Hugging Face's logo

Datasets:
Vikhrmodels
/
programming_books_eng

Modalities:
Text
Formats:
text
Size:
1M - 10M
Libraries:
Datasets
License:
Dataset card Data Studio Files
xet
Community
programming_books_eng / J /expmath.pdf.txt
apsys's picture
apsys
Upload folder using huggingface_hub
08c8a6d verified
raw
history blame contribute delete
178 kB
 Exploring Math
Kenneth E. Iverson
Copyright © 1996-2002 Jsoftware Inc. All rights reserved.
Table Of Contents
Chapter 1
Exploration ...................................................................... 1
A. Introduction .................................................................................... 1
B. Ramble Or Research ....................................................................... 6
What Is Math.................................................................... 9
A. Relations ......................................................................................... 9
B. Proofs.............................................................................................. 13
C. Summary......................................................................................... 14
Function Tables .............................................................. 17
Grammar And Spelling ................................................... 23
A. Introduction .................................................................................... 23
B. The Use Of Grammar...................................................................... 24
C. Punctuation And Other Rules ......................................................... 25
D. Spelling........................................................................................... 27
Reports ............................................................................ 29
A. Introduction .................................................................................... 29
B. Transposition .................................................................................. 31
Terminology .................................................................... 33
Decimal and Other Number Systems ............................ 37
A. Introduction .................................................................................... 37
B. Addition .......................................................................................... 42
C. Multiplication.................................................................................. 44
D. Subtraction...................................................................................... 45
Recursion ........................................................................ 47
Proofs .............................................................................. 53
A. Introduction .................................................................................... 53
B. Inductive Proof ............................................................................... 56
Tools ................................................................................ 61
A. Introduction .................................................................................... 61
B. Editing............................................................................................. 62
C. Script Windows .............................................................................. 62
Coordinates and Visualization....................................... 63
A. Introduction .................................................................................... 63
C. Plotting Multiple Figures ................................................................ 67
D. Plotting Functions........................................................................... 68
Linear Functions ............................................................. 73
A. Distributivity................................................................................... 73
B. Linearity.......................................................................................... 74
C. Linear Vector Functions ................................................................. 75
D. Inner Product .................................................................................. 76
E. Why The Name “Linear”? .............................................................. 77
Representations of Functions ....................................... 81
A. Introduction .................................................................................... 81
Polynomials..................................................................... 85
A. Coefficients Representation............................................................ 85
B. Roots Representation ...................................................................... 86
C. Versatility........................................................................................ 87
D. Parity............................................................................................... 89
E. Linearity .......................................................................................... 90
F. Polynomial Approximations............................................................ 92
Arithmetic ........................................................................ 95
A. Introduction .................................................................................... 95
B. Insidious Inverses ........................................................................... 95
C. Rational Numbers ........................................................................... 96
D. Irrational Numbers.......................................................................... 97
E. Complex Numbers .......................................................................... 97
Complex Numbers .......................................................... 99
A. Introduction .................................................................................... 99
B. Addition .......................................................................................... 100
C. Multiplication.................................................................................. 101
D. Powers and Roots ........................................................................... 103
E. Division........................................................................................... 104
Calculus........................................................................... 107
A. Secant Slope ................................................................................... 107
B. Derivative........................................................................................ 108
C. Polynomials .................................................................................... 109
D. Differential Equations..................................................................... 110
E. The Exponential Family.................................................................. 112
Inverses and Equations.................................................. 115
A. Inverse Functions............................................................................ 115
B. Monotonic Functions ...................................................................... 116
C. Under .............................................................................................. 117
D. Equations ........................................................................................ 118
Readings.......................................................................... 121
A. Introduction .................................................................................... 121
B. Phrases ............................................................................................ 121
C. Sample Topics................................................................................. 121
D. Vocabulary and Definitions............................................................ 122
References ...................................................................... 123
Index ................................................................................ 125
1
Chapter
1
Exploration
Something lost behind the ranges
Lost and waiting for you. Go!
Kipling
A. Introduction
Exploring a city or wild park on foot is more fun, and often more instructive, than
studying it in books, lectures, or pictures. A map or other guide may be helpful,
but it is important to be able to experiment, choosing your own path, approaching
points of interest from various directions. This can give you a sense of the lay of
the land that is more useful, and more lasting, than any fixed tour of “important
points” laid out by someone else.
