocket from the ground at time t. the rocket is on the ground. A very important class of problems where such situations arise is in the control of dynamical systems.Chapter 6 SEQUENTIAL DECISION PROBLEMS: DISCRETETIME OPTIMAL CONTROL In this chapter we apply the results of the last two chapters to situations where decisions have to be made sequentially over time. These equations can be derived from the force equations under the assumption that there are four forces acting on the rocket.
76
CHAPTER 6. DISCRETETIME OPTIMAL CONTROL
decision problem can be formalized as (6.2). Maximize x1 (tf ) subject to x(t) = f (x(t), u(t)), 0 ≤ t ≤ tf ˙ x(0) = (0, 0, M ) u(t) ≥ 0, x3 (t) ≥ 0, 0 ≤ t ≤ tf ,
(6.2)
where x = (x1 , x2 , x3 ) , f : R3+1 → R3 is the righthand side of (6.1). The constraint inequalities u(t) ≥ 0 and x3 (t) ≥ 0 are obvious physical constraints.
x3 x1 = inertia ¨
CD ϕ(x1 )x2 = drag 2
gx3 = gravitational force ˙ CR x3 = thrust Figure 6.1: Forces acting on the rocket. The decision problem (6.2) differs from those considered so far in that the decision variables, which are functions u : [0, tf ] → R, cannot be represented as vectors in a ﬁnitedimensional space. We shall treat such problems in great generality in the succeeding chapters. For the moment we assume that for computational or practical reasons it is necessary to approximate or restrict the permissible function u(·) to be constant over the intervals [0, t1 ), [t1 , t2 ), . . . , [tN −1 , tf ), where t1 , t2 , . . . , tN −1 are ﬁxed a priori. But then if we let u(i) be the constant value of u(·) over [ti , ti+1 ), we can reformulate (6.2) as (6.3): Maximize x1 (tN )(tN = tf ) subject to x(ti+1 ) = g(i, x(ti ), u(i)), i = 0, 1, . . . , N − 1 x(t0 ) = x(0) = (0, 0, M ) u(i) ≥ 0, x3 (ti ) ≥ 0, i = 0, 1, . . . , N .
(6.3)
In (6.3) g(i, x(t1 ), u(i)) is the state of the rocket at time ti+1 when it is in state x(ti ) at time ti and u(t) ≡ u(i) for ti ≤ t < ti+1 . As another example consider a simple inventory problem where time enters discretely in a natural fashion. The Squeezme Toothpaste Company wants to plan its production and inventory schedule for the coming month. It is assumed that the demand on the ith day, 0 ≤ i ≤ 30, is d1 (i) for
6.2. MAIN RESULT
77
their orange brand and d2 (i) for their green brand. To meet unexpected demand it is necessary that the inventory stock of either brand should not fall below s > 0. If we let s(i) = (s1 (i), s2 (i)) denote the stock at the beginning of the ith day, and m(i) = (m1 (i), m2 (i)) denote the amounts manufactured on the ith day, then clearly s(i + 1) + s(i) + m(i) − d(i) , where d(i) = (d1 (i), d2 (i)) . Suppose that the initial stock is s, and the cost of storing inventory s ˆ for one day is c(s) whereas the cost of manufacturing amount m is b(m). The the costminimization decision problem can be formalized as (6.4):
30
Maximize
i=0
(c(s(i)) + b(m(i))) (6.4)
subject to s(i + 1) = s(i) + m(i) − d(i), 0 ≤ i ≤ 29 s(0) = s ˆ s(i) ≥ (s, s) , m(i) ≥ 0, 0 ≤ i ≤ 30 .
Before we formulate the general problem let us note that (6.3) and (6.4) are in the form of nonlinear programming problems. The reason for treating these problems separately is because of their practical importance, and because the conditions of optimality take on a special form.
6.2 Main Result
The general problem we consider is of the form (6.5).
N −1
Maximize
i=0
f0 (i, x(i), u(i))
subject to dynamics : x(i + 1) − x(i) = f (i, x(i), u(i)), i = 0, . . . , N − 1 , initial condition: q0 (x(0) ≤ 0, g0 (x(0)) = 0 , ﬁnal condition: qN (x(N )) ≤ 0, gN (x(N )) = 0 , statespace constraint: qi (x(i)) ≤ 0, i = 1, . . . , N − 1 , control constraint: hi (u(i)) ≤ 0, i = 0, . . . , N − 1 .
(6.5)
Here x(i) ∈ Rn , u(i) ∈ Rp , f0 (i, ·, ·) : Rn+p → R, f (i, ·, ·) : Rn+p → Rn , qi : Rn → Rmi , gi : Rn → R i , hi : Rp → Rsi are given differentiable functions. We follow the control theory terminology, and refer to x(i) as the state of the system at time i, and u(i) as the control or input at time i. We use the formulation mentioned in the Remark following Theorem 3 of V.1.2, and construct the Lagrangian function L by L(x(0), . . . , x(N ); u(0), . . . , u(N − 1); p(1), . . . , p(N ); λ0 , . . . , λN ; α0 , αN ; γ 0 , . . . , γ N −1 )
78
N −1 N −1
CHAPTER 6. DISCRETETIME OPTIMAL CONTROL
=
i=0 N i i=0 0
f0 (i, x(i), u(i)) −
i=0
(p(i + 1)) (x(i + 1) − x(i) − f (i, x(i), u(i)))+
N −1 N
(λ ) qi (x(i)) + (α ) g0 (x(0)) + (α ) gN (x(N )) +
i=0
(γ i ) hi (u(i))
.
Suppose that CQ is satisﬁed for (6.5), and x∗ (0), . . . , x∗ (N ); u∗ (0), . . . , u∗ (N − 1), is an optimal solution. Then by Theorem 2 of 5.1.2, there exist p∗ (i) in Rn for 1 ≤ i ≤ N, λi∗ ≥ 0 in Rmi for 0 ≤ i ≤ N, αi∗ in R i for i = 0, N, and γ i∗ ≥ 0 in Rsi for 0 ≤ i ≤ N − 1, such that (A) the derivative of L evaluated at these points vanishes, and (B) λi∗ qi (x∗ (i)) = 0 for 0 ≤ i ≤ N , γ i∗ hi (u∗ (i)) = 0 for 0 ≤ i ≤ N − 1 . We explore condition (A) by taking various partial derivatives. Differentiating L with respect to x(0) gives f0x (0, x∗ (0), u∗ (0)) − {−(p∗ (1)) − (p∗ (1)) [fx (0, x∗ (0), u∗ (0))] +(λ0∗ ) [q0x (x∗ (0))] + (α0∗ ) [g0x (x∗ (0))]} = 0 , or p∗ (0) − p∗ (1) = [fx (0, x∗ (0), u∗ (x))] p∗ (1) +[f0x (0, x∗ (0), u∗ (0))] − [q0x (x∗ (0))] λ0∗ , where we have deﬁned p∗ (0) = [g0x (x∗ (x))] α0∗ . Differentiating L with respect to x(i), 1 ≤ i ≤ N − 1, and rearranging terms gives p∗ (i) − p∗ (i + 1) = [fx (i, x∗ (i), u∗ (i))] p∗ (i + 1) +[f0x (i, x∗ (i), u∗ (i))] − [qix (x∗ (i))] λi∗ . Differentiating L with respect to x(N ) gives, p∗ (N ) = −[gN x (x∗ (N ))] αN∗ − [qN x (x∗ (N ))] λN∗ . It is convenient to replace αN∗ by −αN∗ so that the equation above becomes (6.9) p∗ (N ) = [gN x (x∗ (N ))] αN∗ − [qN x (x∗ (N ))] λN∗ . Differentiating L with respect to u(i), 0 ≤ i ≤ N − 1 gives [f0u (i, x∗ (i), u∗ (i))] + [fu (i, x∗ (i), u∗ (i))] p∗ (i + l) − [hiu (u∗ (i))] γ i∗ = 0 . We summarize our results in a convenient form in Table 6.1 Remark 1: Considerable elegance and mnemonic simpliﬁcation is achieved if we deﬁne the Hamiltonian function H by (6.10) (6.9) (6.8) (6.7)
(6.6)
N − 1 hi (u∗ (i)) ≤ 0 [f0u (i. . u∗ (i)] − [qix (x∗ (i)] γ i∗ transversality conditions: p∗ (0) = [g0x (x∗ (0))] α0∗ λ0∗ ≥ 0. (λ0∗ ) q0 (x∗ (0)) = 0 λN∗ ≥ 0. . . . . . x∗ (i). u(i)) subject to the constraints below dynamics: i = 0. u∗ (i))] + [fu (i. γ 0∗ . p∗ (N ) = [gN x (x∗ (N ))] αN∗ − [qN x (x∗ (N ))] λN∗ (λN∗ ) qN (x∗ (N )) = 0 λi∗ ≥ 0. . N − 1 p∗ (i) − p∗ (i + 1) = [fx (i. . (λi∗ ) qi (x∗ (i)) = 0 γ i∗ ≥ 0 (γ i∗ ) hi (u∗ (i) = 0 79 . . . u∗ (0). λ0∗ . g0 (x∗ (0)) = 0 ﬁnal conditions: qN (x∗ (N )) ≤ 0. α0∗ . . . x∗ (i)u∗ (i))] . u(i)) Table 6. u∗ (i)] p∗ (i + 1) +[f0x (i. such that f0 (i. . . gN (x∗ (N )) = 0 state space constraint: i = 1. x∗ (i). . .2. . x(i). N − 1 qi (x∗ (i)) ≤ 0 control constraint: i = 0. . x∗ (i). x(i). .6. N − 1 x(i + 1) − x(i) = f (i. . . MAIN RESULT Suppose x∗ (0). λN∗ . u∗ (N − 1) maximizes N −1 i=0 then there exist p∗ (N ). . . . p∗ (i1 ) = [hiu (u∗ (i))] γ i∗ adjoint equations: i = 0. .1: initial condition: q0 (x∗ (0)) ≤ 0. αN∗ . . . . . . . . x∗ (N ). γ N −1∗ .
(6.. For this reason the result is sometimes called the maximum principle. Remark 3: If the f0 (i. u∗ (i). u) + p f (i. u∗ (i).10) becomes [hiu (u∗ (i))] γ i∗ = [Hu (i. (6. (6. Furthermore. (6. ·.9) are called transversality conditions for the following reason.12) Since the homogeneous part of the linear difference equations (6.1 are also sufﬁcient. u∗ (i))]δu(i) . u∗ (i))] r(i + 1) . in this case we see from (6. we note that in this case the initial and ﬁnal conditions specify ( 0 + n ) conditions whereas the transversality conditions specify (n − 0 ) + (n − n ) conditions. u. x. we call (6. p∗ (N ) = [gN x (x(N ))] αN∗ which means that p∗ (0) and p∗ (N ) are respectively orthogonal or transversal to the initial and ﬁnal surfaces.9) become respectively p∗ (0) = [g0x (x∗ (0))] α0∗ . which describe surfaces in Rn . (6. x∗ (i). i.11) p∗ (i) − p∗ (i + 1) = [Hx (i.13). u. 0 ≤ i ≤ N − 1 .13) (6. u∗ (i). x∗ (i). u∗ (i). whereas (6. x∗ (i). u∗ (i))]z(i) . p∗ (i + 1))] . qN ≡ 0. which has for it adjoint the system r(i) − r(i + 1) = [fx (i. we have a total of 2n boundary conditions for the 2ndimensional system of difference equations (6. x∗ . and the necessary conditions of Table 6. x. subject to hi (u) ≤ 0 . p∗ (i + 1))] − [qix (x∗ (i))] λi∗ .80 CHAPTER 6.8) is (6. gN (x(N )) = 0. The dynamic equations then become x∗ (i + 1) − x∗ (i) = [Hp (i. x∗ (i). .8) become (6.6) and (6. 0≤i≤N −1 . (6. Suppose q0 ≡ 0. u) . DISCRETETIME OPTIMAL CONTROL H(i. Thus. Remark 2: If we linearize the dynamic equations about the optimal solution we obtain δx(i + 1) − δx(i) = [fx (i. ·) are concave and the remaining function in (6.7). u∗ (i))]δx(i) + [fu (i. Furthermore.5) are linear.13) that u∗ (i) is an optimal solution of Maximize H(i.7). x∗ (i). p∗ (i + 1))] .12).e. x∗ (i. and the p∗ (i) are called adjoint variables. Remark 4: The conditions (6. x∗ (i). 0≤i≤N −1.8) the adjoint equations.6). whose homogeneous part is z(i + 1) − z(i) = [fx (i. then CQ is satisﬁed. some of them refer to the initial time 0 and the rest refer to the ﬁnal time. but note that these 2n boundary conditions are mixed.6).5). x. so that the initial and ﬁnal conditions read g0 (x(0)) = 0. p∗ (i + 1)). Conditions (6. and the adjoint equations (6. p) = f0 (i. x∗ (i). (i).
x(N ) are ﬁxed. j=1 subject to x(i + 1) − x(i) = Ax(i) + Bu(i). A and B are as ˆ ˆ in Exercise 1. A and B are constant matrices. x(N ) = x(N ) . ˆ ˆ u(i) ∈ Rp . . 0 ≤ i ≤ N − 1 . Here x(0). N −1 i=0 Minimize P (u(i))j  . u(i))j  ≤ 1. ˆ and P = P is positive deﬁnite. 1 ≤ j ≤ p. Exercise 2: Show that the minimal fuel problem. MAIN RESULT Exercise 1: For the regulator problem. ˆ u(i) ∈ Rp . 0 ≤ i ≤ N − 1 x(0) = x(0). 0 ≤ i ≤ N − 1 can be transformed into a linear programming problem.2. x(0) is ﬁxed.6. 1 2 N −1 N −1 81 x(i) Qx(i) + where x(i) ∈ Rn . Q = Q is positive semideﬁnite. 0 ≤ i ≤ N − 1 x(0) = x(0). show that the optimal solution is unique and can be obtained by solving a 2ndimensional linear difference equation with mixed boundary conditions. Maximize 1 u(i) P u(i) 2 i=0 i=0 subject to x(i + 1) − x(i) = Ax(i) + Bu(i).
DISCRETETIME OPTIMAL CONTROL .82 CHAPTER 6.
