Você está na página 1de 38

The Little Polya

A Small Compilation of George Polya’s Heuristic


Techniques

Compiled and Rewritten by A. L. Bruce

1
Contents

Preface…………………………………………………………………………...4

Introduction………….…………………………………………………………6

The Heuristic Method of Polya’s How to Solve It…………………………7

I. Understanding the Problem……………………………………………….10


1. Condition………………………………………………………………….11
2. Can You Derive Something Useful From the Data?.......................11
3. Definitions………………………………………………………………..12
4. Did You Use All the Data?...............................................................13
5. Is it Possible to Satisfy the Condition?............................................13
6. Separating the Condition and Setting up Equations……….……..14
7. Symmetry…………………………………………………………………16
8. Notation…………………………………………………………………..16

II. Devising a Plan……………………………………………………………..18


1. Analogy…………………………………………………………………...19
2. Auxiliary Problems and Auxiliary Elements……………………….19
3. Decomposing and Recombining……………………………………….21
4. Figures……………………………………………………………………22
5. Look at the Unknown…………………………………………………..23
6. Working Backwards…………………………………………………….24

2
III. Carrying Out the Plan…………………………………………………...26
1. Carrying Out…………………………………………………………….27
2. Generalization…………………………………………………………..27
3. Specialization……………………………………………………………28
4. Induction and Mathematical Induction………………….………….28
5. Test by Dimension……………………………………………………...29

IV. Looking Back………………………………………………………………31


1. Can You Check the Result?.............................................................32
2. Can You Derive the Result differently?..........................................32
3. Can You Use the Result?.................................................................33

V. Other Techniques…………………………………….…………………….35
1. Practical Problems……………………………………………………..36
2. Problems to Find and Problems to Prove…………………………..37
3. Progress and Achievement…………………………………………...37

3
Preface
In the process of composing this companion book to Polya’s How to Solve
It, I have become aware of many nuances, both in Polya’s heuristic itself, and in
the use of a book such as this. First and foremost I want to tell the reader that this
is by no means a self-contained and self-explanatory work. I have subtitled it “a
small compilation of George Polya’s heuristic techniques,” yes, but it is not a
complete compilation. In fact, I was highly selective in the material I chose to
include. Polya’s “short dictionary of heuristic,” from which all of these techniques
are taken, is some 72 articles long; certainly to merely rewrite the entire dictionary
would be pointless –why not just read Polya’s original work? Instead I have only
included the material that I think are rigorous and algorithmic techniques, or (in
most of the cases) heuristic knowledge that is absolutely essential. Much more on
this is said in the introduction, to which I now refer the reader who is curious about
my selection of the articles.
On that topic, as evidenced by my statement that this is not a self contained
work, is the fact that one cannot possible hope to find this of any use unless it is
used in conjunction with How to Solve It. This I cannot stress enough. Most of my
readers would probably not be drawn to this work if they had no knowledge of
Polya’s heuristic, but to those who do not, I recommend reading the whole of How
to Solve It as prerequisite to this. For those readers who have knowledge of Polya’s
heuristic but either do not have a copy of How to Solve It or haven’t read the whole
of it, consider the purchase of a copy and reading the whole of it a good decision.
In my opinion, anyone who considers themself learned in the mathematical art
should have at least a good, and desirably a masterful knowledge of Polya.
At this point I would like to discuss what the purpose of this book is. In
short, it is nothing other than a reference for those who are practitioners of Polya’s
ideas. I have found that there is a true need to be able to reference a given
technique at some point in one’s studies, or the even greater need for some easy to
find and useful advice on how one might go about solving a problem which proves
to be exceedingly difficult. How to Solve It itself is in my opinion not quite as
effective for this on-the-spot type of scenario.
The sections of his book are divided into many parts, all of which are
extremely valuable to the student first encountering Polya’s ideas, but which
become unwieldy in the face of a quick way to find and use them “in the heat of
battle” so to speak. The closest any section comes to being sufficiently helpful in
this respect is the afore mentioned “short dictionary of heuristic,” but even that is
burdened down with sections which are nothing more that general thoughts on

4
heuristic and the psychology of problem solving. Once again, these are brilliant
and truly invaluable to the student first studying heuristic, but they make the whole
much more difficult to navigate on-the-go. The relevant entries are many times
mixed with analogies, metaphors and involved examples, meant to aid in the
novice’s understanding, but which make the solid information about the technique
hard to find. Let there be no misunderstanding, I quite frequently have used
examples in this book, many of which are taken from Polya himself (as I will cite
most of the time), but the ones I have included are meant to show simply what it
would be extremely difficult to explain in the abstract. The main point of this book
was to boil the dictionary down, and in doing so to establish a reference for the
entire book, since the dictionary constituted the main part of his thought, as well as
195 of the 253 pages. I have also included a short piece in on the general heuristic
process developed in the first and second parts of the book to make sure all the
main aspects were covered.
I think that the work itself is pretty well suited to the job of on-the-spot
reference where How to Solve It may prove to too cumbersome, but it is in no way
a substitute. If one is interested in any particular technique one should refer to the
article’s counterpart in How to Solve It as every entry here does have one (the
names are usually the same in both). Lastly there may be some articles in How to
Solve It which may be of use in here which are not included. I understand that no
matter how impartial I try to be in their section, discretion is subjective. I plan to
put out further editions of this work which will likely contain additions to these.
This is by no means a work which is incapable of changing with time.

