Escolar Documentos
Profissional Documentos
Cultura Documentos
INTRODUCTION
TO
OPERATIONS
RESEARCH
JOSEPH G. ECKER
MICHAEL KUPFERSCHMID
Rensselaer Polytechnic Institute
;1
t..
Copyright
Published
simultaneously
in Canada.
658.4'03'4
in the Republic
of Singapore
10 9 8 7 6 5 4 3
87-23147
Michael.
ABOUT THE
AUTHORS
,
IJ
(
,
I
is Professor and Chairman of the Department of Mathematical Sciences at Rensselaer Polytechnic Institute in Troy, New York. All of
his degrees were received from the University of Michigan-a B.A. in mathematics in 1964, an M.S. in mathematics in 1966, and a Ph.D. in mathematics in
1968. Dr. Ecker has been a faculty member at RPI since 1968 and teaches a
variety of courses in operations research and mathematics. He is also an affiliated
faculty member in RPI's Department of Decision Sciences and Engineering
Systems. In 1975 he was awarded a Fulbright-Hays Research Award and a
NATO Postdoctoral Fellowship in Science, and spent the year as a visiting
professor in the Center for Operations Research and Econometrics at the Catholic University of Louvain in Belgium. In 1983 he was a visiting professor in the
mathematics department of the Ecole Polytechnique Federale de Lausanne in
Switzerland. His research interests are in operations research and include the
theory and applications of linear and nonlinear programming, multiple objective
programming, geometric programming, and algorithm development. He is the
author or coauthor of more than 45 research publications and is a consultant
for several major corporations. Dr. Ecker has been an associate editor of Operations Research and is currently an editor of SIAM Review.
JOSEPH
G. ECKER
KUPFERSCHMID is a computing
nology Services Department at Rensselaer
faculty member in RPI's Department of
Systems. He received a B.S. in electrical
MICHAEL
viii
worked for the next three years at Sikorsky Aircraft designing and flight testing
helicopter autopilots. After returning to RPI for a Master of Engineering degree,
which was awarded in 1972, he studied theatre engineering at the Yale School
of Drama and worked for six more years in industry as a control systems engineer.
In 1978 he resigned his position as design supervisor for the controls division of
the J. R. Clancy Company and returned to RPI for graduate work in operations
research and statistics, leading to a M.S. in 1980 and a Ph.D. in 1981. Dr.
Kupferschrnid has taught courses in operations research and in computing at
RfI and is an author of seven research papers. His research interests are in the
experimental evaluation of algorithm performance, the development of nonlinear optimization methods, and the applications of mathematical programming.
Dr. Kupferschmid is a registered Professional Engineer.
PREFACE
This textbook is intended for use in a two-semester sequence of courses introducing the mathematical methods of operations research. Part I can also be used
alone for a one-semester course on linear programming. We have chosen to
provide deep and thorough coverage of the most important methods in operations research rather than a superficial treatment of a larger number of topics.
The level of exposition is appropriate for juniors and seniors who are majoring
in engineering, computer science, mathematics, and quantitative methods in
management.
The basic techniques of operations research are simple and straightforward,
and only a small amount of advanced mathematics is needed for a technically
accurate introduction to the subject. This textbook assumes a knowledge of high
school elementary algebra and a familiarity with simple matrix notation such as
would be introduced in the first class of an undergraduate linear algebra course.
In addition, Chapters 9 and 10 assume a knowledge of elementary differential
calculus, and Part III assumes a knowledge of elementary probability and sta.tistics. A concise appendix on matrix notation makes the book accessible to
students who have not previously had any linear algebra.
Both the style of exposition and the mathematical notation have been chosen
to reflect the simplicity of the subject, and the readability of the text is considered
more important than rigor. Examples are used extensively to introduce and
motivate the topics. In this way, the presentation reflects the inductive process
of scientific discovery rather than imitating the retrospective deduction that is
typical in research papers. This approach gives the student the opportunity to
LX
PREFACE
rediscover the important results personally instead of merely reading about them
in theorems. Proofs are given for some results, but only after the result has been
illustrated by example, and only when the proof provides a constructive method
for solving problems. Thus the book is not a treatise on mathematical theory.
The simplicity of the methods used in the book means that they can be
deeply understood even by beginning students of the subject, and the treatment
is mathematically precise even though the results are often stated informally.
Thus the discussion is not a cookbook tabulation of trite formulae, and the
I
stydent should reasonably be expected to understand the mathematical basis for
the techniques in addition to being able to apply them.
Answers to selected exercises are given at the end of the book, and a
separate answer book is available that contains complete solutions to all of the
exercises.
The development of this text benefited greatly from the comments and
suggestions made by our colleagues during its use in preliminary form over the
past four years. In particular, we express our gratitude to Professors Carlton E.
Lemke (Rensselaer Polytechnic Institute), Richard T. Wong (Purdue University), and Thomas M. Liebling (Ecole Polytechnique Federate de Lausanne).
We also thank the many RPI students who used the preliminary versions, proofread the text, and tested out the exercises. Special thanks are due to Richard
Sych, Lori Grieb, Laura Ripans, Carla Bryan, and Robert Bosch.