In Section 2 we study more general boundary conditions. x0 ∈ Rn be ﬁxed and let tf ≥ t0 be a ﬁxed time.1 The Linear Optimal Control Problem We consider a dynamical system governed by the linear differential equation (7. The general nonlinear case is deferred to the next chapter. ˙ where x(t) ∈ Rn and u(t) ∈ Rp are respectively the state and control of the system at time t. The control u(·) is constrained to take values in a ﬁxed set Ω ⊂ Rp . ∞) → Ω will be called an admissible control. Let c ∈ Rn . In Section 1 we present the general linear problem and study the case where the initial and ﬁnal conditions are particularly simple.1) Here A(·) and B(·) are n × n. U denotes the set of all admissible controls.and n × pmatrix valued functions of time. A choice of control has to be made at each instant of time t where t varies continuously over a ﬁnite interval. t ≥ t0 .Chapter 7 SEQUENTIAL DECISION PROBLEMS: CONTINUOUSTIME OPTIMAL CONTROL OF LINEAR SYSTEMS We will investigate decision problems similar to those studied in the last chapter with one (mathematically) crucial difference. To understand the main ideas and techniques of analysis it will prove proﬁtable to study the linear case ﬁrst. x(t). We are concerned with the 83 . and to be piecewise continuous. The evolution in time of the state of the systems to be controlled is governed by a differential equation of the form: x(t) = f (t. ˙ (7. we assume that they are piecewise continuous functions.1): x(t) = A(t)x(t) + B(t)u(t). u(t)) . Deﬁnition: A piecewise continuous function u : [t0 . 7.
e. (See Desoer [1970]. u) = Φ(t2 . (ii) Assuming that U is convex show that K(t2 . Deﬁnition: Let Φ(t. Lemma 2: Suppose c = 0. z) is the set of states reachable at time t2 starting at time t1 in state z and using admissible controls. z) is a convex set.1) at time t2 . τ ) = A(t)Φ(t. t1 . and the control u(·) is applied.84 decision problem (7. t1 . t) ≡ In . Let u∗ (·) ∈ U and let x∗ (t) = φ(t. ˙ initial condition: x(t0 ) = x0 . Then u∗ is an optimal solution of (2) iff (i) x∗ (tf ) is on the boundary of K = K(tf . t1 .2). subject to dynamics: x(t) = A(t)x(t) + B(t)u(t) . u∗ ). We call K the reachable set. ﬁnal condition: x(tf ) ∈ Rn . u) denote the state of (7.1). x0 . t1 )z + t2 t1 Φ(t2 . control constraint: u(·) ∈ U . be the transitionmatrix function of the homogeneous part of (7. x0 ). CONTINUOUSTIME LINEAR OPTIMAL CONTROL Maximize c x(tf ). The next result is wellknown. but then . z) = {φ(t2 . t0 .) Deﬁnition: Let K ⊂ Rn . τ ) . t1 . (See Figure 7.. for any z ∈ Rn . and any t0 ≤ t1 ≤ t2 ≤ tf let φ(t2 . and the boundary condition Φ(t. provided we include in U any measurable function u : [t0 .1. The next result gives a geometric characterization of the optimal solutions of (2). t1 . t0 ≤ t ≤ tf . t0 .) Proof: Clearly (i) is implied by (ii) because if x∗ (tf ) is in the interior of K there is δ > 0 such that (x∗ (tf ) + δc) ∈ K. τ )B(τ )u(τ )dτ . t1 . if a time t1 it is in state z. (It is a deep result that K(t2 . show that U is a convex set. Φ satisﬁes the differential equation ∂Φ ∂t (t. and let x∗ ∈ K. Thus. τ ). z. t1 .) Lemma 1: φ(t2 . t0 ≤ τ ≤ t ≤ tf . z. tf ] → Rp . ∞) → Ω. (ii) Let K(t2 . z. (7. z) is convex even if Ω is not convex (see Neustadt [1963]). K(t2 . u)u ∈ U} . and c x∗ ≥ c x for all x ∈ K . CHAPTER 7. and (ii) c is the outward normal to a hyperplane supporting K at x∗ . We say that c is the outward normal to a hyperplane supporting K at x∗ if c = 0. Exercise 1: (i) Assuming that Ω is convex. i.2) Deﬁnition: (i) For any piecewise continuous function u(·) : [t0 .
Finally. Then u∗ (·) is optimal iff (p∗ (t)) B(t)u∗ (t) = sup{(p∗ (t)) B(t)vv ∈ Ω} . t0 )x0 + t0f Φ(tf . x0 . τ )B(τ )u (τ )dτ t ≥ t0f (p∗ (tf )) Φ(tf .5) which is equivalent to (7. t0 . ˙ ﬁnal condition: p∗ (tf ) = c .1: c is the outward normal to π ∗ supporting K at x∗ (tf ) . Let p∗ (t) be the solution of (7.3) and (7. t (7. except possibly for a ﬁnite set. t0 ≤ t ≤ tf . The beauty and utility of the theory lies in the following result which translates this characterization directly in terms of u∗ .3) (7.4): adjoint equation: p∗ (t) = −A (t)p∗ (t) . THE LINEAR OPTIMAL CONTROL PROBLEM x3 c 85 x∗ (tf ) c x2 K π ∗ = {xc x = c x∗ (tf )} x1 Figure 7. tf ∗ ∗ t0 (p (tf )) Φ(tf . Proof: u∗ (·) is optimal iff for every u(·) ∈ U (p∗ (tf )) [Φ(tf . t0 )x0 + t0f Φ(tf . tf ].6) .4) (7. ♦ The result above characterizes the optimal control u∗ in terms of the ﬁnal state x∗ (tf ). τ )B(τ )u(τ )dτ ] . u∗ ).6).1. c (x∗ (tf ) + δc) = c x∗ (tf ) + δc2 > c x∗ (tf ) . t0 ≤ t ≤ tf .7. τ )B(τ )u∗ (τ )dτ ] t ≥ (p∗ (tf )) [Φ(tf . for all t ∈ [t0 . from the deﬁnition of K it follows immediately that u∗ is optimal iff c x∗ (tf ) ≥ c x for all x∈K. τ )B(τ )u(τ )dτ (7. Theorem 1: Let u∗ (·) ∈ U and let x∗ (t) = φ(t.
tf ]. Taking t1 = t0 in (7. p) = sup{H(t. then x∗ (t) is on the boundary of K(t. giving a contradiction. then (7. ♦ But then from (7.7) and the sufﬁciency of (7. Indeed if this is not the case. x∗ (t).5) is immediate.86 CHAPTER 7.. (7. it follows that there exists δ > 0 such that (p∗ (t)) B(t)u∗ (t) < (p∗ (t)) B(t)v. This condition is known as the maximum principle.7). (7. x. the outward normal p∗ (tf ) at time tf . if c = p∗ (tf ) is the outward normal to a hyperplane supporting K(tf . t ∈ [t0 . tf ∗ t0 (p (τ )) B(τ )u∗ (τ )dτ ≥ tf ∗ t0 (p (τ )) B(τ )u(τ )dτ.5) can be rewritten as H(t. u.10) . This normal is obtained by transporting backwards in time.9) is the following. t0 . x0 ) at x∗ (tf ).9) Exercise 2: Prove Corollary 1.7) we see that u∗ (·) cannot be optimal. x. tf ] u∗ (t) otherwise . and since t∗ is a point of continuity of B(·) and u∗ (·). t1 . Corollary 1: For t0 ≤ t1 ≤ t2 ≤ tf . p∗ (t)) = M (t. x∗ (t1 )). i. To prove the necessity let D be the ﬁnite set of points where the function B(·) or u∗ (·) is discontinuous. then there exists t∗ ∈ [t0 . p) = p (A(t)x + B(t)u) . (7. Deﬁne u(·) ∈ U by ˜ u(t) = ˜ Then (7. t0 . (p∗ (t2 ))x∗ (t2 ) ≥ (p∗ (t2 )) x for all x ∈ K(t2 . we see that if u∗ (·) is optimal. t∗ ∈ D.9). x0 ) at x∗ (t). x∗ (t). t) so that (7. CONTINUOUSTIME LINEAR OPTIMAL CONTROL Now by properties of the adjoint equation we know that p∗ (t)) = (p∗ (tf )) Φ(tf . u. We shall show that if u∗ (·) is optimal then (7.2. for t − t∗  < δ . B(t)˜(t)dt > u tf ∗ t0 (p (t)) B(t)u∗ (t)dt . Remark 2: If we deﬁne the Hamiltonian function H by H(t. x0 ) and p∗ (t) is the normal to a hyperplane supporting K(t. and we deﬁne M by M (t.e. via the adjoint differential equation. The situation is illustrated in Figure 7. p∗ (t)) .5) is satisﬁed for t ∈ D. u∗ (t).8) v t − t∗  < δ. x. and v ∈ Ω such that (p∗ (t∗ )) B(t∗ )u∗ (t∗ ) < (p∗ (t∗ )) B(t∗ )v .8) implies that tf ∗ t0 (p (t)) (7. Remark 1: The geometric meaning of (7.6) is equivalent to (7. p)u ∈ Ω}. t0 .
We say that p is orthogonal to T 0 at z ∗ and we write p ⊥ T 0 (z ∗ ) if . . That is. + γn exp(δn (t)) for some γi . t0 .) 7. : [ . B(t) are constant. Suppose u∗ (·) is an optimal control. ˙ initial condition: G0 x(t0 ) = b0 . x0 ).11) In (7. and then translate these conditions in terms of the control. (7. We will analyze the problem in the same way as before. while b0 ∈ R 0 .11) G0 and Gf are ﬁxed matrices of dimensions 0 xn and f × n respectively. .1. Let f0 : Rn → R be a differentiable function and suppose that the objective function in (7. . β]. Show that u∗ (·) satisﬁes the maximum principle (7.2) is f0 (x(tf )) instead of c x(tf ). p∗ (t)) is a Lipschitz function of t.e. bf ∈ R f are ﬁxed vectors. Exercise 4: Suppose that Ω is bounded and closed. control constraint: u(·) ∈ U .2).) Exercise 6: Assume that K(tf . The notion of the previous section is retained. Show that there exists an optimal control u∗ (·) such that u∗ (t) belongs to the boundary of Ω for all t. (Hint: Show that (dm/dt) ≡ 0. t0 ≤ t ≤ tf . Exercise 5: Suppose Ω = [α. x0 ) is convex (see remark in Exercise 1 above). we ﬁrst characterize optimality in terms of the state at the ﬁnal time.) The next two exercises show how we can obtain important qualitative properties of an optimal control. Deﬁnition: Let p ∈ Rn . so that B(t) is an n × 1 matrix. then f0x (x∗ (tf )(x∗ (tf ) − x) ≤ for all x ∈ K(tf . (Hint: ﬁrst show that (p∗ (t)) B = γ1 exp(δ1 t) + . MORE GENERAL BOUNDARY CONDITIONS 87 Exercise 3: (i) Show that m(t) = M (t. t0 . { ] → ⊗ and u(·)piecewise continuous. Suppose that A(t) ≡ A and B(t) ≡ B are constant matrices and A has n real eigenvalues.3) with the ﬁnal condition p∗ (tf ) = f0 (x∗ (tf )) .2 More General Boundary Conditions We consider the following generalization of (7. .1 to show that if u∗ (·) is optimal. Maximize c x(tf ) subject to dynamics: x(t) = A(t)x(t) + B(t)u(t). Show that there is an optimal control u∗ (·) and t0 ≤ t1 ≤ t2 ≤ . For convenience let T 0 = {z ∈ Rn G0 z = b0 } . show that m(t) is constant.7. i. 0 ≤ i ≤ n.2. ﬁnal condition: Gf x(tf ) = bf . x∗ (t).. δi in R. Let z ∗ ∈ T 0 . ti+1 ). (Hint: Use Lemma 1 of 5. Also show that this condition is sufﬁcient for optimality if f0 is concave. (ii) If A(t). T f = {z ∈ Rn Gf z = bf } . ≤ tn ≤ tf such that u∗ (t) ≡ α or β on [ti .10) where p∗ (·) is the solution of the adjoint equation (7.
t0 . t1 t2 tf t t0 .88 Rn Rn p∗ (t2 ) p∗ (t 1) Rn Rn x0 0) = x∗ (t1 ) K(t1 . x0 ) K(tf . t0 . CONTINUOUSTIME LINEAR OPTIMAL CONTROL Figure 7. x0 ) p∗ (tf ) = c x∗ (tf ) CHAPTER 7. t0 .9) for t1 = t0 . t0 . x0 ) x∗ (t2 ) K(t2 . x0 ) x∗ (t = K(t0 .2: Illustration of (7.
such that p0 r 1 + p x1 ≥ p0 r 2 + p x2 for all (r i .13) are satisﬁed. S 1 is convex since T f is convex. Let x∗ (t) = φ(t. Also from (7. p ∈ Rn . (7. and p such that (7.18) implies that p0 r + p x∗ (tf ) ≥ p0 c x + p x for all x ∈ X(tf ). MORE GENERAL BOUNDARY CONDITIONS p (z − z ∗ ) = 0 for all z ∈ T 0 .19) can hold only if p0 ≥ 0. p ˆ p ˆ p ⊥ T f (x∗ (tf )) . (i) Suppose the Ω is convex.18) we get (7.14) (ii) Conversely if there exist p0 > 0. i = 1. u(·) ∈ U}. If u∗ (·) is optimal. xi ) ∈ S i . t0 . On the other hand letting r → ˆ c x∗ (tf ) we see that (7.15) (7. ˆ ˆ ˆ ˆ In particular (7. Since Ω is convex by hypothesis it follows by Exercise 1 of Section 1 that S 2 is convex.7.12) (7. Similarly if z ∗ ∈ T f . 89 Lemma 1: Let x∗ (t0 ) ∈ T 0 and u∗ (·) ∈ U . x)r > c x∗ (tf ). x)r = c x .20) (7. t0 )] (ˆ0 c + p) ⊥ T 0 (x∗ (t0 )) . p ˆ (7.17) First of all S 1 ∩ S 2 = φ because otherwise there exists x ∈ X(tf ) ∩ T f such that c x > c x∗ (tf ) contradicting optimality of u∗ (·) by (7. ˆ [Φ(tf . (i) Suppose that u∗ (·) is optimal. there exist p0 ∈ R. u)z ∈ T 0 . r > c x∗ (tf ). not ˆ ˆ both zero. S 2 by S 1 = {(r.14) is also satisﬁed. t0 )z + wz ∈ T 0 . Exercise 1: X(tf ) = {Φ(tf . z. such that (ˆ0 c + p) x∗ (tf ) ≥ (ˆ0 c + p) x for all x ∈ X(tf ) . w ∈ K(tf . Deﬁnition: Let X(tf ) = {Φ(tf . and suppose that x∗ (tf ) ∈ T f . ˆ ˆ ˆ ˆ (7. S 2 = {(r. x ∈ X(tf )} . In R1+m deﬁne sets S 1 . p0 ≥ 0 and p ∈ Rn . t0 .12). 2. p ⊥ T f (z ∗ ) if p (z − z ∗ ) = 0 for all z ∈ T f .19) can hold only if p0 c x∗ (tf ) + p x∗ (tf ) ≥ p0 c x + p x for all x ∈ X(tf ) . 0)}.15). Proof: Clearly u∗ (·) is optimal iff c x∗ (tf ) ≥ c x for all x ∈ X(tf ) ∩ T f .12) and (7.2. Secondly. t0 .16) (7. x ∈ T f } . ˆ ˆ ˆ ˆ which is the same as (7. then u∗ (·) is ˆ ˆ optimal and (7.18) .13) (7. u∗ ).5) there exists p0 ∈ R.19) Letting r → ∞ we conclude that (7. But then by the separation theorem for convex sets (see 5. not both ˆ ˆ ˆ zero. x∗ (t0 ).