Adam L. Bruce,
Eureka, August 2009

5
Introduction

Most esteemed friends and colleagues:


If you are reading this little book, it’s probably because I have given it to
you in the hope that you may benefit by it greatly. It contains what I believe to be
the core of George Polya’s heuristic techniques as they are applied to various
mathematical and scientific problems, which are arranged in logical sections for
the reader. I hope it will serve as a good reference.
I have divided it into five main sections each representing a step in Polya’s
heuristic process. For example, you’ll find that the techniques under the
supersection of “Devising a Plan” have to do with creating a strategy to solving the
problem and the techniques under “Looking Back” have to do with learning from
the problem. The fifth section is devoted to the techniques that aren’t specific to
any one step in the process, but which are still valuable on the whole. Many of
those in the fifth section may not be algorithmic techniques, but general ideas
about various heuristic topics
All of these are my reworking of selected articles from the “Short Dictionary
of Heuristic” found in How to Solve It. Their selection I have based on the
following criteria:
1. Is it a self contained technique?
2. Is it applicable to an interdisciplinary set of problems?
3. Is it helpful as a specific tool in heuristic thinking?
Just as the author of a book on the Calculus would include a section on logarithmic
differentiation but may or may not include a biography of Gauss, I have chosen to
include only those essential techniques which have a specific heuristic function and
in so doing feel that I have made it a lot more useful.
In bringing all of these elements together I hope to have created a small
reference of heuristics for anyone who like myself can consider themself a
“disciple” of Polya –at least in his approach to problem solving. I think this will be
very handy for that purpose and will fill the need for a readily available place that
one can find various heuristic techniques. I will not lie that I have mostly done this
for myself, but I feel that others who are also familiar with the Polyan heuristic
process will undoubtedly find this useful.

6
The Heuristic Method of How to Solve It

I have decided to include a short piece on the basic heuristic process set
forth by Polya. It consists of four steps, each one designed to carry the practitioner
to the next, and ultimately through the whole of the problem.
1. Understanding the Problem:
It’s almost useless to attempt to solve a problem if you don’t fist understand
what it is asking you to accomplish, and therefore understanding the problem is the
first step of the process. One asks themselves such questions as “what is/are the
unknown(s)?”, “what are the data?”, and “what is/are the condition(s)?”. These
questions help to create a thorough understanding of the problem. Other techniques
such as drawing a figure to act as a visual representation can also aid in the
understanding of the problem; also, the introduction of suitable notation for the
next step is crucial, and will be discussed at length in 1.8. Once this has been done
we proceed to the next step.
2. Devising a Plan:
The second step is to devise a plan that will find the unknown. One knows they
have a plan when they either know completely, or know a rough outline, of the
steps they must take in order to obtain the solution. In a problem to prove, which is
Polya’s name for a problem involving a mathematical proof, this may not be an
exact unknown variable, but rather the main ideas of the proof.
The step of devising the plan is, in my opinion, the most overlooked by novice
students. I find that they tend to form a general understanding of what the problem
is asking them to accomplish, and from there do no more thinking on it but to
rigidly apply a given technique and proceed to the third step, which is carrying out
the plan (or in their case technique). This type of approach might work for some
elementary problems, but is utterly useless in any complex one. Instead when one
confronts a complex problem, the solution to which is eluding them, they should
do things such as looking for a problem “related to [their]’s and solved before” in
Polya’s words, or look at the problem from a number of different perspectives
taking note of the subtleties exposed by each. This will not solve all problems (for
indeed not all problems can be solved), but it will certainly bring much more
clarity and method to the way one goes about attempting to solve them.

7
3. Carrying Out the Plan:
In all respects the issue of carrying out the plan is the most important of all.
Why? Because one can have as thorough an understanding of a problem as
possible and one can also have devised the perfect plan to solve it as well, but if
one is incapable of putting that plan into action all else is meaningless. The
problem cannot be solved.
This being said, in mathematics, carrying out the plan is often the simplest step
of the process. Once one knows all the background information about the problem
to be solved (step 1), and the procedure, or at least an outline of the procedure to be
taken in solving it (step 2). One must implement the procedure. The only true
danger is that one might forget or become lost in trying to carry out their plan. This
is either because the plan is faulty or because one doesn’t completely understand
how to use it. A remedy is to look back on every step of the plan and consider its
validity. This can show either a flaw in the plan or a flaw in one’s understanding of
the plan at the same time as checking on minor errors that one might have made in
the process (computational errors, etc). During this phase, ask questions such as
“can you see clearly that each step is correct?”, “can you prove that each step is
correct?”
4. Looking Back:
Looking back affords one the opportunity of learning from the problem and of
consolidating their understanding of the solution. No problem is ever exhausted if
one is willing to look back and consider the process they took in solving it. Even
though one has looked at their plan step by step, and hopefully examined their
mundane computational work for errors, this gives the opportunity to reexamine
the work. Questions such as “can you check the result?”, “can you check the
argument?” are employed, expanding on the more particular questions in step 3 to
include the entire solution. These are not so much as to insure the plan itself is
correct, but more to make sure that the solution one has found is sufficient to
satisfy the problem. Thus, if one can check the result, they check both the plan and
the solution.
Other questions such as “can you derive the result differently” serve to help one
consolidate their knowledge of the topic. In almost any sufficiently complex
problem there are multiple ways of obtaining identical (and hopefully correct)
solutions. In asking oneself this question and more importantly in finding different
ways of obtaining the same result, one achieves a thorough understanding not only
of the single problem, but of the subtleties of the topic in general. This leads to an
enhanced understanding of another problem which belongs to that topic, and
8
ultimately supplies one with a problem “related to yours and solved before” when
one does encounter another similar problem.
The essence of Polya’s method is understanding; understanding of the problem,
understanding of methods employed in solving the problem, understanding of how
to apply those general methods to the problem at hand, and finally understanding
why such a method was employed in the first place. All of these general steps have
specific heuristic techniques which are associated with them and which aid in their
use. It is the most useful of these which I now present to you in the forthcoming
part of this little book.