Joseph G. Ecker
Michael Kup[erschmid
Ii
CONTENTS
XU
CONTENTS
Selected References
Exercises
3
3.1
~.2
29
30
FOR LINEAR
3.4
3.5
,
3.6
3.7
3.8
37
40
41
Canonical Form
42
42
44
45
46
47
Optimal Form
Unbounded Form
Two Infeasible Forms
47
48
49
51
51
52
54
55
58
Getting an Identity
The Subproblem Technique
Pivoting to Form a Subproblem
Summary of the Subproblem Technique
58
62
62
63
63
55
59
64
Maximization Problems
Inequality Constraints
Free Variables
64
64
65
69
Getting b ~ 0
The Artificial Problem
Feasibility of the Original Problem
Canonical Form for the Original Problem
3.9
37
Standard Form
The Simplex Tableau
Pivoting on a Simplex Tableau
Canonical Form
Finding a Better Basic Feasible Solution
The Simplex Rule for Pivoting
The Geometry of a Pivot
3.3
37
69
70
71
71
74
Pivot Matrices
The Revised Simplex Method
Tableaus with the Same Basic Sequence
74
Selected References
Exercises
76
78
79
79
80
81
82
83
83
CONTENTS
4
4.1
4.2
I
4
4.3
91
91
91
93
96
97
98
Convex Sets
99
99
100
101
101
102
103
105
105
Selected References
Exercises
5
5.1
5.2
5.3
5.4
5.5
5.6
5.7
6
6.1
6.2
6.3
I
\
Xlll
DUALITY IN LINEAR
The Standard Dual Pair
PROGRAMMING
109
109
Duality Relations
110
111
111
113
Economic Interpretation
115
of Dual Variables
118
119
120
121
122
124
127
US
128
130
Selected References
Exercises
132
133
139
141
142
144
145
145
146
148
150
139
141
XLV
6.4
6.5
CONTENTS
151
153
154
New Products
Technology Changes
New Constraints
154
155
156
157
157
Selected References
Exercises
I
J,
7.1
7.2
7.3
7.4
7.5
7.6
7.7
161
166
167
174
175
176
178
181
181
183
186
187
190
191
193
195
198
199
Selected References
Exercises
201
201
PART II
INTEGER, NONLINEAR AND DYNAMIC
PROGRAMMING
8
8.1
8.2
8.3
8.4
161
209
INTEGER PROGRAMMING
The Integer Programming Problem
Implicit Enumeration
Solution by Branch and Bound
211
211
214
217
217
218
225
227
227
A 0-1 Example
Looking Ahead
How Far to Look Ahead
Getting Nonnegative, Increasing Cost Coefficients
The Branch-and-Bound Algorithm for 0-1 Programs
228
233
237
238
239
CONTENTS
8.5
8.6
I
Ii
242
242
243
243
244
246
247
250
Selected References
Exercises
250
251
251
253
253
NONLINEAR PROGRAMMING
A Nonlinear Programming Problem
259
259
Contour Plots
Assumption of Continuity
Algorithms for Nonlinear Programming
261
263
263
Unconstrained Minimization
264
264
267
Equality Constraints
272
Parametric Representation
of Equality Constraints
Several Equality Constraints
Several Parameters
The Method of Lagrange
Classifying Lagrange Points
An Important Use of the Lagrange Multiplier Theorem
273
278
280
280
283
286
Inequality Constraints
287
288
290
291
Convexity
295
Convex Functions
Convexity and Minima
Checking Whether a Function is Convex
295
296
297
299
300
302
304
Line Searching
The Method of Steepest Descent
The Generalized Reduced-Gradient
The Ellipsoid Algorithm
Conclusion
306
307
311
315
322
Mixed-Integer Programs
Enforcing Logical Conditions
9
9.1
9.2
9.3
9.4
9.S
9.6
9.7
9.8
xv
Method
322
Quadratic Programming
Geometric Programming
Separable Programming
Other Theoretical Aspects of Nonlinear
322
326
329
332
Programming
XVl
CONTENTS
10
10.1
110.2
10.3
10.4
10.5
10.6
Selected References
Exercises
333
334
DYNAMIC PROGRAMMING
An Introductory Example
347
347
347
349
351
352
Formulation
A Loading Problem
The Boxes Problem
An Equipment Replacement Problem
Problems with Several State Variables
353
356
359
363
364
364
A Simple Example
A More Complicated
364
366
Example
Selected References
Exercises
PART III
PROBABILISTIC
11
11.1
11.2
369
369
MODELS
375
QUEUEING MODELS
A Simple Example and Basic Terminology
A Simple Model Suggested by Observation
377
377
379
11.3
11.4
11.5
11.6
11.7
11.8
11.9
11.10
12.1
382
387
390
391
394
397
399
402
405
407
409
411
412
412
12
Times
Distributions
413
414
Selected References
Exercises
415
415
INVENTORY MODELS
Economic Order-Quantity Models
421
421
Economic
Economic
Economic
Economic
Order-Quantity
Order-Quantity
Order-Quantity
Order-Quantity
Models
Models
Models
Models
with
with
with
with
No Shortages Allowed
Shortages Allowed
Price Breaks
Several Inventories
422
424
426
427
,
r
CONTENTS
12.2
A Single-Period Model
A Two-Period Model
429
432
437
441
441
449
Selected References
Exercises
452
452
SIMULATION
Next-Event Simulation
Selected References
Exercises
457
458
459
459
463
463
464
464
464
465
466
467
468
469
471
473
475
475
477
480
480
484
487
488
APPENDIX
491
12.3
IJ
13
13.1
13.2
w,
13.3
..
13.4
13.5
A.1
,
'1
NOTATIONAL
ALGEBRA
Matrices
CONVENTIONS
FOR MATRIX
t
A.2
A.3
A.4
A.5
A.6
XVll
491
492
492
493
494
494
494
495
495
497
505