and (7. t0 )(z − x∗ (t0 )) + x∗ (tf )} ∈ X(tf ) for all z ∈ T 0 . Clearly then for some c (in particular for the c in Figure 7. t0 )(z − x∗ (t0 )) for all z ∈ T 0 .12) we get p0 c x∗ (tf ) ≥ p0 c x . x0 . Intuitively p0 = 0 means that it is so difﬁcult to satisfy ˆ the initial and ﬁnal boundary conditions in (7. so that (7. ˆ ˆ ˆ ˆ which can hold only if p1 c x∗ (tf ) + p x ≥ p0 c x∗ (tf ) + p x∗ (tf ) for all x ∈ T f . Exercise 2: Suppose there exists z in the interior of X(tf ) such that z ∈ T f . we illustrate a 2dimensional situation where T 0 = {x0 }. (7.14) holds.21) But {x − x∗ (tf )x ∈ T f } = {zGf z = 0} is a subspace of Rn .12). (ii) Now suppose that p0 > 0 and p are such that (7.3) we are forced to set p0 = 0.12) we get 0 ≥ (ˆ0 c + p) Φ(tf . (7. ˆ ˆ˜ so that from (7. But in Figure ˆ 7. ˆ Remark 2: it would be natural to conjecture that in part (i) p0 may be chosen > 0.3 below. ˆ and (7. Let x ∈ X(tf ) ∩ T f . but basically if T ˆ0 f Exercise 2 below). In higher dimensions the reasons may be more ˆ f is “tangent” to X(t ) we may be forced to set p = 0 (see complicated.90 CHAPTER 7. ˆ which is the same as (7. ˆ ˆ ˜ but then by (7. t0 . In particular. so that from (7. because by the deﬁnition of X(tf ) and Exercise 1. Finally. t0 . p ˆ which can hold only if (7. T f is the vertical line. since then (7. u∗ ) ∈ T f is optimal for any c. ˆ ˆ ˜ Then from (7. It follows that the control u∗ (·) ∈ U for which x∗ (tf ) = φ(tf . .13) are satisﬁed.12). 0) is convex even if Ω is not (see Neustadt [1963]).12). x ∈ T f . But it is known that K(tf . {Φ(tf .13). in part (ii) of the Lemma we may assume p0 = 1. and T f ∩ X(tf ) consists of just one vector. ˆ We now translate the conditions obtained in Lemma 1 in terms of the control u∗ .14) hold for any vector c whatsoever.14).15) u∗ (·) is optimal. ˆ ˆ ˆ ˆ or p (x − x∗ (tf )) ≥ 0 for all x ∈ T f ˆ (7.21) can hold only if p (x − x∗ (tf )) = 0 for all x ∈ T f . p = (ˆ/ˆ0 ) will also satisfy (7.11) that optimization becomes a secondary matter. Then in part (i) we must have p0 > 0.13).13). Remark 3: In (i) the convexity of Ω is only used to guarantee that K(tf .14). we note that part (i) is not too useful if p0 = 0. ♦ Remark 1: If it is possible to choose p0 > 0 then p0 = 1. t0 . ˆ ˆ ˆ p p (7. 0) is convex. Finally (7.12) always implies (7.13) we conclude that p x∗ (tf ) = p x . CONTINUOUSTIME LINEAR OPTIMAL CONTROL p0 r + p x ≥ p0 c x∗ (tf ) + p x∗ (tf ) for all r > c x∗ (tf ).
x.26) are satisﬁed. tf ] → Rn . u∗ (t). (i) Suppose that Ω is convex. p) = sup{H(t. x. (7.24). p)v ∈ Ω}. x∗ (t0 )) .23) and (7. p∗ (t)) = M (t. t0 . If u∗ (·) is optimal for (7. not ˆ ˆ both zero.22). Let p0 = p∗ and p = ˆ ˆ 0 0 p∗ (tf ) − p∗ c.25). so that (7. p∗ (t)) . p) = p (A(t)x + B(t)u).12). (ii) Conversely suppose there exist p∗ > 0 and p∗ (·) satisfying (7.24).2. t0 )z + w) . and a 0 function p∗ : [t0 .12). (7. 0 and the maximum principle H(t.24) becomes equivalent to (7.22). u. (7. (7. whereas since K(tf . and (7. Next if x ∈ X(tf ) we have 0 (ˆ0 c + p) x = (p∗ (tf )) x p ˆ = (p∗ (tf )) (Φ(tf . then (7.14) are satisﬁed. MORE GENERAL BOUNDARY CONDITIONS 91 Theorem 1: Let x∗ (t0 ) ∈ T 0 and u∗ (·) ∈ U . (7.25) holds for all t ∈ [t0 .13) and (7.23).7. not both identically zero. (7. t0 . x. [Here H(t. Let x∗ (t) = φ(t.13) are respectively equivalent to (7.22). p ∈ Rn . . and (7. 0 Then u∗ (·) is optimal.22) with the ﬁnal condition p∗ (tf ) = p∗ c + p = p0 c + p . (ii) Suppose p∗ > 0 and (7. t0 ≤ t ≤ tf ˙ (7.13). such that (7.26): (p∗ (tf )) x∗ (tf ) ≥ (p∗ (tf )) x for all x ∈ K(tf . x∗ (t). x∗ (t0 )) ⊂ X(tf ).23) ﬁnal condition: (p∗ (tf ) − p∗ c)⊥T f (x∗ (tf )) . (7. Then by Lemma 1 there exist p ≥ 0. t0 . Let p∗ = p0 and let p∗ (·) be the solution ˆ 0 of (7. ˆ ˆ ˆ 0 Then (7.] Proof: A repetition of a part of the argument in the proof of Theorem 1 of Section 1 show that if p∗ satisﬁes (7. M (t.23). tf ] except possibly for a ﬁnite set.25) is equivalent to (7.24) (7. v. then there exist a number p∗ ≥ 0. (7.24). satisfying adjoint equation: p∗ (t) = −A (t)p∗ (t) . x∗ (t).26) (i) Suppose u∗ (·) is optimal and Ω is convex.22) initial condition: p∗ (t0 )⊥T 0 (x∗ (t0 )) (7. u∗ ) and suppose that x∗ (tf ) ∈ T f .26) is implied by (7. x∗ (t0 ).14) and (7.11).
x0 ) = X(tf ) f f Tf Figure 7.92 CHAPTER 7.3: Situation where p0 = 0 ˆ x0 = T 0 c t . x (t ) = X(t ) ∗ K(tf . t0 . CONTINUOUSTIME LINEAR OPTIMAL CONTROL Tf .
t0 )(z − x∗ (t0 )) p ˆ +(p∗ (tf )) (w + φ(tf . Exercise 4: How would you use Exercise 3 to solve Example 3 of Chapter 1? . t0 )x∗ (t0 )) ∈ K(tf . x∗ (t0 )). t0 . Show that Theorem 1 also holds for this case where. and we require that u(t) ∈ Ω(t) for all t. v. p)v ∈ Ω(t)}.2. ♦ Exercise 3: Suppose that the control constraint set is Ω(t) which varies continuously with t.7.25). x. 93 But by (7. it follows from (7. Hence (ˆ0 c + p) x = (p∗ (f )) Φ(tf . MORE GENERAL BOUNDARY CONDITIONS for some z ∈ T 0 and some w ∈ K(tf . t0 . in (7.26) that the second term is bounded by (p∗ (tf )) x∗ (tf ). t0 )x∗ (t0 )) = (p∗ (t0 )) (z − x∗ (t0 )) +(p∗ (tf )) (w + Φ(tf . M (t. p) =sup{H(t. t0 )x∗ (t0 )) . and since (w+φ(tf . p ˆ p ˆ and so u∗ (·) is optimal by Lemma 1. 0).23) the ﬁrst term on the right vanishes. x. Thus (ˆ0 c + p) x∗ (tf ) ≥ (ˆ0 c + p) x for all x ∈ X(tf ) .
CONTINUOUSTIME LINEAR OPTIMAL CONTROL .94 CHAPTER 7.
(For complete proofs see (Lee and Markus [1967] or Pontryagin. Unfortunately when f is nonlinear such a characterization is not available.1 Main Results 8. This comparison was possible because we had an explicitly characterization of x(·) in terms of u(·). We are interested in the optimal control of a system whose dynamics are governed by the nonlinear differential equation x(t) = f (t.1).2 is presented in Section 1.. We assume throughout that the function f : [t0 .1) where x(t) ∈ is the state and u(t) ∈ is the control. Section 3 deals with the minimumtime problem and Section 4 considers the important special case of linear systems with quadratic cost. it is possible to convey the main ideas of the proofs at an intuitive level and we shall do so. ˙ Rn Rp u∗ (·) (8. But ﬁrst we need to impose some regularity conditions on the differential equation (8.1.Chapter 8 SEQUENTIAL DECISION PROBLEMS: CONTINUOUSTIME OPTIMAL CONTROL OF NONLINEAR SYSTEMS We now present a sweeping generalization of the problem studied in the last chapter. (t). In the case of linear systems we obtained the necessary and x conditions for optimality by comparing x∗ (·) with trajectories x(·) corresponding to other admissible controls u(·). tf ] × Rn × Rp → Rn satisﬁes the following conditions: 95 . An alternative form of the objective function is discussed in Section 2. We can then estimate the difference between x(·) and x∗ (·) by the solution to a linear differential equation as shown in Lemma 1 below. which is a direct generalization of Theorem 1 of 7. However. Instead we shall settle for a comparison between the trajectory x∗ (·) and trajectories x(·) obtained by perturbing the control u∗ (·) and the initial condition x∗ (t0 ). in Section 5 we discuss the socalled singular case and also analyze Example 4 of Chapter 1.) The principal result. Unfortunately we are forced to omit the proofs of the results since they require a level of mathematical sophistication beyond the scope of these Notes. u(t)) .1 Preliminary results based on differential equation theory. x. Finally. 8. [1962]. Suppose is an optimal control ∗ (·) is the corresponding trajectory. t0 ≤ t ≤ tf . et al.
. Let u∗ (·) ∈ U be ﬁxed and let D∗ be the set of discontinuity points of u∗ (·). m. z. . . u(t))]Φ(t. except for a ﬁnite subset D ⊂ [t0 . tm < tf . . . t1 . m is a nonnegative integer. and 4. The following result is proved in every standard treatise on differential equations. t1 . i = 1. ·. . fx . u(·)). u. t1 ≤ t ≤ tf . x. CONINUOUSTIME OPTIMAL CONTROL 1. m (recall that D is the set of ≥ 0. i = 1. for ﬁxed t1 ≤ t2 in [t0 . (t). t1 . for every ﬁnite α. 1. the n × n matrixvalued function Φ deﬁned by Φ(t2 . . u(·)) is the solution of the linear homogeneous differential equation ∂Φ ∂t (t. fu are continuous on [t0 . . . 0 Deﬁnition: π = (t1 . there exists a unique solution x(t) = φ(t. . . tf ] and ﬁxed u(·). tf ]×Rn × Rp . and ti ∈ D ∗ discontinuity points of f ). x ∈ Rn .96 CHAPTER 8. t1 . of the differential equation x(t) = f (t. . u(·)) = In . x(t). z. tf ]. the functions f. . . . . . tf ]. um ) is said to be a perturbation data for u∗ (·) if 1. ∂x and the initial condition Φ(t1 . t1 . i D. z. ·) : Rn xRp → Rn is continuously differentiable in the remaining variables (x. . z. . u). (·)) = [ ∂f (t. tf ] → Ω. 2. tm . ui ∈ Ω. and 3. t1 ≤ t ≤ tf . tf ]. . t1 . t1 ≤ t ≤ tf . u(·)) = ∂φ ∂z (t2 . . . the function φ(t2 . i = 1. . ˙ satisfying the initial condition x(t1 ) = z . u(t)) . Theorem 1: For every z ∈ Rn . for each ﬁxed t ∈ [t0 . u) ≤ β + γx for all t ∈ [t0 . . u1 . and every piecewise continuous function u(·) : [t0 . z. . Moreover. there exist ﬁnite number β and γ such that f (t. 3. u(·)) . t1 . . x. 2. Furthermore. for every t1 ∈ [t0 . Now let Ω ⊂ Rp be a ﬁxed set and let U be set of all piecewise continuous functions u(·) : [t0 . . z. ·. tf ] → Rp . u ∈ Rp with u ≤ α . tf ]. m. f (t. t0 < t1 < t2 < . m. u(·)) : Rn → Rn is differentiable. Let x∗ ∈ Rn be a ﬁxed initial condition.
uj ) − f (tj . t1 ) 1 . t1 . 0 More precisely we have the following result which we leave as an exercise. z. u(π. let Q(t) = {h(π. Let Φ(t2 .ε) (t) = ui for all t ∈ [ti − ε i . t0 . x∗ ). t ∈ [t0 .ξ) (·) the linearized (trajectory) perturbation corresponding to (π. The proof of the lemma is a straightforward exercise in estimating differences of solutions to differential equations. = Φ(t. tf ] let K(t. z) = {φ(t. . Lemma 1: lim xε (t) − x∗ (t) − εh(π.ε) (·) ∈ U corresponding to π is deﬁned by u(π. MAIN RESULTS 97 Let ε(π) > 0 be such that for 0 ≤ ε ≤ ε(π) we have [ti − ε i . starting at time t0 in state z. . Deﬁnition: For each t ∈ [t0 .0) (·)is the linearized perturbation corresponding to(π. t0 )ξ + Φ(t.ε) (·) is given by ε→0 h(π. t ∈ [tm .ε) (·)). 0)} . tj )[f (tj . x∗ (tj ). x(ξ. t1 ) = 0 Φ(t2 . t0 . t0 . t0 )ξ + j=1 m Φ(t. 0 Now let x∗ (t) = φ(t. ti ] . t ∈ [ti . . tf ]. x∗ (t1 ). x∗ (t1 ). u∗ (tj ))] Φ(t. t0 )ξ = Φ(t. the set x∗ (t) + Q(t) can serve as an approximation to the set K(t. ti+1 ) . x∗ (tj ). i = 1. x(ξ.ε) (t) = Φ(t.ε) (t) = 0 for t ∈ [t0 . uj ) − f (tj . and h(π. u(·))u(·) ∈ U} be the set of states reachable at time t. ti ] ⊂ [t0 .8. u∗ (t1 ))] i . where h(π.ε) . u∗ (·)) and let xε (t) = φ(t. u∗ (·)). Then for 0 ≤ ε ≤ ε(π). t0 )ξ + j=1 j (See Figure 8. t0 . u1 ) − f (t1 . t2 ) j = Φ(t. tj ] = φ for i = j. t ∈ [t0 . t0 . x∗ (tj ). .ε) = x∗ . tf ] for all i. t1 ]. In particular for ξ = 0. and using controls u(·) ∈ U . t1 )[f (t1 . ti ] [tj −ε j .ε) ) up to an error of order o(ε). m u∗ (t) otherwise .0) (t)πis a perturbation data for u∗ (·). and it is omitted (see for example (Lee and Markus [1967])). t0 . Deﬁnition: Any vector ξ ∈ Rn is said to be a perturbation for x∗ . tm )[f (tj . tf ] .1.1. u∗ tj ))] .ε) deﬁned for 0 ε > 0 is said to be a perturbed initial condition if lim x(ξ. x∗ . ξ). x∗ (t1 ). and [ti −ε i . x∗ (tj ). t ∈ [t1 . The following lemma gives an estimate of x∗ (t) − xε (t). and a function x(ξ.the perturbed control u(π.ε) − x∗ ) = ξ .) We call h(π. . Remark: By Lemma 1 (x∗ (t)+εh(π. 0 ε→0 and ε→0 lim 1 ε (x(ξ. Deﬁnition: For z ∈ Rn .ξ) ) belongs to the set K(t.