9
Understanding the Problem

10
1.1 Condition:
“The principle part of a problem to find” (my italics). In sum, the information
given by the problem constitutes the condition insofar as that information modifies
the process taken in solving it (if not it is data). By this logic it should seem that
any problem is merely a set of conditions, all of which are satisfied through a given
method, but there is a distinction between the main unknown of the problem and its
subsequent conditions. Thus the problem “find two numbers whose sum is 60 and
whose quotient is 4” is made up of an unknown and two conditions.
1. Find two number numbers (unknown)
2. Their sum is 60 (condition1)
3. Their quotient is 4 (condition 2)
This is solved easily with a simple translation into algebraic notion (see 1.7),
60=x+y, x/y=4; so x=48, y=12.
A condition is redundant when there is more information than required. So to
add the third condition of “their product is 80” to the equation above would create
a linear system comprised of more equations than variables; therefore it would be
redundant and may not have a solution. A condition is insufficient where there is
less information than required. To delete one of the conditions would render the
problem above too vague, and therefore incapable of a solution. A condition is
contradictory when two or more elements are mutually opposed. To add a third
condition of “their sum is 43” to the problem above would be to render it
contradictory, because two numbers cannot have a sum of 60 and 43
simultaneously.

1.2 Can You Derive Something Useful from the Data?


In any problem there are two places to start. The first and most prevalent is
from the unknown, but another place is from the data. Polya uses the analogy that
these two are a separated by a gap for which a bridge must be built –where one can
start from either side. The main question of this technique is “Can you derive
something useful from the data?”
A. Beginning the Inquiry
The inquiry begins with the usual set of questions:
1. What are the data?
2. What is the unknown?
11
3. What is/are the condition(s)?
The usual procedure would be to start from the unknown, i. e. set up something
like an x = y, where x is the unknown and y is the data and condition. At that point
one usually can obtain a solution, if it is still elusive, than one should consider that
data more.
B. Considering the Data
In considering the data one thinks of the problem in terms of y=x or at least y
implies x. It is a deductive process, where x must be equivalent to one’s solution
because of the data, that is, x supports the data and not vice-versa. Use the data in
isolation when possible, so that the relationship between various parts of it may be
observed. The example of the line through three points is how Polya shows
working from the data in How to solve it, see the section in the heuristic dictionary
which bears the same title as this section for the problem itself.

1.3 Definitions:
“The definition of a term is a statement of its meaning in other terms which
are supposed to be well known”. Thus says Polya.
A. General Information
There are two types of definitions, or technical terms.
Primitive terms, or terms which are not defined, are the first. These are those such
as “point” and “line” in Euclidian geometry, and “number” and “variable” in many
algebraic systems. These terms cannot be defined because there is nothing to
define them with.
Derived terms are terms which can be defined using primitive terms. Therefore, a
circle can be defined as “the locus of points equidistant from a single point in
Cartesian two-space”.
B. Going Back to Definitions
The main application of primitive terms vs. derived terms is in restating a
problem using primitive terms for derived terms, which Polya calls going back to
definitions. The point of this is to make the problem easier to understand.
One starts with a problem which uses many unfamiliar derived terms, and slowly
works it so that the derived terms are restated in primitive terms, this may not be
12
accomplished at once, but over the interval where other heuristic techniques are
coming into play as well. Drawing a figure, for example, may help one to restate
the derived terms of the problem in the primitive terms of the figure. The heuristic
question is “Can you restate the problem?” The restatement is the deflated
problem, where one has “deflated” the unfamiliar derived terms.

1.4 Did You Use All the Data?


Essential to the understanding of a problem as a whole is the understanding
of its data. When one is finding themself unable to solve a problem, it may be
because one has not taken all the data into account. One must ask themselves “did
you use all the data?”, or “did you use all of the conditions/the whole hypothesis?”,
with the divide in the latter for a problem to solve and problem to prove
respectively. It is a check and nothing more. Any strategy for solving the problem
that does not take into account all of the data is likely flawed, excepting those
circumstances where some part of the data is extraneous.
One might have a flawed or incomplete notion of the problem if one does
not take all the data into account, or does not consider in the right way some part of
it. Therefore one must also ask “have you taken into account all of the essential
notions involved in the problem?” This cannot be directly accomplished if one
does not know what the essential notions of the problem are, of course, but rather
can be accomplished through considering the different ways all of the data
function, both individually and as a whole.

1.5 Is it Possible to Satisfy the Condition?


Since it is pointless to work toward an end which one cannot achieve, one
must appraise the condition of the problem as to whether or not it can be satisfied.
Thus one asks themself “Is the condition sufficient to determine the unknown, or is
it insufficient, redundant, or contradictory?” In any reasonable problem, the
condition can be satisfied. This is most useful at the beginning of considering the
problem, and because of that should only be a plausible guess which gives a
provisional answer.

13
1.6 Separating the Condition and Setting up Equations:
“To set up equations means to express in mathematical symbols a condition
that is stated in words; it is translation from ordinary language into the language of
mathematical formulas.” He later compares it to a translation between two
languages. The problem can either be a simple translation or a complex one. The
process itself is ultimately an algorithmic one, which precious few of Polya’s
techniques are. Primarily this technique is used for word problems.
A. Separating the Condition
In easy cases, the word problem will split into successive parts which can
then be translated into an equation. In difficult cases, there is often some nuance
that cannot be directly translated into mathematical symbols.
First one may have to rearrange or separate the condition; to do this one
must completely understand the condition, which may require one to deflate the
problem (see 1.3). One then separates the various parts of the condition, the
question being “Can you write [the independent parts of the condition] down?”
B. Setting Up Equations
At this point one makes the translation into the mathematical notation, if it is still
not clear as to the form the translation will take, it is likely that one has not made
the appropriate divisions. One can use a vertical line down the page to separate the
statement in words from its equivalent mathematical representation such as:
|
(Statement in words) | (Statement in Mathematics)
|
|

At this point one can proceed to solve the problem.


C. Example
This is an example taken from Polya (see pages 175-176 of How to Solve It).
Find the breadth and height of a right prism with a square base, given it has a
volume of 63 in3, and a surface area of 102in2.