1. u∗ (·)).4) and (8.1.. Let u∗ (·) ∈ U be an optimal control and let x∗ (t) = φ(t. u(t)) . t0 ≤ t ≤ tf .) Show that Q(t) ⊂ C(K(t. tf ] → Ω and u(·) piecewise continuous .1. t0 . CONINUOUSTIME OPTIMAL CONTROL u1 u(πε) (·) u∗ (·) ε  1 u2 t1   u3 ε 2 ε   3 t0 x t2 t3 tf x∗ (·)   xε (·) x( ξ.3): Maximize ψ(x(tf )) subject to dynamics: x(t) = f (t. 0 ﬁnal condition: x(tf ) ∈ Rn . ε) εhπξ t1     t2 t3 tf Figure 8. u∗ (t))] p∗ (t). ˙ ∂x (8. x∗ (t). t0 . t0 ≤ t ≤ tf . i. be the 0 corresponding trajectory. Exercise 1: (Recall the deﬁnition of the tangent cone in 5. be the solution of (8. 0 We can now prove a generalization of Theorem 1 of 7. x(t). Let p∗ (t).5): adjoint equation: p∗ (t) = −[ ∂f (t. Theorem 2: Consider the optimal control problem (8. ˙ initial condition: x(t0 ) = x∗ . x∗ ). t0 ≤ t ≤ tf .1: Illustration for Lemma 1. t0 ≤ t ≤ tf . x∗ .98 u CHAPTER 8.4) .e. control constraint: u(·) ∈ U .2) (8. u : [t0 . x∗ (t)) .3) where ψ : Rn → R is differentiable and f satisﬁes the conditions listed earlier. (8.
0 and in particular from (8. Exercise 2: Show that (i) Q(t) is a cone. p) = p f (t.0) (t∗ ) > 0 .6) does not hold from some t∗ ∈ D ∗ ∪ D. if h ∈ Q(t) and λ ≥ 0. MAIN RESULTS ﬁnal condition: p∗ (tf ) = Then u∗ (·) satisﬁes the maximum principle H(t. t0 . x. 0 and so by Lemma 1 of 5. x. The proof of the lemma depends upon some deep topological results and is omitted. below. x∗ (t). x(t∗ ). The problem involving more general boundary conditions is much more complicated and requires more reﬁned analysis. i. (ii) for t0 ≤ t1 ≤ t2 ≤ tf .e. x(t∗ ). Instead we offer a plausibility argument. t∗ ).9) is equivalent to p∗ (tf ) h(π. Exercise 3: Show that (i) C(t) is a convex cone. If we consider the perturbation data π = (t∗ . p) = sup{H(t. Speciﬁcally. p∗ (t)) = M (t.7). then λh ∈ Q(t). Remark: From Lemma 1 we know that if h ∈ C(t) then (x∗ (t) + εh) belongs to K(t. t0 . x∗ (t0 )) up to an error of order o(ε). ♦ (8. In Theorem 2 the initial condition is ﬁxed and the ﬁnal condition is free. Then for all ε > 0 sufﬁciently small.1. .. Φ(t2 . p)v ∈ Ω}]. v) − f (t∗ . v). tf ] except possibly for a ﬁnite set. x. t1 )C(t1 ) ⊂ C(t2 ) . 99 (8.2 More general boundary conditions. u∗ (t∗ ))] > 0 . Proof: Since u∗ (·) is optimal we must have ψ(x∗ (tf )) ≥ ψ(z) for all z ∈ K(tf . x∗ (t). But ﬁrst we need some simple properties of the sets Q(t) which we leave as exercises. ). Lemma 2: Let h belong to the interior of the cone C(t). Then there exists v ∈ Ω such that (8. Φ(t2 . (8. x∗ ) . u∗ (t).2) ψx (x∗ (tf ))h ≤ 0 for all h ∈ Q(tf ) . Lemma 2.1 ψ(x∗ (tf ))h ≤ 0 for all h ∈ C(K(tf .9) Now from (8.6) for all t ∈ [t0 .7) Now suppose that (8. x∗ (tf )) . (ii) for t0 ≤ t1 ≤ t2 ≤ tf . Deﬁnition: Let C(t) denote the closure of Q(t).8.1. x∗ ). u. x∗ (t0 )) for ε > 0 sufﬁciently small. [Here H(t. asserts further that if h is in the interior of C(t) then in fact (x∗ (t) + εh) ∈ K(t. t∗ )h(π. M (t.4) we can see that p∗ (t∗ ) = p∗ (tf ) Φ(tf . then (8. Lemma 1 needs to be extended to Lemma 2 below. t0 .0) (tf ) = Φ(tf . 1. t1 )Q(t1 ) ⊂ Q(t2 ) . p∗ (t)) ψ(x∗ (tf )) . t0 .0) (tf ) > 0 which contradicts (8.5) (8. u.8) 8.8) is equivalent to p∗ (t∗ ) h(π. x. p∗ (t∗ ) [f (t∗ . v.1.0) (t∗ ) so that (8. Also h(π.
100
CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
(x∗ (t) + εh) ∈ K(t, t0 , x∗ ) . 0 (8.10)
Plausibility argument. (8.10) is equivalent to εh ∈ K(t, t0 , x∗ (t0 )) − {x∗ (t)} , where we have moved the origin to x∗ (t). The situation is depicted in Figure 8.2. ˆ K(ε) ˆ C(ε) o(ε) K(t1 , t0 , x∗ ) − {x∗ (t)} εh 0
(8.11)
h
δε
C(t)
Figure 8.2: Illustration for Lemma 2. ˆ Let C(ε) be the crosssection of C(t) by a plane orthogonal to h and passing through εh. Let ˆ K(ε) be the crosssection of K(t, t0 , x∗ ) − {x∗ (t0 )} by the same plane. We note the following: 0 ˆ ˆ (i) by Lemma 1 the distance between C(ε) and K(ε) is of the order o(ε); ˆ (ii) since h is in the interior of C(t), the minimum distance between εh and C(ε) is δε where δ > 0 is independent of ε. ˆ Hence for ε > 0 sufﬁciently small εh must be trapped inside the set K(ε). (This would constitute a proof except that for the argument to work we need to show that there ˆ are no “holes” in K(ε) through which εh can “escape.” The complications in a rigorous proof arise precisely from this drawback in our plausibility argument.) ♦ ∗ ) in a neighborhood of x∗ (t) when we Lemmas 1 and 2 give us a characterization of K(t, t0 , x0 perturb the control u∗ (·) leaving the initial condition ﬁxed. Lemma 3 extends Lemma 2 to the case when we also allow the initial condition to vary over a ﬁxed surface in a neighborhood of x∗ . 0 0 Let g0 : Rn → R 0 be a differentiable function such that the 0 × n matrix gx (x) has rank 0 n 0 0 0 ∗ 0 0 for all x. Let b ∈ R be ﬁxed and let T = {xg (x) − b }. Suppose that x0 ∈ T and let 0 (x∗ ) = {ξg 0 (x∗ )ξ = 0}. Thus, T 0 (x∗ ) + {x∗ } is the plane through x∗ tangent to the surface T x 0 0 0 0 0 T 0 . The proof of Lemma 3 is similar to that of Lemma 2 and is omitted also. Lemma 3: Let h belong to the interior of the cone {C(t)+Φ(t, t0 )T 0 (x∗ )}. For ε ≥ 0 let h(ε) ∈ Rn 0 1 be such that lim h(ε) = 0, and lim ( )h(ε) = h. Then for ε > 0 sufﬁciently small there exists ε→0 ε x0 (ε) ∈ T 0 such that (x∗ (t) + h(ε)) ∈ K(t, t0 , x0 (ε)) .
8.1. MAIN RESULTS
101
We can now prove the main result of this chapter. We keep all the notation introduced above. f Further, let gf : Rn → R f be a differentiable function such that gx (x) has rank f for all x. f ∈ Rn be ﬁxed and let T f = {xg f (x) − bf }. Finally, if x∗ (t ) ∈ T f let T f (x∗ (t )) = Let b f f f {ξgx (x∗ (tf ))ξ = 0}. Theorem 3: Consider the optimal control problem (8.12): Maximize ψ(x(tf )) subject to dynamics: x(t) = f (t, x(t), u(t)) , t0 ≤ t ≤ tf , ˙ initial conditions: g0 (x(t0 )) = b0 , ﬁnal conditions: gf (x(tf )) = bf , control constraint: u(·) ∈ U , i.e., u : [t0 , tf ] → Ω and u(·) piecewise continuous .
(8.12)
Let u∗ (·) ∈ U , let x∗ ∈ T 0 and let x∗ (t) = φ(t, t0 , x∗ , u∗ (·)) be the corresponding trajectory. 0 0 Suppose that x∗ (tf ) ∈ T f , and suppose that (u∗ (·), x∗ ) is optimal. Then there exist a number 0 p∗ ≥ 0, and a function p∗ : [t0 , tf ] → Rn , not both identically zero, satisfying 0 adjoint equation: p∗ (t) = −[ ∂f (t, x∗ (t), u∗ (t))] p∗ (t), t0 ≤ t ≤ tf , ˙ ∂x initial condition: p∗ (t0 )⊥T 0 (x∗ ) , 0 ﬁnal condition: (p∗ (tf ) − p∗ ψ(x∗ (tf )))⊥T f (x∗ (tf )) . 0 Furthermore, the maximum principle H(t, x∗ (t), u∗ (t), p∗ (t)) = M (t, x∗ (t), p∗ (t)) (8.16) (8.13) (8.14) (8.15)
holds for all t ∈ [t0 , tf ] except possibly for a ﬁnite set. [Here H(t, x, p, u) = p f (t, x, u, ), M (t, x, p) = sup{H(t, x, v, p)v ∈ Ω}]. Proof: We break the proof up into a series of steps. Step 1. By repeating the argument presented in the proof of Theorem 2 we can see that (8.15) is equivalent to p∗ (tf ) h ≤ 0 for all h ∈ C(tf ) . (8.17)
Step 2. Deﬁne two convex sets S1 , S2 in R1+m as follows: S1 = {(y, h)y > 0, h ∈ T f (x∗ (tf ))}, S2 = {(y, h)y = ψx (x∗ (tf ))h, h ∈ {C(tf ) + Φ(tf , t0 )T 0 (x∗ )}} . 0 We claim that the optimality of (u∗ (·), x∗ ) implies that S1 ∩ Relative Interior (S2 ) = φ. Suppose 0 this is not the case. Then there exists h ∈ T f (x∗ (tf )) such that ψx (x∗ (tf ))h > 0 , (8.18)
102
CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL
h ∈ Interior{C(tf ) + Φ(tf , t0 )T 0 (x∗ )} . 0 (8.19)
f f Now by assumption gx (x∗ (tf ) has maximum rank. Since gx (x∗ (tf ))h = 0 it follows that the Implicit Function Theorem that for ε > 0 sufﬁciently small there exists h(ε) ∈ Rn such that
gf (x∗ (tf ) + h(ε)) = bf ,
(8.20)
and, moreover, h(ε) → 0, (1/ε)h(ε) → h as ε → 0. From (8.18) and Lemma 3 it follows that for ε > 0 sufﬁciently small there exists x0 (ε) ∈ T 0 and uε (·) ∈ U such that x∗ (tf ) + h(ε) = φ(tf , t0 , x0 (ε), uε (·)) . Hence we can conclude from (8.20) that the pair (x0 (ε), uε (·)) satisﬁes the initial and ﬁnal conditions, and the corresponding value of the objective function is ψ(x∗ (tf ) + h(ε)) = ψ(x∗ (tf )) + ψx (x∗ (tf ))h(ε) + o(h(ε)) , and since h(ε) = εh + o(ε) we get ψ(x∗ (tf ) + h(ε)) = ψ(x∗ (tf )) + ε)ψx (x∗ (tf ))h + o(ε) ; but then from (8.18) ψ(x∗ (tf ) + h(ε)) > ψ(x∗ (tf )) for ε > 0 sufﬁciently small, thereby contradicting the optimality of (u∗ (·), x∗ ). 0 Step 3. By the separation theorem for convex sets there exist p0 ∈ R, p1 ∈ Rn , not both zero, such ˆ ˆ that p0 y 1 + p1 h1 ≥ p0 y 2 + p1 h2 for all (y i , hi ) ∈ S1 , i = 1, 2 . ˆ ˆ ˆ ˆ (8.21)
Arguing in exactly the same fashion as in the proof of Lemma 1 of 7.2 we can conclude that (8.21) is equivalent to the following conditions: p0 ≥ 0 , ˆ p1 ⊥T f (x∗ (tf )) , ˆ Φ(tf , t0 ) (ˆ0 ψ(x∗ (tf )) + p1 )⊥T 0 (x∗ ) , p ˆ 0 and (ˆ0 ψx (x∗ (tf )) + p1 )h ≤ 0 for all h ∈ C(tf ) . p ˆ (8.24)
(8.22)
(8.23)
If we let p∗ = p0 and p∗ (tf ) = p0 ψ(x∗ (tf )) + p1 then (8.22), (8.23), and (8.24) translate respecˆ0 ˆ ˆ ˆ tively into (8.15), (8.14), and (8.17). ♦
u∗ (t))] p∗ (t) . x∗ . [Here H(t. Finally. ﬁnal conditions: gf (x(tf )) = bf . and suppose that x∗ (tf ) ∈ T f . t0 ≤ t ≤ tf . let x∗ ∈ T o and let x∗ (t) = φ(t. the maximum principle ˜ ˜ H(t. INTEGRAL OBJECTIVE FUNCTION 103 8. u(t))dt (8. x(t). To this end we deﬁned the augmented system with state variable x = (x0 . x∗ (t). x∗ (t). x(t). p. x. ˜ ˜ ∂x ˜ initial condition: p∗ (t0 )⊥T 0 (x∗ ) .26): tf Maximize t0 f0 (t. x∗ ) is optimal. Evidently then the problem of maximizing (8. u∗ (·)). u(t)) x0 (t) ˙ ˜ ˜ . x. tf ] except possibly for a ﬁnite set. p. x subject to the augmented dynamics and constraints which is of the form treated in Theorem 3 of Section 1.25) is equivalent to the ˜ x problem of maximizing ψ(˜(tf )) = x0 (tf ) . and M (t. x. u∗ (t)) = M (t. then there exists a function p∗ = (p∗ .26) subject to dynamics: x(t) = f (t. p∗ (t)) ≡ 0.2. ˙ initial conditions: g0 (x(t0 )) = b0 . then M ˜ ˜ Exercise 1: Prove Theorem 1. Theorem 1: Consider the optimal control problem (8. t0 . and control constraints are the same as before. p∗ (t)) ≡ constant.] ˜ ˜ ˜ (t. u) = ˜ ˜˜ ˜ ˜ p0 f0 (t. x. if f0 and f do not explicitly depend on t. x. u).25) The dynamics of the state.8. and we get the following result. p∗ (t). x(t). u(t)) The initial and ﬁnal conditions which are of the form g0 (x) = b0 . x. p) = sup{H(t. u) + p f (t. 0 ﬁnal condition: p∗ (tf )⊥T f (x∗ (tf )) . u) = p f (t. We proceed to show how such objective functions can be treated as a special case of the problems of the last section.2 Integral Objective Function In many control problems the objective function is not given as a function ψ(x(tf )) of the ﬁnal state. satisfying 0 0 (augmented) adjoint equation: p∗ (t) = −[ ∂ f (t. x(t). not identically ˜ 0 0 zero. and with p∗ (t) ≡ constant and p∗ (t) ≥ 0. v)v ∈ Ω}.) ˜ · ˜ . x(t). control constraint: u(·) ∈ U . x= ˜ = f (t. Futhermore. x) ∈ ˜ R1+m as follows: · f0 (t. If 0 0 (u∗ (·). u(t)) = x(t) ˙ f (t. the boundary conditions. p∗ (t)) ˜ ˜ ˜ holds for all t ∈ [t0 . u(t)). u(t))dt . (Hint: For the ﬁnal part show that (d/dt) M (t. p∗ ) : [t0 . x∗ (t). x∗ (t). Let u∗ (·) ∈ U. x∗ (t). x(t). gf (x) = bf are augmented g0 (˜) = ˜ x x0 g0 (x) = ˜0 = b 0 b0 and gf (˜) = gf (x) = bf . tf ] → R1+m . (8. but rather as an integral of the form tf t0 f0 (t.
0 ≤ s ≤ 1 . u(t)). (8.27) We analyze (8. x(t). tf ] into a ﬁxedtime interval [0. ∞) .104 CHAPTER 8. t0 ≤ t ≤ tf . In many important cases the ﬁnal time is itself a decision variable. tf ] → [0. One such case is the minimumtime problem where we want to transfer the state of the system from a given initial state to a speciﬁed ﬁnal state in minimum time. consider the optimal control problem (8. where s(·) : [t0 . (8. CONINUOUSTIME OPTIMAL CONTROL 8. s(t0 ) = 0. with initial condition t(0) = t0 .1 Main result. Now if x(·) is the solution of x(t) = f (t. then it is easy to see that z(·) is the solution of dz ds (s) (8.28) = α(s)f (s. ˙ initial condition: g0 (x(t0 )) = b0 . ﬁnaltime constraint: tf ∈ (t0 . tf Maximize t0 f0 (t. ∞). ﬁnal condition: gf (x(t)f )) = bf . x(t). 1]. In the problem considered up to now the ﬁnal time tf is assumed to be ﬁxed. x(t0 ) = x0 ˙ and if we deﬁne z(s) = x(t(s)). in fact. u(t))dt subject to dynamics: x(t) = f (t. (t). z(s). 1] is the functional inverse of s(t). . v(s) = u(t(s)) . ˙ .29) Conversely from the solution z(·) of (8. 1]. v(s)) . 0 ≤ s ≤ 1 . This change of timescale is achieved by regarding t as a new state variable and selecting a new time variable s which ranges over [0. u(t)) .29) we can obtain the solution x(·) of (8.3 Variable Final Time 8. t0 ≤ t ≤ tf .28) by x(t) = z(s(t)) .27). Here α(s) is a new control variable constrained by α(s) ∈ (0. control constraint: u(·) ∈ U . x. The equation for t is dt(s) ds = α(s) . More generally. t0 ≤ t ≤ tf . s(·) is the solution of the differential equation s(t) = 1/α(s(t)). 0 ≤ s ≤ 1 z(0) = x0 .27) by converting the variable time interval [t0 .3.
31) (8. α∗ (·)). ˙ initial constraint: g0 (z(0)) = b0 . p∗ ) : [t0 . If (u∗ (·).30) such that the correspond∗ ing trajectory (t∗ (·). and α∗ (·) by ∗ z0 = x∗ 0 ∗ (s) = u∗ (t + s(t∗ − t )) v . t0 ≤ t ≤ t∗ .30).30) Lemma 1: (i) Let x∗ ∈ T 0 . and suppose that x∗ (t∗ ) ∈ T f . α∗ (·)). Suppose that x∗ (t∗ ) ∈ T f . u∗ (t))] p∗ (t) . ﬁnal constraint: gf (z(1)) = bf . z(s). t∗ ) is optimal for 0 f f ∗ (8. t∗ ∈ (t0 . let x∗ ∈ T 0 . Deﬁne x∗ . t0 . z) ∈ R1+m . t∗ ) is optimal for (8. t(1) ∈ R . f Exercise 1: Prove Lemma 1. u∗ (·) ∈ U. let t∗ ∈ (0. u∗ (·)). α∗ (·)) be an admissible control for (8.32) . f ∗ where s∗ (·) is functional inverse of t∗ (·). and suppose that (u∗ (·). not identically zero. u∗ (·) ∈ U . f ∗ Then ((v ∗ (·). and let (v ∗ (·). t∗ ] → R1+m .30).30). v(s))α(s)ds (8. The relation between problems (8. ˜ ˜ ∂x ˜ initial condition: p∗ (t0 )⊥T 0 (x∗ ) .30) is established in the following result. Then (u∗ (·). α(·) piecewise continuous. and ˜ 0 f with p∗ (t) ≡ constant and p∗ (t) ≥ 0. 0 u∗ (t) = v ∗ (s∗ (t)) . α(s)) ∈ Ω × (0. 0 ·∗ ˜ (8. satisfying 0 0 (augmented) adjoint equation: p (t) = −[ ∂ f (t. u∗ (·)) be the 0 0 f corresponding trajectory. 0≤s≤1. 0 f0 (t(s). ∗ (ii) Let z0 ∈ T 0 . then there exists a function p∗ = (p∗ .27) and (8.27). ∞) and let x∗ (t) = φ(t. z0 ) is optimal for (8.8. α(s)). v) ∈ R1+p : 1 Maximize subject to ˙ dynamics: (z(s). control constraint: (v(s). t∗ ) is optimal 0 0 f f for (8.27). x∗ (t). ∞). t(s)) = (f (t(s). and let 0 f x∗ (t) = φ(t. t(0) = t0 . x∗ . x∗ . Deﬁne z0 . z(s). t0 . ∞) for 0 ≤ s ≤ 1 and v(·). f t∗ = t∗ (1) . Theorem 1: Let u∗ (·) ∈ U. x∗ . t0 ≤ t ≤ tf . z0 . 0 0 f α∗ (s) = (t∗ − t0 ) . z ∗ (·)) satisﬁes the ﬁnal conditions of (8. Suppose that ((v ∗ (·).27).30).3. v ∗ (·). v(s))α(s). and t∗ by 0 f ∗ x∗ = z0 . where the state vector (t. VARIABLE FINAL TIME 105 With these ideas in mind it is natural to consider the ﬁxedﬁnaltime optimal control problem (8. 0≤s≤1. x∗ . z0 ) is optimal for (8. and the control (α.
z ∗ (s). ∞)} n+1 (8. p∗ is not identically zero. v ∗ (s))] λ∗ (s) n+1 0 +[ ∂f (t∗ (s). ˜ ∗ Proof: By Lemma 1. v ∗ (s) = u∗ (t0 + s(t∗ − t0 )) and α∗ (s) = (t∗ − t0 ) for 0 ≤ s ≤ 1 0 f f constitute an optimal solution for (8. there exists a function λ∗ = (λ∗ . tf ] except possibly for a ﬁnite set. and with λ0 0 (8. w)β + λ∗ (s)β]w ∈ Ω. w)β 0 +λ∗ (s) f (t∗ (s). ˜ ˜ holds for all t ∈ [t0 . Let s∗ (t) = (t − t0 )/(t∗ − t0 ). λ∗ ≡ constant. v ∗ (s))α∗ (s) 0 +λ∗ (s) f (t∗ (s).37) (8. z ∗ (s). the maximum principle λ∗ (s)f0 (t∗ (s). then from (8. t∗ must be such that f ˆ H(t∗ . u∗ (t∗ )) = 0 . λ∗ . λ∗ ) : [0. x∗ (t). z ∗ (s). p∗ (t)) ≡ 0. 0 ≤ s ≤ 1 .40) we have (λ∗ . x∗ (t). p∗ (t) = λ∗ (s∗ (t)). but from (8.38). not 0 n+1 ∗ (s) ≡ constant and λ∗ (s) ≥ 0. t0 ≤ t ≤ t∗ .38) ﬁnal condition: λ∗ (1)⊥T f (z ∗ (1)) . z ∗ (s). z ∗ (s). n+1 Furthermore. p∗ (t∗ ). 1] → R1+n+1 . The resulting trajectory is z ∗ (s) = x∗ (t0 + s(t∗ − t0 )). 1] except possibly for a ﬁnite set. if f0 and f do not explicitly depend on t. x∗ (t).106 CHAPTER 8. Because if p∗ ≡ 0. CONINUOUSTIME OPTIMAL CONTROL ﬁnal condition: p∗ (t∗ )⊥T f (x∗ (t∗ )) .33) Also the maximum principle ˜ ˜ H(t. f f ˜ f f ˆ Finally. t∗ ] → R1+n by ˜ 0 f f f p∗ (t) = λ∗ (s∗ (t)). v ∗ (s))α∗ (s) + λ∗ (s)α∗ (s) n+1 = sup{[λ∗ (s)f0 (t∗ (s). then M (t. p∗ (t).34) (8. x∗ (t∗ ). z0 = x∗ . p∗ (t)) . v ∗ (s))] λ∗ (s) ˙ λ∗ (t) 0 ∂z = − +[ ∂f (t∗ (s). z ∗ (s). t∗ (s) = t0 + s(t∗ − t0 ).35) ˙ 0 λ∗ (t) 0 {[ ∂f0 (t∗ (s).36). f ˜ By Theorem 1 of Section 2. 0 0 f (8. Furthermore. v ∗ (s))] λ∗ (s)}α∗ (s) adjoint equation: ∂z ∂f0 ∗ λ∗ (t) ˙ {[ ∂t (t (s).39) holds for all s ∈ [0.36) (8. satisfying identically zero. t0 ≤ t ≤ t∗ . β ∈ (0.40) First of all. so that in particular f f z ∗ (1) = x∗ (t∗ ). p∗ ) : [t0 . u∗ (t)) = M (t. v ∗ (s))] λ∗ (s)}α∗ (s) ∂t ∗ initial condition: λ∗ (0)⊥T 0 (z0 ) (8. λ∗ ) ≡ 0 and ˜ ˜ 0 ˜ then from (8. λ∗ (1) = 0 so that we would have λ∗ ≡ 0 n+1 n+1 . and deﬁne p∗ = (p∗ .30). f f (8. z ∗ (s). z ∗ (s). λ∗ (1) = 0 .
u∗ (t))] p∗ (t). and. f Also the maximum principle H(t.38) respectively imply (8. Let x∗ (·) be the corresponding f f trajectory. z ∗ (s).2 Minimumtime problems . x∗ (t).35) and the fact that M (t. We consider the following special case of (8. x∗ (t). 0 (8. ∞) and let u∗ : [t0 . ∞) . v ∗ (s)) + λ∗ (s) = 0 n+1 and λ∗ (s)f0 (t∗ (s).45) (8. w) + λ∗ (s) f (t∗ (s). p∗ . t∗ ] → Rn . u(t)). f Finally. z ∗ (s). (t)) ≡ constant . xf are ﬁxed. n+1 ˜ Finally. p∗ (t). Applying Theorem 1 to this problem gives Theorem 2.39) is equivalent to λ∗ (s)f0 (t∗ (s). the last assertion of the Theorem follows from (8. (8.34) and (8. control constraint: u(·) ∈ U . p∗ (t)) holds for all t ∈ [t0 .27): tf Maximize t0 (−1)dt subject to dynamics: x(t) = f (t.33).42) (8. on the other hand (8. z ∗ (s).43). (8. v ∗ (s)) 0 = Sup {[λ∗ (s)f0 (t∗ (s).32) and (8. not identically zero. ﬁnal condition: x(tf ) = xf . z ∗ (s). Theorem 2: Let t∗ ∈ (t0 . so that the optimal control problem consists of ﬁnding a control which transfers the system from state x0 at time t0 to state xf in minimum time. v ∗ (s)) + λ∗ (s) f (t∗ (s). v ∗ (s)) 0 +λ∗ (s) f (t∗ (s).43) In (8. VARIABLE FINAL TIME 107 which is a contradiction. p∗ (tf )) ≥ 0 f and if f does not depend explicitly on t then M (t.41) and the fact that λ∗ (1) = 0. x∗ (t). It is trivial to verify that p∗ (·) satisﬁes (8. t0 ≤ t ≤ tf ˙ initial condition: x(t0 ) = x0 . x0 . t∗ ] except possibly for a ﬁnite set.46) (8.35) follows from (8. f are not explicitly dependent on t. x∗ (tf ).8. z ∗ (s). u∗ (t)) = M (t. x∗ (t). ﬁnaltime constraint: tf ∈ (t0 . p∗ (t)) ≡ ˜ constant if f0 . Next.3. z ∗ (s). M (t∗ .44) . w)]w ∈ Ω}. Then there exists a function p∗ : [t0 . ♦ 8. ˙ f ∂x initial condition: p∗ (t0 ) ∈ Rn . satisfying f adjoint equation: p∗ (t) = −[ ∂f (t. x(t).41) Evidently (8.3. (8. x∗ (t). t0 ≤ t ≤ t∗ . t∗ ] → Ω be optimal.42) is equivalent to (8.31).37) ˜ and (8. ﬁnal condition: p∗ (t∗ ) ∈ Rn .