14
What are the unknowns? The breadth and the height.
What are the Data? Same as the condition.
What is the Condition? The volume of the prism is 63 cubic inches and the surface
area is 102 square inches.
First: Draw a figure (see 2.4).
Second: Reduce the terms to primitives; one can deflate the unfamiliar terms
breadth and height (see 1.3).
The breadth of a right prism is defined as a side of the base, one can call it x. The
height is defined as the altitude, say y. Therefore we restate the problem:
Given a parallelogram with a square base that has a volume of 63 in3, and a
surface area of 102in2. Find the length of the side of the base and the altitude of
the prism
Third: The problem now needs better organization, particularly of the condition.
Therefore, separate the various parts of the condition.
There are two parts: first the volume, then the surface area, both of which one
needs knowledge of Euclidian geometry to fully understand, but no more. Thus
one has successfully separated the condition so to facilitate a greater understanding
of the problem at hand.
Fourth: Make the transition from words to mathematical symbols:
Find the length of the side of the base | x
Find the altitude | y
The volume is 63 in3 | x2y = 63
The surface area is 102 in2 | 2x2 + 4xy = 102
From this one can easily solve the linear system and obtain a solution to the
problem. Notice that the equations are in terms of the main variables rather than
their Euclidian definitions; it would be helpful in any circumstance, and even
necessary in others to define say x2 as equaling the base of the parallelogram etc.
These can also be included in the “translation” with the dividing line in the middle.

15
1.7 Symmetry:
Symmetry, in a general sense, is the idea that certain parts of a problem are
interchangeable with others, thus given the sum xy +xz +yz one can interchange
any two variables without changing the meaning of the expression. Any symmetry
one encounters should be noted, and one should be careful not to destroy any
natural symmetry without cause, since it can help one to understand the problem.

1.8 Notation:
Notation is one of the most important aspects of problem solving. Essential
to the process of understanding the problem is to “introduce suitable notation”,
using this one can better formulate a way of dealing with the problem. Most of all a
notation cannot be redundant or ambiguous –one can easily hamper themselves in
their ability to work the problem if they are using a poor notation, and it makes it
near impossible to go back and check one’s work. This is all summed up by Polya
when he says:
“A good notation should be unambiguous, pregnant, easy to remember; it
should avoid harmful second meanings and take advantage of useful second
meanings; the order and connection of signs should suggest the order and
connection of things.”
A. General Principles
The general principles of devising a notation follow Polya’s quote.
First, none of the elements of the notation can be ambiguous:
One symbol cannot denote two or more objects.

written ܽ × ܾ, ܽ ∙ ܾ, or simply ܾܽ. In some cases this is advantageous, but it


Two or more symbols can denote one object. Multiplication, for instance, can be

should never be done without cause.


Second, the notation should be easy to remember. It should remind one of
the quantity it denotes when that is appropriate.
A simple devise is to use the initial of the quantity, V for volume, a for
acceleration, etc. This doesn’t work when one works with two quantities which
have the same initial, rate and radius for example, but for this there are other
methods.

16
Third, the order and connection of the signs should suggest the order and
connection of the objects. These are shown many times through the alphabet used.
Letters near the beginning of the alphabet, a, b, c, usually denote constants or other
given quantities, while letters near the end, x, y, z, usually denote variables. Thus if
in a problem one is given length, width, and height, it may be more useful to write
them as a, b, c, rather than l, w, h, to show that they are given constants and not
variables.

thus in Euclidian geometry: A, B, C, are all points; a, b, c, are all lines; ߙ, ߚ, ߛ, are
Objects belonging to the same class are usually written within the same alphabet,

all angles.
These are the foremost and most necessary principles of formulating a
notation. There are others, but they mostly consist of nuances and so will not be
included here (see How to Solve It 138-141).

17
Devising a Plan

18
2.1 Analogy:
Analogy is a very important aspect of problem solving; it allows one to make
inferences about the problem at hand through problems which are related to it.
Much of the time, a good analogy makes use of an auxiliary problem (2.2) and
vice-versa. One, when desiring to make use of a problem analogous to the present
one, should ask themself if there is a “simpler problem related to the present one”.
To state the obvious, simpler problems are easier to solve, and therefore one
can possibly derive some useful insight or information from them, without
burdening oneself with yet another difficult problem and “losing sight of the goal”.
Therefore, given the choice between a simple problem which is somewhat
analogous to the present one, and a another problem, just as difficult as the first,
which is much more related, one should choose the simpler one because one
cannot solve the difficult one; If the solution to the difficult one was obtainable
than why was the solution to the present problem evasive? There are instances
where this is not true (2.2 (D)) but for the most part this is a good general rule.
Polya devotes pages to the subject, including numerous examples and anecdotes,
but this is the heart of his argument (see How to Solve It 137-146).

2.2 Auxiliary Problems and Auxiliary Elements:


Both auxiliary problems and auxiliary elements are used as an analogy (2.1)
to the present problem, usually in the format of a problem related to the present
one and solved before, but in some cases merely as a simpler problem which one
can solve easier.
1. Auxiliary Elements
An auxiliary element is defined by Polya as “an element we introduce in the
hope that it will further the solution”. There are various types of auxiliary
elements, in Euclidian geometry elements such as auxiliary lines and auxiliary
polygons, in algebra auxiliary unknowns and auxiliary theorems, etc.
There are also various reasons for introducing auxiliary elements, such as
using them to make the present problem similar to another problem, otherwise
related and solved before. Thus, given a related problem whose solution has to do
with triangles, yet having no triangles in one’s figure, one can introduce a triangle
as an auxiliary element to take full advantage of the related problem. And when
going back to definitions (1.3), if one finds that the primitive term contains some