CONINUOUSTIME OPTIMAL CONTROL Exercise 2: Prove Theorem 2. x2 = x we rewrite the particle dynamics as ˙ x1 (t) ˙ x2 (t) ˙ = 0 1 0 −α x1 (t) x2 (t) + 0 b u(t) . p∗ (t). The control constraint set is Ω = [−1. ˙ Solution: Taking x1 = x.44). We now study a simple example illustrating Theorem 2. (8.49) . p∗ (t) ≡ p∗ (0) .45). x = 0 in minimum time.47) where α = (σ/m) > 0 and b = (1/m) > 0. 1 2 (8.46) hold.47) is 1 0 1 α (1 Φ(t. x ˙ where m = mass. u ∈ R and u(t) constrained by u(t) ≤ 1.48) such that (8. 1 1 αt α (1 − e ) 0 eαt p∗ (0) 1 p∗ (0) 2 . Example 1: The motion of a particle is described by m¨(t) + σ x(t) = u(t) . By Theorem 2 there exists a nonzero solution p∗ (·) of p∗ (t) ˙1 p∗ (t) ˙2 =− 0 0 1 −α p∗ (t) 1 p∗ (t) 2 (8. 1].48) is p∗ (t) 1 p∗ (t) 2 or = e−α(t−τ ) − e−α(t−τ ) ) . v) = (p∗ (t) − αp∗ (t))x∗ (t) + bp∗ (t)v 1 2 2 2 = eαt (p∗ (0) − αp∗ (0))x∗ (t) + pb∗ (t)v . Starting with an initial condition x(0) = x01 .108 CHAPTER 8. τ ) = so that the solution of (8. For simplicity we suppose that x ∈ R. 1 1 and p∗ (t) = 2 The Hamiltonian H is given by H(x∗ (t). (8. and (8. Suppose that u∗ (·) is optimal and x∗ (·) is the corresponding trajectory. and x = position of the particle. x(0) = x02 we wish to ﬁnd an admissible control which brings the ˙ particle to the state x = 0. Now the transition matrix function of the homogeneous part of (8. 1 2 2 2 1 ∗ α p1 (0) 1 + eαt (− α p∗ (0) + p∗ (0)) . u = applied force. σ = coefﬁcient of friction.
we must 1 2 2 1 have in this case p∗ (0) = 0. Case 3.3. 2 ∗ (t) > 0 for t > t. 2 ∗ −1 if p∗ (t) < 0.47) does not depend on t explicitly we must also have eαt (p∗ (0) − αp∗ (0))x∗ (t) + bp∗ (t)u∗ (t) ≡ constant. 2 109 (8.50) Furthermore. p∗ (·) 1 2 (8.50). 2 Case 2. Also since p∗ (t) ≡ 0.49) we see that p∗ (t) must be a strictly 1 2 2 monotonically increasing function so that from (8. <0. −p∗ (0) + αp∗ (0) < 0 : Evidently u∗ (·) can behave in one of two ways: 1 2 either u∗ (t) = or u∗ (t) ≡ −1 and p∗ (t) < 0 for all t. from (8.8. 2 ˆ and p∗ (t) < 0 for t > t. Case 1. First of all since p∗ (t) ≡ 1 can have three qualitatively different forms. u (t) = 2 ? if p∗ (t) = 0 . ˆ ˆ +1 for t > t and p2 u∗ (t) = or u∗ (t) ≡ +1 and p∗ (t) > 0 for all t.50) u∗ (·) can behave in one of two ways: either ˆ ˆ −1 for t < t and p∗ (t) < 0 for t < t. . since the righthand side of (8. Hence u∗ (·) we can behave in one of two ways: 1 either u∗ (t) ≡ +1 and p∗ (t) ≡ 2 or u∗ (t) ≡ −1 and p∗ (t) ≡ 2 1 ∗ α p1 (0) 1 ∗ α p1 (0) ˆ ˆ +1 for t < t and p∗ (t) > 0 for t < t.51) We now proceed to analyze the consequences of (8. 1 2 2 2 p∗ (0). −p∗ (0) + αp∗ (0) = 0 : In this case p∗ (t) ≡ (1/α)p∗ (0).49) and (8. VARIABLE FINAL TIME so that from the maximum principle we can immediately conclude that +1 if p∗ (t) > 0. −p∗ (0) + αp∗ (0) > 0: Evidently then. ˆ −1 for t > t 2 >0.
which is the curve OA in Figure 8. This gives b ξ1 (t) = α (−t + eαt −1 α ) . CONINUOUSTIME OPTIMAL CONTROL Thus.53) (8. p∗ (0) such that the solution of the 1 2 differential equation x = x2 ˙ 1 1 x2 = −αx2 + b sgn[ α p∗ (0) + eαt (− α p∗ (0) + p∗ (0))] . The optimal control is given by u∗ (t) = sgn p∗ (t) 2 1 1 = sgn [ α p∗ (0) + eαt (− α p∗ (0) + p∗ (0))] . One way is to guess at the value of p∗ (0) and then integrate (8. which is the curve OB. If (8. ξ2 (t) = − α (1 − eαt ) . f f There are at least two ways of solving the twopoint boundary value problem (8. f f (8. then u∗ (t) ≡ −1 1 2 2 and we get b ξ1 (t) = − α (−t + eαt −1 α ) b . . (8. An alternative is to guess at the value of p∗ (0) and then integrate (8. ξ2 (t) = b α (1 − eαt ) .52) and (8. x2 (t∗ ) = 0 .52).52) for some t∗ > 0.52) and (8. Integrating (8. 1 1 2 Thus the search for the optimal control reduces to ﬁnding p∗ (0).54) backward in time give us a trajectory ξ(t) where u ˙ ˙ ξ1 (t) = −ξ2 (t) ˙ ξ2 (t) = αξ2 (t) − b . The latter approach is more advantageous because we know that any trajectory obtained by this procedure is optimal for initial conditions which lie on the trajectory. Suppose we choose p∗ (0) such that −p∗ (0) = αp∗ (0) = 0 and p∗ (0) > 0.110 CHAPTER 8.54) is satisﬁed. the optimal control u∗ is always equal to +1 or 1 and it can switch at most once between these two values.54) (8. Then we must have 1 2 2 ∗ (t) ≡ 1.54) backward in time and check of (8.53) is satisﬁed. ˙ 1 1 2 with initial condition x1 (0) = x10 . and then t∗ is the minimum time.54) is not satisﬁed then modify p∗ (0) and repeat.52) and (8. and (8. On the other hand.3.54). Let us follow this procedure. with ξ1 (0) − ξ2 (0) = 0 . if p∗ (0) is such that −p∗ (0) + αp∗ (0) = 0 and p∗ (0) < 0.53) forward in time and check if (8.53). x20 = x20 also satisﬁes the ﬁnal condition x1 (t∗ ) = 0.
Then [(1/α)p∗ (0) + 1 2 2 1 ˆ + p∗ (0))] will have a negative value for t ∈ (0. b for t > t with ξ1 (0) = 0.52). ∞). We see then that the optimal control u∗ (·) has the following characterizing properties: u∗ (t) = 1 if x∗ (t) is above BOA or on OA −1 if x∗ (t) is below BOA or on OB . VARIABLE FINAL TIME u∗ ≡ −1 B C D u∗ ≡ 1 O E 111 ξ1 ξ2 u∗ ≡ 1 A F u∗ ≡ −1 Figure 8. (8. Next suppose p∗ (0) is such that −p∗ (0) + αp∗ (0) > 0. and we get the 2 2 curve OEF . ξ2 (0) = 0. if we integrate (8.4: Optimal trajectories of Example 1. Finally if p∗ (0) is such that −p∗ (0) + 1 ˆ ˆ αp∗ (0) < 0. Hence we can synthesize the optimal control in feedback from: u∗ (t) = ψ(x∗ (t)) where the B u∗ ≡ −1 x2 u∗ ≡ 1 u∗ ≡ −1 O x1 u∗ ≡ 1 A Figure 8. then u∗ (t) = 1 for t < t and u∗ (t) = −1 for t > t. This give us the curve OCD.3. and p∗ (0) < 0. t) and a positive value for t ∈ 2 ˆ (t.8.3: Backward integration of (8.54). and p∗ (0) < 0.52) and (8. .54) backwards in time we get trajectory ξ(t) where eαt (−(1/α)p∗ (0) 1 ˙ ξ(t) = −ξ2 (t) ˙ ξ2 (t) = αξ2 (t)+ ˆ −b for t < t ˆ . Hence.
such that p p∗ (t) = −p∗ (−P (t)x∗ (t)) − A (t)p∗ (t) . The Hamiltonian function is H(t. this will imply 0 u∗ (t) = whereas if p∗ = 0. positive semideﬁnite matrix whereas Q(t) is a p × p symmetric.57) (8. T is a ﬁxed ﬁnal time. x2 ) is below BOA or on OB . Gf is a given f × n matrix.58) (t)p∗ (t) . x2 ) is above BOA or on OA −1 if (x1 . p∗ (t).58) cannot have a maximum. CONINUOUSTIME OPTIMAL CONTROL function ψ : R2 → {1. positive deﬁnite matrix. Quadratic Cost An important class of problems which arise in practice is the case when the dynamics are linear and the objective function is quadratic. bf ∈ R f are given vectors. then we must have 0 p∗ (t) B(t) ≡ 0 because otherwise (8.55): T Minimize 0 1 [x (t)P (t)x(t) + u (t)Q(t)u(t)]dt 2 (8. ﬁnal condition: Gf x(t) = bf . 0 ≤ t ≤ T . x∗ (t). x2 ) = 1 if (x1 . not both zero. so that we must search for a number p∗ ≥ 0 and a function 0 ∗ : [0. control constraint: u(t) ∈ Rp . ˙ initial condition: x(0) = x0 .112 CHAPTER 8. (8.60) 1 −1 p∗ Q (t)B 0 (8.4) ψ(x1 . ˙ 0 and p∗ (t)⊥T f (x∗ (t)) = {ξGf ξ = 0} . Speciﬁcally. 8. −1} is given by (see Figure 8. In (8. consider the optimal control problem (8. 2 0 If p∗ > 0.55) subject to dynamics: x(t) = A(t)x(t) + B(t)u(t). T ] → Rn . We apply Theorem 1 of Section 2. v) = − 1 p∗ [x∗ (t) P (t)x∗ (t) + v Q(t)v] ˜ 2 0 +p∗ (t) [A(t)x∗ (t) + B(t)v] so that the optimal control u∗ (t) must maximize − 1 p∗ v Q(t)v + p∗ (t) B(t)v for v ∈ Rp .56) (8.59) . (8.4 Linear System. and x0 ∈ Rn .56) we assume that P (t) is an n × n symmetric. u(·) piecewise continuous.
For further details regarding the solution of this boundary value problem and for related topics see (See and Markus [1967]). (p∗ (t)/p∗ )) will satisfy all the necessary ˆ 0 0 conditions so that we can assume that p∗ = 1. p∗ > 0. implies ξ = 0 . Then the controllability assumption is equivalent to the statement that for any ξ ∈ Rn ξ Φ(t. then we must have p∗ (t) ≡ 0 which is a ˜ 0 contradiction. THE SINGULAR CASE 113 We make the following assumption about the system dynamics. 8.8. p∗ (T )⊥T f (x∗ (T )) . Thus.60) (p∗ (t)) Φ(T. under the controllability assumption. 0 ≤ τ ≤ T . 0 ≤ t ≤ T initial constraint: k(0) = k0 . Hence if p∗ = 0. t)B(t) = 0 . The problem can be summarized as follows: T Maximize subject to ˙ dynamics: k(t) = s(t)f (k(t)) − µk(t) .61) Next we claim that if the system is controllable then p∗ = 0.5 The Singular Case In applying the necessary conditions derived in this chapter it sometimes happens that H(t. Assumption: The control system x(t) = A(t)x(t) + B(t)u(t) is controllable over the interval ˙ [0. and hence the optimal control is 0 given by (8. The optimal trajectory and the optimal control is 0 obtained by solving the following twopoint boundary value problem: x∗ (t) = A(t)x∗ (t) + B(t)Q−1 (t)B (t)p∗ (t) ˙ p(t) = P (t)x∗ (t) − A (t)p∗ (t) ˙ x∗ (0) = x0 . τ ) be the transition matrix function of the homogeneous linear differential equation x(t) = ˙ A(t)x(t). but then from (8. control constraint: s(t) ∈ [0. v) is independent of v for values of t lying in a nonzero interval. because if p∗ = 0 then from (8. Now if p∗ > 0 it is trivial that p∗ (t) = (1.61) we get p∗ (T ) = 0. (8. τ )B(τ ) = 0 . We illustrate this by analyzing Example 4 of Chapter 1.5. 1]. p∗ (t).) Let Φ(t. ﬁnal constraint: k(t) ∈ R . T ]. 0 ≤ t ≤ T . Gf x∗ (T ) = bf . 0 c(t)dt = T 0 (1 − s(t))f (k(t))dt . s(·) piecewise continuous.56) 0 0 we can see that p∗ (t) = (Φ(T. x∗ (t). t)) p∗ (T ) and hence from (8. (See (Desoer [1970]) for a deﬁnition of controllability and for the properties we use below. In such cases the maximum principle does not help in selecting the optimal value of the control. We are faced with the socalled singular case (because we are in trouble–not because the situation is rare).59).
such that 0 p∗ (t) = −p∗ (1 − s∗ (t))fk (k∗ (t)) − p∗ (t)[s∗ (t)fk (k∗ (t)) − µ] ˙ 0 with the ﬁnal condition p∗ (T ) = 0 . s∗ (t) = 1 : Then the dynamic equations become ˙ k∗ (t) = f (k∗ (t)) − µk∗ (t) . p∗ (t) > 1. 0 ≤ t ≤ T . p)−. CONINUOUSTIME OPTIMAL CONTROL We make the following assumptions regarding the production function f : fk (k) > 0.114 CHAPTER 8. which immediately implies that 1 if p∗ (t) > 1 ∗ s (t) = 0 if p∗ (t) < 1 ? if p∗ (t) = 1 We analyze separately the three cases above. and a function p∗ : [0. (See Figure 8. T ] → R.62) that kG < kM .68) is depicted in the (k. Case 1. fkk (K) < 0 for all k .64) and (8. (k. be the corresponding trajectory of the capitaltolabor ratio. p∗ ) by (1/p∗ )(p∗ . kH are the solutions of fk (kG ) − µ = 0 and f (kM ) − µk = 0.65) we must also 0 have p∗ (t) ≡ 0.63). T ] → [0. we note from (8.66) (8.6.62) k→0 (8.63) is mainly for technical convenience and can be dispensed with without difﬁculty. (8. Here kG . t)−planes in Figure 8. k∗ (t). p∗ (t) = −p∗ (t)[fk (k∗ (t)) − µ] . Assumption (8. ˙ The maximum principle says that H(t.68) (8. lim fk (k) = ∞ . Hence we must have p∗ > 0 and then by replacing (p∗ . p∗ (t).62) says that the marginal product of capital is positive and this marginal product decreases with increasing capital. s) = (1 − s)f (k∗ (t)) + p∗ (t)[sf (k∗ (t)) − µk∗ (t)] is maximized over s ∈ [0. 1] is an optimal savings policy and let k∗ (t). Such solutions exist and are unique by virtue of the assumptions (8. there exist a number p∗ ≥ 0. t)− and (p. Now suppose that s∗ : [0.67) The behavior of the solutions of (8.65) (8. ˙ (8.64) simpliﬁes to 0 p∗ (t) = −1(1 − s∗ (t))fk (k∗ (t)) − p∗ (t)[s∗ (t)fk (k∗ (t)) − µ] . if p∗ = 0 then from (8. and fk (k) − µ > 0 according as k > kG whereas f (k) − µk < 0 according as k > kM . Then by Theorem 1 of Section 2. 1] at s∗ (t). so that (8. not both identically zero. First of all.) < < < > .62) and (8. (8.64) and the maximum principle holds. p∗ ) we 0 0 0 0 can assume without losing generality that p∗ = 1. Futhermore.5.63) Assumption (8.
8.70) . so −fk (k∗ (t)) + µ = 0 for t ∈ I .66) ˙ we get −(1 − s∗ (t))fk (k∗ (t)) − [s∗ (t)fk (k∗ (t)) − µ] = 0 for t ∈ I . But then we have p∗ (t) = 0 for t ∈ I so that from (8. or k∗ (t) = kG for t ∈ I .7.5. p∗ (t) < 1. where I is a nonzero interval. Case 3. (8.) Evidently if p∗ (t) = 1 only for a ﬁnite set of times t then we do not have to worry about this case. p∗ (t) = 1. p∗ (t) = −fk (k∗ (t)) + µp∗ (t) . s∗ (t) = 0: Then the dynamic equations are ˙ k∗ (t) = −µk∗ (t) . s∗ (t) =?: (Possibly singular case. ˙ giving rise to the behavior illustrated in Figure 8. THE SINGULAR CASE 115 p l fk > µ fk < µ k kM f > µk kG p kM f < µk k t l t Figure 8.5: Illustration for Case 1. kG s∗ (t) = µ f (kG ) for t ∈ I . ˙ In turn then we must have k∗ (t) = 0 for t ∈ I so that s∗ (t)f (kG ) − µKG = 0 for t ∈ I .69) (8. Case 2. We face the singular case only if p∗ (t) = 1 for t ∈ I. and hence.