19
auxiliary element, one should not hesitate to introduce it to the problem. As well as
many more reasons.
2. Auxiliary Problems
One takes up an auxiliary problem to illuminate the solution to the present
problem.
A. Profiting from the Problem
There are two ways to profit from the problem.
When one profits from the result they use a solution obtained in the auxiliary to
make clear the solution to the present problem. Such as, given x4+x2+32=0, if one
lets y=x2, the solution becomes clear, and once one solves y2+y+32=0 one also
obtains x by ඥ‫ݕ‬.
When one profits from the method however, one takes a notion involved in the
auxiliary problem and applied it to the present problem. Thus one observes the
condition (1.1). Polya’s example of this is the problem: find length of the diagonal
of a rectangular parallelepiped being given the lengths of the three edges drawn
from the same corner. The appropriate auxiliary problem is that of finding the
diagonal of a parallelogram, which introduces the same notion that the
Pythagorean Theorem is to be used.
B. Equivalence and Bilateral Reduction
Many auxiliary problems are equivalent, defined by Polya by saying “the
solution of each requires the solution of the other”. In essence, two equivalent
problems involve the same or extremely similar mathematical notions in their
solutions. Thus the two problems
1. x2+x=0
2. 3x2+3x=0
are equivalent. They are not identical, but the same notions are involved with
each. Thus if one can realize the answer to (1) is zero, one knows the answer to (2)
must be zero, and furthermore the answer to any problem of the form ax2+ax=0
must be zero. The process of proceeding from (1) to (2) is termed bilateral
reduction.

20
C. Chains of Equivalent Problems
Using the idea of bilateral reduction, it is possible to set up a large chain of
equivalent problems which stretches to a problem which is either already solved, or
one to which to solution is easily obtained. Since each problem is equivalent to the
one before it, each and every problem is equivalent to the first. Thus, all algebraic
manipulation can even be termed bilateral reduction.
D. Unilateral Reduction
Given two problems, both unsolved, where the solution of the first would
solidify the solution of the second, but not vice-versa, thus we should solve the
first problem first then the second. If the first problem is “more ambitious” than the
second, there are however two ways to proceed. When one proceeds from a given
problem to a “more ambitious” one or a “less ambitious” one, the process is termed
unilateral reduction.
Of the two ways, the first is to deal with the “less ambitious” then the “more
ambitious.” Even though in the scenario above this would be less advantageous,
there are many circumstances where this is helpful (the second example in (A) for
instance). This is by far the easiest and most common.
The second is to proceed from the “less ambitious” to the “more ambitious.”
This was almost forbid in the analogy section (2.1), but in some special instances
this can be helpful. The ability to solve a difficult problem before a simple one is
called by Polya the inventor’s paradox (see How to Solve it 121-122).

2.3 Decomposing and Recombining:


Decomposing is the process by which one examines the problem by looking
at each of its elements individually. After this is accomplished, reconstructing
them, called recombining, can give one a better idea of the problem. One primarily
looks at the three principle parts of the problem, namely what is the unknown?
what are the data? And what is the condition? Once each of these has been
examined individually, they can be further decomposed by going back to the
definitions (1.3), etc. The main point of decomposing a problem however, is to
construct an auxiliary problem which will be able to aid in obtaining a solution to
the first. There are three primary ways of doing this: 1. Keeping the unknown and
changing the data and condition, 2. Keeping the data and changing the unknown
and the condition, and 3. Changing both the unknown and the data.

21
A. Keeping the Unknown and Changing the Rest:
The problem has the same unknown as the original, but other things are
slightly changed. One considers “what data is appropriate to determine the
unknown.” One could also keep part of the data and part of the condition, changing
as little as possible, but dropping some part of either and consider the problem
then.
B. Keeping the Data and Changing the Rest:
The data is retained and a new condition and unknown are created. The
unknown should be useful and accessible, acting as a median point between the
data and the original unknown. Since it is hard however, to conceive of an
unknown which is both useful and accessible, one can introduce a new unknown
which is related to the original, but that may not be as accessible, in the hope that it
will yield to a solution more easily than the original and vice-versa.
C. Changing Both the Data and Unknown:
This type is a more radical change than those which preceded it. The new
problem however, might have a good chance of success, and thus one considers if
they can “…change the data or unknown, or both if necessary, so that the new data
and unknown are closer to each other.” This is done through considering the
principles for changing the unknown and data found in (A) and (B).

2.4 Figures:
The drawing of figures is essential to one’s ability to solve the problem.
Many, if not all, geometric problems have a figure which is associated with the
problem, but in other problems, it is very useful to introduce a figure which then
acts as a visual aid in one’s heuristic process.
A. Drawing Figures Exactly
Exact figures are not absolutely necessary, but one should draw them as
exactly as possible. A good freehand sketch of a figure should be enough for most
problems. A badly drawn figure will suggest a false conclusion, and therefore hurt
more than it will help. It is important to consider the problem; does the figure fit
the problem?

22
B. Order of Construction
The various elements of a figure must be constructed in the correct relations
and measurements, or at least close to the correct ones. The order of their
construction however, is up to the problem itself. Therefore, given two angles a
and b to be constructed so that a=3b, it is not possible to construct first a and then
b, one must rather construct b first and obtain a from the first. The figure in either
will illustrate the same concept. In many other problems, the order of construction
will not have an effect on the figure, in which case it is optional.
C. Erroneous Conclusions
The construction of a figure should not introduce any symmetry or relation
that is not shown in the problem. Lines and angles which are not equal should not
appear so; otherwise this may lead to false conclusions.
Note: the best triangle to construct for a general triangle is one with the angles 45,
60, 75. This is the most remote from both an isosceles and a right triangle
D. Shading
Shading is a very important part of drawing a figure. One can shade a line or
area which has a special significance to the problem. Creating darker lines bring
out a certain area, while dotted lines hint that there is some relation between two
other things in the figure.

2.5 Look At the Unknown:


It is advantageous in any problem to consider the final goal of the problem.
Without this in mind one can easily be sidetracked, thus the Latin saying “respice
finem,” or Polya’s more understandable “look at the unknown” are used to
describe the consideration of one’s goal.
A. The Unknown as an Auxiliary Problem
It is very helpful to consider an auxiliary problem (2.2) which has the same
unknown as the present one. Thus if the unknown to the present problem is the
length of a line, the unknown to the auxiliary problem should also be the length of
a line. By doing this one might very well be able to profit from both the method
and the solution to the auxiliary problem.

23
First one looks at a schematic of the problem, where all other parts of it
except the unknown are omitted. The example Polya gives is:
“Given……….Find the length of the line”
This focuses our attention to the nature of the unknown, which in this case is a line
Second one can consider all of the other problems which have unknowns
related to theirs; there is “an economy of choice”, as Polya puts it, where one
considers the simplest and most familiar first. In the example one sees that the
length of the line could easily be obtained if it were a side of a triangle. Thus we
must introduce the auxiliary element (2.2) of a triangle into our figure, etc…
B. General Unknowns in Auxiliary Problems
For any problem the process in (A) can be of use. One must consider the
typical types of problems which involve a certain unknown, thus for the unknowns:
1. “Given……….Find the angle”
2. “Given……….Find the area of the cube”
3. “Given……….Construct the point”
We can think of the problems for
1. To be concerned with some triangle
2. To be concerned with some side or given distance
3. To be concerned with some locus of points
Thus the heuristic question is “given a problem, can you think of an auxiliary
problem having the same or a similar unknown?”

2.6 Working Backwards:


When one is confronted with a problem which is extremely perplexing, one
should consider it from a variety of perspectives. An important one of these is
called “working backwards” by Polya, where one considers the solution and from
the solution derives the problem. This may seem strange, but is a good and useful
skill nonetheless, thus the saying of Pappus, “let us assume what is sought as
already found.” Many times the actual work is through a series of auxiliary
problems (2.2) each derived from first the solution and then each other. This is
illustrated best in the following example.
A. Example
24
Polya’s example for this technique is so poignant and well conceived that I
would be a fool to not include it at present. Thus it is this example I set forth.
How can you bring up from a river exactly six quarts of water having only
two containers to hold it, a four quart pail and a nine quart pail?
What are the unknowns? Six quarts of water, or more correctly the method
for obtaining six quarts of water.
What are the Data? Same as the conditions
What are the conditions? You only have a four quart pail and a nine quart
pail.
First: Draw a figure(s). I shall not include any in my explanation as I
currently have no practical way to do this at my disposal. Refer to Polya (How to
solve It pg 226-229) for the figures or draw your own (it’s not that difficult).
Second: The problem is much more difficult than it appears at first, and
there seems no way to obtain the solution from the problem, therefore we work
backwards.
Working backwards one has 6 qts in the 9 qt container (the 4 qt container
cannot hold enough water).
How could one have come by this?
If one had 1qt in the 4 qt container and had filled the 9 qt, then one could
have poured 3 qts on top of the 1 qt in the 4 qt container (until it was full) and then
poured all of the contents of the 4 qt into the river, leaving them with 6 qts.
How could one get the 1 qt in the 4 qt container?
If the 4 qt was empty and one had 1 qt in the 9 qt container one could have
transferred the 1 qt from the 9 qt container into the 4 qt container.
How could one have 1 qt in the 9 qt container?
Then the solution is realized. If one had filled the 9 qt container, and then
poured its contents into the 4 qt container, emptied the 4 qt container into the river,
and repeated that process one could obtain the one qt. The solution consists mostly
of simple arithmetic, 9 – 4(2) = 1. Thus one has obtained the correct solution,
which might never have occurred to them if they hadn’t thought of “working
backwards.”

25
Carrying Out the Plan

26
3.1 Carrying Out:
The step of carrying out the plan is different from merely devising it. One
can notice nuances to the problem and sees things about it that one was not aware
of when the plan was first devised. Devising the plan, one makes use of many
plausible guesses and intuition. Once the plan is to be carried out however, these
must be replaced by a more rigorous set of standards.
One should pay special attention to the order which one carries out the steps
in their plan. The major aspects to the argument should be checked before one
starts to go into its details. There should be no detail which is omitted, and the
relationship between various details should be noted.
One must also verify their argument at every step of the plan if one is going
to be sure if its validity. There are two ways to do this. The first is by a direct
rigorous proof that the step is correct, and the second is by an intuitive notion that
one “sees” how the step is correct. Both of these must be verified when carrying
out the plan, since the plan cannot possible true if one cannot prove a step, and
cannot possible be practical if one cannot see in intuitively. Thus one asks themself
the questions “can you prove that [the step] is correct” and “can you see clearly
that [the step] is correct.”

3.2 Generalization:
Generalization is, in sum, the technique of realizing that the problem at hand
is a member of a greater set of problems, all of which have similar solutions, and
then using the general solution for that set to solve the problem at hand. Polya
defines it as “passing form the consideration of a certain object to the consideration
of a set that contains that object.” When one is confronted with a problem which is
burdened with data, one can generalize the data so as to give a clear indication as
to the solution. For given lengths one can create letters, for specific objects, one
can substitute a general class of object. Thus the problem is likely to resemble
some abstract idea which one is familiar with. Then the solution becomes apparent
and the problem can be solved.

27
3.3 Specialization:
Just as generalization (3.2) moves from the consideration of a specific object
to a set of objects, specialization moves from the consideration of a set of objects
to a specific object or from one member of the set to another in the set which is
closely related to it. Polya outlines a specific method by which one should create
an auxiliary problem (2.2) through a specialization.
A. Specialization Though Auxiliary Problems
Given an original problem which is too difficult because of a given
condition or set of conditions, one can specialize the problem. Invent an auxiliary
problem which ignores one or more of the conditions so that it can be solved
easily. What is to be learned from this problem? How does it relate to the original
problem? Once this connection has been made, one can many times go on to solve
the original problem without much trouble. Since one part of the problem is solved,
solving the others, i. e. the conditions which were previously ignored, is more
manageable. Thus, the specialized problem serves as a stepping stone to the
original.