1] such that ˆ ˆ − µk = 0. In particular we must have k < k . or (B) there exists t2 ∈ (0. (8. CONINUOUSTIME OPTIMAL CONTROL µk line of slope µ f (k) .69) and (8. T ] and then s∗ (t) = 0.6: Illustration for assumptions (8. p∗ (t) < 1 so that we are in Case 2.63). T ) such that p∗ (t3 ) = 1. s 2 0 G (Bii) k∗ (t2 ) > kG : then p∗ (2 ) > 0 but then p∗ (t2 + ε) > 1 for ε > 0 sufﬁciently small and since ˙ p∗ (T ) = 0 there must exist t3 ∈ (t2 . This contradicts the deﬁnition of t2 so that this possibility cannot arise. The reason for this term is contained in the following exercise. The capitaltolabor ratio kG is called the golden mean and the singular solution is called the golden path.62). T ) such that p∗ (t2 ) = 1 and p∗ (t) < 1 for t2 < t ≤ T . For t < t1 we either have p∗ (t) > 1. Show that kG is the unique sustainable capitaltolabor ratio which maximizes ˆ sf (k) ˆ sustainable consumption (1 − s)f (k). We face two possibilities: Either (A) p∗ (t) < 1 for all t < [0. . or we have p∗ (t) < 1. We can now assemble separate cases to obtain the optimal control. The various possibilities are illustrated in Figure 8. s∗ (t) > 1 if k0 < kG . for 0 ≤ t ≤ T .116 f CHAPTER 8. from the ﬁnal condition (8. as in Figure 8. s∗ (t) = 0 if k > kG . Thus in the singular case the optimal solution is characterized by (8. k∗ (t) = kG . k∗ (t) = k0 e−µt . First of all.70). and s∗ (t) = µ(kG /f (kG )) for t ∈ (t1 . k kG kM Figure 8.65) we know that for t close to T.9. t2 ) so that p∗ (t) = 1. (Biii) k∗ (t2 ) − kG : then we can have a singular arc in some interval (t1 . We then have three possibilities depending on the value of k∗ (t2 ): (Bi) k∗ (t2 ) < kG : then p∗ (t2 ) < 0 so that p∗ (t) > 1 for t < t2 and we are in Case 1 so that ˙ ∗ (t) = 1 for t < t .8. ˆ Exercise 1: A capitaltolabor ratio k is said to be sustainable if there exists s ∈ [0. t2 ).
7: Illustration for Case 2. On the one hand these include extensions to inﬁnitedimensional state spaces and on the other hand they allow for constraints on the state more general than merely initial and ﬁnal constraints. treatment see (Neustadt [1969]). Several important generalizations of the maximum principle have appeared. and (Balakrishnan and Neustadt [1964]). et al.6 Bibliographical Remarks The results presented in this chapter appeared in English in full detail for the ﬁrst time in 1962 in the book by Pontryagin. [1968]). For applications of the maximum principle to optimal economic growth see (Shell [1967]).. et al. That book contains many extensions and many examples and it is still an important source.6. [1967]). but mathematically difﬁcult. (Kelley [1962]). also consult (Jacobson and Mayne [1970]). Among the many useful techniques which have been proposed see (Lasdon. et al. 8.. cited earlier. [1971]).8. (McReynolds [1966]). For an applicationsoriented treatment of this subject the reader is referred to (Athans and Falb [1966]) and (Bryson and Ho [1969]). the derivation of the maximum principle given in the book by Lee and Markus is more satisfactory. BIBLIOGRAPHICAL REMARKS p k 117 l k kG p t l t Figure 8. For a less rigorous treatment of statespace constraints see (Jacobson. . et al. For a uniﬁed. whereas for a discussion of the singular case consult (Kelley. and (Polak [1971]). However. There is no single source of computational methods for optimal control problems.
8: Case 3. CONINUOUSTIME OPTIMAL CONTROL p k 1 .118 CHAPTER 8. k kG p kG t 1 t Figure 8. The singular case. .
t Case (Biii) .6.8. t t2 T s∗ 1 s∗ µkG f (kG ) t k∗ kG k0 t k∗ .9: The optimal solution of example. t Figure 8. BIBLIOGRAPHICAL REMARKS 119 p∗ 1 t s∗ 1 t k∗ T T p∗ 1 t s∗ 1 t2 T t k∗ t2 T t Case (A) p∗ T p∗ Case (Bi) t2 T t . . . t1 . . t t1 t2 T . .
120 CHAPTER 8. CONINUOUSTIME OPTIMAL CONTROL .
the necessary conditions for optimality were obtained by comparing the optimal decision with decisions in a small neighborhood of the optimum.e. Finally f0 (i. 1. 1. . i.1 Discretetime DP We consider a problem formulation similar to that of Chapter VI. N − 1 . the state x(i) and the control u(i) belong to arbitrary sets X and U respectively. or ﬁnitedimensional vector spaces (as in the previous chapters). However.. therefore. 9. or even inﬁnitedimensional spaces. leads to optimality conditions which are sufﬁcient. initial condition: x(0) = x0 . x0 ∈ X is ﬁxed. N −1 Maximize i=0 f0 (i. x(i). . u(i)) . for notational convenience we neglect ﬁnal conditions and statespace constraints. i = 0. . This global comparison. The only disadvantage (which unfortunately often rules out its use) of DP is that it can easily give rise to enormous computational requirements. . ·) : X × U → R. ·. . . Φ : X → R. The main advantage of DP.Chapter 9 Dynamic programing SEQUENTIAL DECISION PROBLEMS: DYNAMIC PROGRAMMING FORMULATION The sequential decision problems discussed in the last three Chapters were analyzed by variational methods. Some general remarks and bibliographical references are collected in the ﬁnal section. ·) : X × U → X are ﬁxed functions. In the ﬁrst section we develop the main recursion equation of DP for discretetime problems. X and U may be ﬁnite sets. i = 0.1). The Ωi are ﬁxed subsets of U . . f (i. besides the fact that it give sufﬁciency conditions. ·. x(i). control constraint: u(i) ∈ Ωi .1) subject to dynamics: x(i + 1) = f (i. The second section deals with the continuoustime problem. 121 . In (9. N − 1 . u(i)) + Φ(x(N )) (9. Dynamic programming (DP is a technique which compares the optimal decision with all the other decisions. . is that DP permits very general problem formulations which do not require differentiability or convexity conditions or even the restriction to a ﬁnitedimensional state space.
. initial condition: x(k) = x. . . k ≤ ≤ N − 1. i = k.1).x∗ ( ) . . . u(i)) + Φ(˜(n)) ˜ ˜ x i=k −1 = > i=k f0 (i. . N − 1. u(i)) + Φ(ˆ(N )) ˆ ˆ x i= i=k N −1 f0 (i. u(N − 1) with ˜ ˜ u(i) ˜ u∗ (i) . . control constraint: u(i) ∈ Ωi . N − 1 . . such that ˆ ˆ ˆ N −1 f0 (i. . ˆ The value of the objective function corresponding to this control for the problem (9. . . x( + 1). N − 1 . u( + 1). x∗ (i). .2)k. into a family of optimal control problems with the same dynamics. x∗ (k + 1). . u∗ ( ). . . . x(i). u∗ (i)) + N −1 f0 (i. DYNAMIC PROGRAMING The main idea underlying DP involves embedding the optimal control problem (9. We begin with an elementary but crucial observation. − 1 u(i) . . More precisely. x(N ). . . . . .x is N −1 f0 (i. we will sometimes use the index (9. x(i) . . .2)k. ·. . . . ˆ and the corresponding trajectory. . i = . ∗ ∗ ∗ > i= But then consider the control u(k). Since the initial time k and initial state x are the only parameters in the problem above. Lemma 1: Suppose u∗ (k). . u(i)) + Φ(x(N )) . x∗ (i).122 CHAPTER 9.2) . . is x(k). . . u(i)). u (i)) + Φ(x (N )) . u∗ (N − 1) is an optimal control for (9. and control constraint as in (9. .2) subject to dynamics: x(i + 1) = f (i. k + 1. . i = k. i = + 1.3) f0 (i. . k + 1. x(i). u(i)) + Φ(ˆ(N )) ˆ ˆ x i= N −1 (9. x∗ (N ) be the corresponding optimal trajectory. . . . consider the following problem: N −1 Maximize i=k f0 (i. i = k. . . . . in which the system starts in state x0 at time 0. . u∗ (N − 1) is an optimal control for (9. and let x∗ (k) = x. x(i). N . . x(i). Proof: Suppose not. . . i = k. x (i). u∗ (i)) + Φ(x∗ (N )) . .x to distinguish between different problems.1) but with different initial states and initial times.x . u(N − 1). . Then for any . x(i). starting in state x at time k.2)k. Then there exists a control u( ). x(N ) where ˜ ˜ x(i) = ˜ x∗ (i) . (9. for each x ∈ X and k between ) and N − 1. with corresponding ˆ ˆ ˆ trajectory x( ) = x∗ ( ). . objective function.
u(i)) + Φ(x(N )) = f0 (k. and all x ∈ X. u∗ (N − 1) deﬁned by u∗ ( ) = ψ( . ·). . u( )) + V ( + 1. xu∗ (k)) + V (k + 1. N − 1.. x(N ). . .2)k. x)) = Max{f0 (k. k ≤ ≤ N − 1 . x( ). . for all u(k) ∈ Ωk . contradicting the hypothesis. . and equality holds for all k ≤ ≤ N − 1 if and only if the control is optimal for (9. . . x(i). .4). . ≥ i=k By Lemma 1 the lefthand side of (9. Theorem 1: Deﬁne V (N. V (k. ))u ∈ Ωk }. u.9. x(i).x . . u(k)) i=k N +{ i=k+1 f0 (i. x∗ .x and let x(k) = x. u∗ (N − 1) cannot be optimal for 9. . . . (k)) + V (k + 1.x(k+1) . . f (k. let ψ(k.(end theorem) Corollary 1: Let u(k). . x. x. . which is equivalent to (9.5) is equal to f0 (k. u))u ∈ Ωk } . Let V (k. x)) + V (k + 1. f (k. x. by the deﬁnition of V we have N −1 f0 (i. . x∗ (N ) be the corresponding trajectory be x(k) = x. Then V ( . . . . ·) by (V (N. u∗ (k)) . f (x. u(i)) + Φ(x(N )) ≤ f0 (k. DISCRETETIME DP 123 by (9. Corollary 2: For k = 0. u(N − 1) is optimal for (9. . x∗ (i).5) f0 (i.1. . ·) : X → Ωk be such that f0 (k. u) + V (k + 1. u) + V (k1 . . x. . x. x. u∗ (k)) + V (k + 1. u( )). .x exists for all 0 ≤ k ≤ N − 1. . N − 1 is an optimal feedback control.2)k. . . We call V the (maximum) value function. .2)k. k ≤ ≤ N − 1.2)k. k. On the other hand. . u∗ (i)) + Φ(x∗ (N )) (9. ψ(k.2)k . x. ψ(k. . with equality if and only if u(k + 1). . u∗ (N − 1) be an optimal control for (9.2)k+1. let u∗ (k). x the control u∗ (k). . u(k)) + V (k + 1. . . (k. x.x . f (k. u(i)) + Φ(x(N )) . f (k. u(k))} .x . f (k. f (k. u. k = 0. x. u(N − 1) be any control for the problem (9. x(N ) be the corresponding trajectory. for any k. 1.2)k.e. x. x) = Φ(x). i. f ( . .3). x. We have N −1 i=k N −1 f0 (i. 0 ≤ k ≤ N − 1 . x) satisﬁes the backward recursion equation V (k. x) be the maximum value of (9. x( ). u(k))) .4) Proof: Let x ∈ X. where . and let x∗ (k) = x. . x∗ ( )). so that u∗ (k). Then ψ(k. u∗ (k))) ≥ f0 (k. . . x. (9. x(i). Combining these two facts we get f0 (k. x. . (end theorem) From now on we assume that an optimal solution to (9. x( )) ≤ f0 ( . x) = Max{f0 .
x(t + ∆))u : [t. x) = Max{ t t+∆ f0 (τ. ψ( . But now suppose X = Rn and we approximate each dimension of x by 20 values. How many units of each item should be placed in each knapsack so as to maximize total utility? Formulate this problem by DP. unless we can ﬁnd a “closedform” analytic solution to (9. u ∈ Rp . These numbers represent the relative utility of that item during the hike. tf ] → Ω and u(·) piecewise continuous.4) analytically. which is quite impractical for existing computers. and for n = 5 it is 32. t + ∆] → Ω}.4). u(τ ))dτ (9.4) allows us to compute the value function. which is a reasonable amount. x∗ (k) = x . x(t). x) = Φ(x) . Then it is easy to see that V must satisfy V (t.2 Continuoustime DP We consider a continuoustime version of (9. x(τ ). x∗ ( ). He assumes that each person can take up to W pounds in his knapsack. for t0 ≤ t ≤ tf and x ∈ Rn .7) +V (t + ∆. x) be the maximum value of the objective function over the interval [t.1.124 CHAPTER 9. However.2): Maximize 0 f f0 (t.6) In (9.000.4) we also obtain the optimum feedback control. u(t))dt + Φ(x(tf )) subject to dynamics: x(t) = f (t.1. we have to compute and store 10x(20)n values of V . As before.8) . • Exercise 1: An instructor is preparing to lead his class for a long hike. k ≤ ≤ N − 1 . let V (t. The instructor assigns a number Ui > 0 for each unit of item i. ∆ ≥ 0 . tf ] starting in state x at time t. x ∈ Rn .x . Remark: Theorem 1 and Corollary 2 are the main results of DP.000. control constraint: u : [t0 . u(t)) . Φ : Rn → R is assumed differentiable and f0 . This “curse of dimensionality” seriously limits the applicability of DP to problems where we cannot solve (9. For n = 3 this number is 80. t (9. f are assumed to satisfy the conditions stated in VIII. and in evaluating the maximum in (9. For instance. 9. and V (tf . (9. Note that this feedback control is optimum for all initial conditions.6). DYNAMIC PROGRAMING x∗ ( + 1) = f ( . There are N possible items to choose from. Each unit of item i weighs wi pounds. x∗ ( )). the DP formulation may necessitate a prohibitive amount of computation since we would have to compute and store the values of V and ψ for all k and x. The recursion equation (9. x(t). Then for N = 10. Then we have to compute and store 10 × 20 values of V .000. Ω ⊂ Rp . is optimal for (α)k. suppose n = 10 and the statespace X is a ﬁnite set with 20 elements. t0 ≤ t ≤ tf ˙ initial condition: x(0) = x0 .