3.4 Induction and Mathematical Induction:


There are two types of induction in mathematics, the first is the style of
formal proof using the n + 1 model, and the other is an informal way of creating
and supporting plausible guesses. In How to Solve It Polya discusses both types,
especially the formal one in great detail, but since the formal type is common
knowledge, and not truly of any heuristic value, I will not include it here, only the
informal.
A. The Informal Type of Induction
In solving a problem one may chance upon some discovery without being
sure of its validity. A common way to form a plausible guess as to its validity is to
test it in other scenarios, if a formula is good for one configuration of numbers, is it
good for another? How many others? If there are a substantial amount of other
cases where the formula holds true, then it is most likely true (one proves nothing
however, until one has devised a rigorous argument, and it is only a plausible guess
until that occurs). It is a step toward a generalization (3.2), and supplying other
cases of a given idea’s validity is a specialization (3.3). The best use of this is as
one is carrying out a plan, if one makes some discovery, to verify the validity of it

28
for any plausible purpose, so that one may use it in the future, or even in the
problem at hand.

3.5 Test by Dimension:


The test by dimension is one of the most useful ways to check a physical or
geometrical formula. Any physical concept has its own dimension associated with
it, thus for area it is cm2, m2 etc… If one is given a formula, but there is some
reason to question its validity, one can test it by testing the dimension, that is, by
finding an equivalence between the units on each side. There are two main uses of
the test by dimension, the first is to check the accuracy of a formula which is
already given, and the second is to help find a formula which one doesn’t
completely know. Both of these are illustrated by examples.

A. Checking the Validity of a Given Formula

Suppose one is given the question “find the volume of a sphere being given
its radius as 4cm.” This is an extraordinarily simple problem of plane geometry;
there is a formula, but also suppose one cannot quite recall the formula. Is it
ସ ସ
ߨ‫ ݎ‬ଶ or ߨ‫ ݎ‬ଷ ? If one is trying to find volume, then one finds a measurement in
ଷ ଷ
cm3. Since the measure of r is in cm, then it follows that r2 will yield a
measurement of cm2, but that r3 will yield a measurement of cm3. Thus one can be
sure the second one is correct, we have established an equivalence between the
dimensions.

B. Finding a Formula Which One Doesn’t Completely Know

Suppose one has a simple pendulum. One must find a formula which
expresses the period, T, of the pendulum in terms of l, the length of the string and
the gravitational force exerted upon it, g. From a physics class, one can remember
that the period involves the product of a constant and both of the other variables
raised to some power, but not the exact power, that is:

ܶ = ݈ܿ ௠ ݃௡

One knows however, the dimensions of each measurement, thus T= sec,


l=cm, and g=cm(sec-2). Thus one can produce the equation,

29
sec = cmmcm(sec-2)n

Distributing we get:

sec = cmm+nsec-2n

To eliminate the cm terms one sets them to 0, such as:

m+n=0

And then to complete the system, one writes the remaining equivalence:

1 = -2n

Thus we obtain n = -½ and m = ½. The formula is now:


݈
ܶ = ܿඨ
݃

One has now ascertained much more about this formula than was previously
known, without actually memorizing it. True, there is no way to find the constant
c, but it is still better that the previous form of it.

30
Looking Back

31
4.1 Can You Check the Result?
Checking the result is a very important aspect of any problem. If one cannot
check the validity of their result, the result itself is most likely invalid. There are a
number of facets to this technique.
A. Checking An Unknown
In problems to solve (5.2), where the unknown is a number, one should
consider if the answer is reasonable. Therefore finding the age of a man who you
know is a grandfather to be 9.3 years is incorrect, without considering the problem
or solution itself we know it is incorrect, because the answer is unreasonable. This
does not always work, but it should give one reason to consider their work.
One can also check the validity of a “problem in letters,” a formula or
algebraic manipulation. This can be done by specialization (3.3) of the formula. Is
there a specialized case where the formula reduces down to another; where another
formula becomes the first? Can one let one variable equal another so as to reduce
the expression to a known equation?
B. Checking the Argument
When one checks the argument it should not be mere repetition of the checks
while carrying out (3.1). Instead one should focus on specific parts of their
argument, what are the weaker portions? Why? How can one make them stronger?
Thus, in answering these questions, one does not stumble over any part of the
problem, and one’s argument is consistent and provable.

4.2 Can You Derive the Result Differently?


In most problems, there are multiple methods for obtaining the correct
answer to the problem. In any problem where the solution is long and involved,
one naturally looks for a solution which is simpler, easier to follow and clearer.
Thus the questions “can you derive the result differently?” and “can you see it at a
glance?” are asked by the good mathematician. One may not be satisfied with
merely one method by which the solution is obtained, and will look for another to
obtain that same solution so to build the credibility of the solution itself, thus it is
said by Polya “two proofs are better than one.”
In considering a different derivation of the result one may see the result in a
new light, certain relationships and attributes of it may become obvious, whereas
before they were not known. In examining the various parts of the problem, and
32
reassembling them one is likely to get a better understanding of the problem as a
whole, thus allowing one to see clearly the solution and be sure of its validity.

4.3 Can You Use the Result?


“To find the solution to a problem by your own means is a discovery.” So
the intelligent mathematician considers what use to make of this discovery; one
asks is what other problems can the solution to the present one aid? This many
times calls for one to invent new problems, analogous to the first one. There are
two main processes for achieving this: Retaining the unknown and changing the
conditions, and changing the unknown and retaining the conditions and
equivalence.

A. Retaining the Unknown and Changing/Adding Conditions

The process itself is exactly what the name implies. Given a solved problem,
one retains the unknown and changes, or adds some conditions. Polya’s example of
this is: given the length, width, and height of a rectangular parallelepiped, find the
diagonal. In this case the diagonal of the parallelepiped is the unknown. Once this
is solved, one then can be sure that the following questions are all partially if not
completely solved:

1. Given the length, width, and height of a rectangular parallelepiped, find the
radius of the circumscribed sphere.
2. Given the altitude of a pyramid and the sides of its base, find the lateral
edges.
3. Given the rectangular coordinates (x1, y1, z1) and(x2, y2, z2) of two points in
space, find the distance between the two points

Thus one has used the solution from the first problem to find the solutions for
many other problems.

B. Changing the Unknown and Retaining the Conditions/Equivalence

In the fore mentioned problem, once one has obtained a value for the
unknown one has also established an equivalence among the four parts, thus given
any three of the parts, one can always find the fourth. Thus from the solution to the
first problem one also obtains the solutions to the problems given the length, width
33
and diagonal of a rectangular parallelepiped, find the height etc… for all of the
other quantities. These new problems retain all of the conditions in the first one,
and most of all the equivalence established between the various parts of the first
problem, but changes the unknown, substituting one part for another.

34
Other Techniques

35
5.1 Practical Problems:
Outside of mathematics there are many other sciences which make use of
mathematical problem solving with relation a physical system. Many times the
information in a practical problem can be staggering, with hundreds of unknowns,
conditions, and thousands of data. This is because a practical problem is usually
not as clear as purely mathematical problems.
The complexity of the unknowns, the data, and the conditions differentiate a
practical problem from a purely mathematical one the most sharply. It is usually
thought that practical problems need much more experience to tackle than purely
mathematical problems, but this is usually in the knowledge needed that the actual
heuristic approach to the problem. In either type, one applies their knowledge of
related problems, thus the questions “have you seen the same problem in a slightly
different form?” and “do you know a related problem?” should be asked in any
case.
Another sharp difference is in the ability to completely understand all of the
notions related to the problem. In a mathematical problem, these are clear, and one
can fairly easily gain an understanding of them. In practical problems however,
these are usually very “hazy” as Polya puts it, and the clarification of them is
essential, thus the questions “have you take into account all essential notions
involved in the problem?”
In pure mathematical problems one must include all data and all conditions,
but not so in a practical problem. Think of the engineer in charge of building a
power plant. They must take into account things such as cost, environmental
impact, and efficiency, but not the petty grievances of the local residents. Thus one
asks themself “did you include all the data/conditions which could influence
appreciably the solution?” rather than just the first part of the question.
The last consideration is “can you make the problem simpler by using a
reasonable approximation rather than completely accurate data.” A burdensome
amount of detail in calculations may make the problem more difficult and involved
than it should be.

5.2 Problems to Find Problems to Prove:


There are two principle types of mathematical problems. The first, problems
which revolve around some unknown bit of information, are called “problems to
find” by Polya, and the second, problems which revolve around proving or
36
disproving a given proposition, are called “problems to prove.” This entry shall
take up both separately.
A. Problems to Find
The unknown of a problem to find could be anything. In geometry, a shape
or measurement, in algebra a number, in the calculus an antiderivitive to a given
function, in a crossword puzzle a word of a given amount of characters, etc. One
must know and be able to decompose (2.3) the three main parts of a problem to
find, the unknown, the data, and the condition. In simple problems there may be
only one or two of each, but in more complex problems one might have many sets
of data and conditions. Either way, in solving a problem to find one asks
themselves “what is/are the unknown(s)/data/condition(s)?”, “can you separate the
various parts of the condition?” (1.6), “can you establish a link between the data
and the unknown?” etc (there is a complete list in How to Solve it pg 155-156).
B. Problems to Prove
Unlike the unknown of a problem to find, the unknown of a problem to
prove is “to show conclusively that a certain clearly stated assertion is true, or else
to show that it is false.” Thus problems to prove depend much more on the content
of the problem for their variation rather than the nature of what is sought. Usual
problems to prove can be separated into two parts, the hypothesis and the
conclusion. This is not true for all problems to prove however, note the problem
“prove there is infinitely many prime numbers.” In solving a problem to prove,
understanding these parts is very important, and furthermore, recognizing the
nature of the notions involved is essential. Thus one asks questions like “did you
use the whole hypothesis?” and “is there another problem with a similar
conclusion?” etc (once again, there is a complete list of these questions in How to
Solve it pg 156).

5.3 Progress and Achievement:


There are certain steps one takes as they progress toward the solution of a
problem. There are three very important ones called mobilization, organization,
and changes in the mode of conception. Each of these has a given heuristic value,
and shall be discussed separately.

37
A. Mobilization
When first confronted with a problem one must assemble the necessary
knowledge in order to correctly form a plan to solve it. This requires either
recalling knowledge that one already has or gaining knowledge which one does not
have, but is conducive to the solution. This process of gathering knowledge is
termed mobilization by Polya. It is a process of extraction
B. Organization
Once one has mobilized their knowledge of the problem, the knowledge
must be ordered correctly so as to find a solution. One must take the isolated facts
which are the products of the mobilization and combine them in a suitable way so
as to create a path to the solution. This process is called organization by Polya, and
is used to “…construct and argument containing the materials recollected to t\a
well adapted whole.”
C. Changes in the Mode of Conception
Once one has mobilized and organized their knowledge, one sets off
carrying out the steps to find a solution. Throughout this process one becomes
aware of things which were not known before, realizes connections which one was
previously unaware of, and recalls helpful ideas which previously didn’t occur to
them. Thus, in the process of solving a problem, one’s idea of the problem and the
contents of thee problem changes from the outset. These changes are called
changes in the mode of conception by Polya. With several changes in the mode of
conception, one becomes aware of various standpoints of the problem, thus helping
to smooth the solution to the problem by recognizing all of its parts.

38

Você também pode gostar