x)) ∂x = Max{f0 (t. x. x. x)) + ∂V f (t. ψ(t. and V is the value function. u∗ (τ ))dτ + Φ(x∗ (τ )) f0 (τ. (τ )). (9. ψ(τ.2. u)∆ + V (t.7). (9. x∗ (τ ). Proof: Let t ∈ [t0 . u) + ∂V ∂x (t. u(τ ))dτ + Φ(ˆ(tf )) .8). x) + Max{f0 (t. Let u : [t. ∂t 125 Dividing by ∆ > 0 and letting ∆ approach zero we get the HamiltonJacobi. t ≤ τ ≤ t + ∆ . x. t ≤ τ ≤ tf .Bellman partial differentiable equation for the value function: ∂V ∂t (t. ˙ x(t) = x . u) + ∂V f (t. satisfying f0 (t. u∗ (τ ))dτ . ˆ x (9. x. ψ(t. Let us suppose that V is differentiable in t and x.14) F − 0(τ. x. x)f (t. tf ] × Rn → R which satisﬁes (9. x∗ (τ ))) . x∗ (τ ))dτ dτ ∂V ∂V ∗ { (τ.6). x∗ (τ ) + x (τ )}dτ ˙ ∂τ ∂x t f (9. x) + ∂V f (t. x) = Max{f0 (t. tf ] and x ∈ Rn . x(τ ). Let u∗ (τ ) = ψ(τ. x. x∗ (τ ). tf ] → Ω be any piecewise continuous control and ˆ let x(τ ) be the solution of ˆ x (τ ) = f (τ. t ≤ τ ≤ tf . x(τ ) is the solution of x(τ ) = f (τ. ˆ Let x∗ (τ ) be the solution of x∗ (τ ) = f (τ.9) Theorem 1: Suppose there exists a differentiable function V : [t0 . x(τ ). x∗ (t)) = f tf =∈t tf =− dV (τ. x)∆ + o(∆)u ∈ Ω}. ∆ > 0 . ˆ ˆ ˆ x(t) = x . tf ] × Rn → Ω with ψ piecewise continuous in t and Lipschitz in x. ∂x (9. u(τ )) . CONTINUOUSTIME DP In (9.9) and the boundary condition (9.7) we get V (t.13) ≤ To this end we note that V (tf . x∗ (tf )) − V (t. u)u ∈ Ω} = 0. x. u)u ∈ Ω} . x∗ (τ ). Then from (9. x.10) Then ψ is an optimal feedback control for the problem (9. u(τ )) . ˙ x∗ (τ ) = x .9.12).12) · (9. t ≤ τ ≤ tf . Suppose there exists a function ψ : [t0 . t . x∗ (τ ). To show that ψ is an optimal feedback control we must show that tf t tf t f0 (tτ. u)∆ ∂x + ∂V (t.11) Note that the hypothesis concerning ψ guarantees a solution of (9. x∗ .
x(τ )) + ˆ ∂τ ∂x f0 (τ.15).3 Miscellaneous Remarks There is vast literature dealing with the theory and applications of DP. x. (9. x(τ ). From (9. x(τ ). x(tf )) − V (t. On the other hand.13) is proved. ˆ ˆ (9.9). The most elegant applications of DP are to various problems in operations research where one can obtain “closedform” analytic solutions to be recursion equation for the value function.9). control constraint: u(t) ∈ Rp . DYNAMIC PROGRAMING { ≤− t tf ∂V ∂V · x (τ )}dτ ˜ (τ. u(τ ))dτ ˆ ˆ ♦ so that (9. For an excellent introduction to this area of application see (Howard [1960]). For an important application of DP to computational considerations for optimal control problems see (Jacobson and Mayne [1970]).] 9. Finally.14). x) = Φ(x∗ (tf )) + ≥ Φ(ˆ(tf )) + x t tf t tf f0 (τ. • Exercise 1: Obtain the value function and the optimal feedback control for the linear regulatory problem: Minimize 1 x (T )P (T )x(t) + 2 1 2 T0 {x (t)P (t)x(t) +u (t)Q(t)u(t)}dt subject to dynamics: x(t) = A(t)x(t) + B(t)u(t) . u∗ (τ )) f0 (τ. Larson [1968] has developed computational techniques which greatly increase the range of applicability of DP where closedform solutions are not available. x) = x R(t)x where R is unknown. (9.8) and the fact that x∗ (t) = x(t) = x we conclude that ˆ V (t. (t)) = ˆ ˆ t tf CHAPTER 9. and Q(t) = Q (t) is positive deﬁnite. where P (t) = P (t) is positive semideﬁnite. [] . x) and try a solution of the form V (t.15) using (9. the book of Bellman [1957] is still excellent reading. (9. x∗ (τ ). 0 ≤ t ≤ T . V (tf . [Hint: Obtain the partial differential equation satisﬁed by V (t. It also follows that V is the maximum value function.126 using (9. See (Bellman and Dreyfus [1952]) and (Wagner [1969]). u∗ (τ ))dτ . ˙ initial condition: x(0) = x0 . In the case of sequential decisionmaking under uncertainties DP is about the only available general method.10).
W. [6] R. Bruns. Princeton University Press. IEEE Trans. IEEE Trans. [8] J. 1960. 1954. 1962. Desoer. 1971.D. Falb. [15] S. Girshick. 1970. 1969a. McGrawHill. [9] A.A.A. 1964.W.W. Dreyfus. 1966. [14] S.A. Dynamic Programming. Academic Press. Director and R. Princeton University Press. and E. 1963. The generalized adjoint network and network sensitivities. Director and R. Bellman and S. [7] D. on Circuit Theory. CT16(3). Decentralization and Computation in Resource in resource Allocation. John Wiley. Hurwicz. Rohrer. [4] K. Banerjee. Neustadt. CT16(3). Generalized Lagrange Multipliers in Dynamic Programming. Automated network design–the frequencydomain case. Blaisdell.). Dantzig. [3] A.Bibliography [1] J. [13] S. University of California.L. [10] J.W. Optimal Control. on Circuit Theory. SSC6(4). on Circuit Theory. University of North Carolina Press. McGrawHill. Rohrer. Director and R. Applied Optimal Control. 1969b. On the design of resistance nport networks by digital computer. IEEE Trans. in Pfouts R.W.E. Blackwell and M.E. PhD thesis. Polak. C. [12] C. on Systems Science and Cybernetics. Bellman. Theory of Optimal Control and Mathematical Programming. 1957. Arrow and L. Applied Dynamic Programming. Theory of Games and Statistical Decisions. [2] M.J. Essays in Economics and Econometrics. IEEE Trans. Princeton University Press. Linear Programming and Extensions. Athans and P.C.E. Cullum.E. Rohrer. Van Nostrand Reinhold.E. Computing Methods in Optimization Problems. Balakrishnan and L. 1970. Cannon. The function of operations research specilists in large urban schools.D. (ed. Berkeley. Notes for a Second Course on Linear Systems. [11] G. College of Engineering. 1969c. Bryson and Y. 1970.A. 127 . Ho.A. 1969. CT16(3). [5] R.V.
H. John Wiley. Rinehart. [25] F. Surrogate mathematical programming. volume 1. P.Kushner. Rohrer. Karlin.R. G.J. Isaacs. [30] D. Markham Publishing Co. M. 1971. 1960. Duality in Nonlinear Programming: a Simpliﬁed ApplicationOriented Treatment.(ed. Public Expenditures and Policy Analysis. Math. [18] R. Lele. 1969. XXXIV(1).J. on Circuit Theory. 1965. 18. Automated design of biasing circuits. Mayne. Fleming.Q. 1958. Samuelson. and J. J. Programming. 1965. AddisonWesley. Introduction to Stochastic Control.H. Memo RM6134PR. Dorfman and H. A geometric duality theorem with economic applications. and Winston. and Margolis.L.M. AddisonWesley. 1971.A. Linear Programming and Economic Analysis. Operations Research. [22] D. Review of Economic Studies.). 1970.J. 1962. [20] W. Differential Dynamic Programming. Analysis and Applications. Functions of Several Variables.M.H. [19] Dowles Foundation Monograph.D. Math. Production Theory and Indivisible Commodities. Dynamic Programming and Markov Processes. 1967. 1970b. IEEE Trans. CT18(1). John Wiley. Optimization Techniques. Academic Press. Dowell and R. 1970a. Geoffrion. [21] C. Method of Gradients. 18. [28] R. 1971. The Economic Theory of Teams. [33] H. [17] R.A. Holt. Princeton University Press. R.A. and Economics. 1970. The Rand Corporation. Operations Research. Geoffrion. Jacoby. 1969.P. to appear. Greenberg and W. [31] D. Gale. [26] J.). MIT Press.I.H. Dorfman. [24] A. Howard. American Elsevier Publishing Co. and R. 1970. in Leitmann. Speyes. McGrawHill. [29] R.M. . A Model of Public Decisions Illustrated by a Water Pollution Policy Problem. Frank. 1959. New necessary condtions of optimality for problems with statevariable inequality constraints. Mathematical Methods and Theory in Games.M. Pierskalla. to appear. [23] A. Extensions of lagrange multipliers in nonlinear programming. Gould. Differential Games.J. Jacobson and D. [27] H. 1971. Kelley.Solow. Primal resource directive approaches for optimizing nonlinear decomposable programs. (eds. SIAM J. Apl.128 BIBLIOGRAPHY [16] R. J. in Haveman. Jacobson.. 17. [32] S.
[52] E.B. 1967.F. G. The successive sweep method and dynamic programming. Game Theory. The Simplex Method for Local Separable Programming. Larson.H. American Elsevier Publishing Co. Louis.G. Control Conf. Analysis and Applications. Waren. [42] D.D. Boltyanski.). IEEE Trans. Raiffa. 1962. R. Lee and L. Miller. constraints. Pontryagin. 1967. John Wiley. Objectives. 1968. R. 1967. AC12(1).W. Math. [49] L. Statistics and Probability. McGrawHill. Nonlinear programming. Academic Press. [53] L. [37] H. in Leitman. Markus. W. Nonlinear Programming. Foundation of Optimal Control Theory. [48] L. 1969. Water quality regulation with multiple polluters. 19.G. Academic Press. Polak.BIBLIOGRAPHY 129 [34] J. . Recent Advance Programming. In Proc. (ed. Mesarovic. 1967.A.D. and H. and C. and E. S. Math. Rao. [36] T.. [47] C. University of California Press. Singular Extremals. [51] G. J. P. Luenberger.S. Autom.. 1963.Mischenko. State Increment Dynamic Programming. J. [46] M. (eds. Econometrica. 7. Saunders & Co. Second Berkeley Symp. McGrawHill. on Automatic Control. 1957.B. D. August 1113 1971. McReynolds.. Macho. [40] E. Topics in Optimization. 1970. 3(1).K. Owen. Nikaido. Analysis and Applications. [50] H. and outcomes in optimal growth models. Applied Math. 1969. 1968. In Proc. Multilevel Systems. [41] R. Convex Structures and Economic Theory.W. Kuhn and A. Takahara. John Wiley. The conjugate gradient method for optimal control problems.C. 1970. and Wolfe. [44] S. Tucker. Computational Methods in Optimization: A Uniﬁed Approach.S. Computer and System Sciences. Wells. Academic Press. [43] O.W. Luce and H.V. Games and Decisions. [35] D. 1971 Jt. Kendrick.E. McGrawHill. The existence of optimal controls in the absence of convexity conditions. and A. J. Theory of Hierarchical. Academic Press.E. A general theory of extremals. Neustadt. 1968.. Mayer. [39] L. Washington U.E. H.).S. [38] R. 1968. R. Interscience. 1963. Mitter. Neustadt. Berkeley. Stochastic Optimal Linear Estimation and Control. in Graves. Meditch. 1969.S. The Mathematical Theory of Optimal Processes. Kelley. 1971. R.R. 16. 35(1).W. on Math. Koopmans. Kopp. Quasiconvex programming. Siam J.L. Mangasarian. St. Lasdon.L. [45] J. and Y. 1951.H. Gamkrelidze.V.
Shell. Science. The economist’s approach to pollution and its control. PrenticeHall. Princeton University Press. SIAM J.M. Sakarovitch. 173(3996). 1970. Wonham. 1968. Econometrica. on Control. The simplex method for quadratic programming. Notes on Linear Programming. 6(2). SIAM J. The Foundation of Statistics. 1959. 1969. 1971. PrenticeHall. [60] H.M. [59] D. 1967. 1969. Wagner. Convex Analysis. 5(2). On the seperation theorem of optimal control. [61] P.M. Van Nostrand Reinhold. Rockafeller. [56] L.T.M. Zangwill. Principles of Operations Research. BIBLIOGRAPHY [55] M. [58] R. Solow. MIT Press. . 1971. Essays in the Theory of Optimal Economic Growth. on Control.130 [54] R. John Wiley. Savage. Topkis and A. [63] E. Veinnott Jr.J. [57] K. 1954. 1967. 27. Nonlinear Programming: A Uniﬁed Approach. [62] W. Wolfe. On the convergence of some feasible directions algorithms for nonlinear programming.
5 Gradient. 123 Discretetime optimality control sufﬁcient condition. 49 Game theory . 21 Linear programming. 77 sufﬁcient condition. 45. 54. 8 Design of resistive network. 91 discretetime. 37 properties. 34 Constraint qualiﬁcation deﬁnition. 39 Certaintyequivalence principle. 33. 125 ˜ Hamiltonian H. 37 Derivative. 61 Equilibrium of an economy. 101. 35 problem formulation. 53 sufﬁcent conditions. 8 HamiltonJacobiBellman equation. 32 Feasible direction. LP duality theorem. 34 Maximum principle continuoustime. 85 Adjoint equation augmented. 80 Minimum fuel problem. 107 . 39 basic variable. 61 Knapsack problem. 5 Complementary slackness. 80 Dual problem. DP optimality conditions. 101.LP optimality condition. 91. 42 Linear programming. 54 Langrangian multipliers. 64 Farkas’ Lemma. 33. 91. 23 Afﬁne function. 15 Discretetime optimal control necessary condition. 105 continuoustime. 86. 33. 72 algorithm. 58 131 Duality theorem. 31 theory of the ﬁrm. 21. 81 Minimumtime problem.Index Active constraint. 101 Hypograph. 123. 121. 80 Adjoint network. 78. 78 problem formulation. 67 Convex function deﬁnition. 55 Continuoustime optimal control necessary condition. 103 sufﬁcient condition. 103 discretetime. 33. 101.H. 55 Convex set. 71 Feasible solution. 54 Basic feasible solution. 124 Epigraph. 103 problem formulation. 124 Lagrange multipliers. 35 Langrangian function. 50 Adjoint Equation augmented. 37 Lagrangian function. 63 Dynamic programming. 125 problem formulation. 98 continuoustime. 99 ˜ Hamiltonian HH. 125 Control of water quality. 37.
70 Wolfe algorithm. 70 problem formulation. 108 Nondegeneracy condition. 45. 49 suﬁcient condition. 112 Resource allocation problem. 84 Tangent. 80 Value function. 17 sufﬁcient condition. 33. 41 Phase II. 21 Optimum tax. 11 sufﬁcient condition. 124 Regulator problem. 5 Shadow prices. 58 Wolfe algorithm. 125 Optimization over open set necessary condition. 2. 54 Optimal decision. 39 Simplex algorithm. 113 Slack variable. 4 Optimization with equality constraints necessary condition. 1 Optimal economic growth. 50. 81. 91 discretetime problem. 50 Transversality condition continuoustime problem. 63 necessary condition. NP duality theorem. 53 problem formulation. 71 INDEX . 33 Quadratic cost. 32 Statespace constraint continuoustime problem. 60 Supergradient. 103 Vertex. 13 Optimization under uncertainty. 39 Singular case for control. 71 Recursion equation for dynamic programming.132 example. 77 Subgradient. 112 Quadratic programming. 70 Primal problem. 113. 123 Variable ﬁnal time. 37. QP optimality condition. 81. 60 Supporting hyperplane. 73 Separation theorem for stochastic control. 61. 70 Shadowprices. 39 Nonlinear programming. 37 Phase I. 65 Separation theorem for convex sets. 123. 38 Weak duality theorem. 117 Optimal feedback control. 117 discretetime problem.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.