Matters other than landscapes may also be explored, effectively and enjoyably.
For example, to learn about clockwork, begin not with diagrams and discussions
of balance wheels, springs, and escapements, but rather with an actual old-style,
wind-up alarm clock. Explore it by first finding what can be done with it. Can
you: reset the time? make it run faster? stop it? or reset the hour hand
independently of the minute hand?
Having learned what it can do, explore the matter of how it does it, by removing
its cover, studying the works, and finally taking it apart and re-assembling it. You
may, of course, not be skillful enough to get it working again.
Exploration can also be applied to other devices that may be more interesting or
more easily available to you: toasters, typewriters, electrical toggle switches, or
door locks. But do not forget your own safety—danger lurks in electrical devices
as well as in wilderness parks. Finally, in choosing a device for exploration,
favour the older models: modern typewriters and digital clocks may be totally
inscrutable. At least one author (Ivan Illich) has claimed to see a sinister motive
in this, claiming that modern design is deliberately inscrutable in order to keep
ordinary people like us in ignorance.
But can exploration be applied to abstract, non-physical notions such as math?
Yes it can. With an ordinary hand-calculator you can explore the relation between
multiplication and addition by using it to multiply two by three, then to add two
plus two plus two, and then comparing the results. If the calculator has a button
for power, you can even explore that less-familiar notion by doing two to the
power three, and comparing the result with two times two times two.
2
Exploring Math
But the abilities of a calculator are limited, and for a general exploration of math
we will use a computer equipped with suitable software called J. It is available
from Website http://www.jsoftware.com . We will assume that you have J at hand
on a computer, and will simply show examples of exploring math with it:
3+2
5
3*2
6
3-2
1
These examples are in a uniformly-spaced font (Courier) that differs from the
Roman font used elsewhere. We will use this difference to append comments to
some of the examples. In typing the examples on your computer, enter only the
part in Courier (followed by pressing the Enter key), but do not enter anything
that appears in Roman. Thus:
3+2
5
three=:3
three+2
5
Addition
Assign the name three to 3
Use the assigned name in a sentence
b=:2
b*b
4
In experiments on a sequence of numbers, it will be easier to make the entries and
to compare the results if we treat them as a list. This may be illustrated as follows:
2*0
0
2*1
2
2*2
4
2*0,1,2,3,4,5
0 2 4 6 8 10
a=:0,1,2,3,4,5
2*a
0 2 4 6 8 10
a+a
0 2 4 6 8 10
Comparisons can be shown more clearly by using the equals function as follows:
(2*a)=(a+a)
1 1 1 1 1 1
a^2
0 1 4 9 16 25
a*a
0 1 4 9 16 25
(a^3)=(a*a*a)
1 1 1 1 1 1
Chapter 1 Exploration
3
The list a to the power 2 (that is, the square)
The cube equals a product of three factors
Lists of integers (whole numbers) are so useful that a special function is provided
for making them. Enter the following expressions, and comment on the results:
The first six non-negative integers (whole numbers)
Read aloud as a is (the list) i.6
The integers in (repeatable) random order
The even numbers (divisible by 2)
The odd numbers
i.6
0 1 2 3 4 5
a=:i.6
b=:?.~6
b
5 1 2 4 3 0
a+b
5 2 4 7 7 5
a*b
0 1 4 12 12 0
2*a
0 2 4 6 8 10
1+2*a
1 3 5 7 9 11
a=b
0 1 1 0 0 0
As shown by the last result, the lists a and b are not equal, but they are similar in
the sense that one can be obtained from the other by shuffling or permuting the
items. It is rather easy to see that a and b are similar, but for longer lists similarity
is not so easy to spot. For example, are the following lists similar?
p=:2 15 9 10 4 0 13 13 18 7 10 16 0 1 10 13 0 7 1 8
q=:7 4 7 13 0 10 1 1 2 13 13 15 0 10 9 18 10 8 0 16
A good general method for determining similarity is to first sort each list to
ascending order, and then compare the results:
sort=: /:~
sort p
0 0 0 1 1 2 4 7 7 8 9 10 10 10 13 13 13 15 16 18
sort q
0 0 0 1 1 2 4 7 7 8 9 10 10 10 13 13 13 15 16 18
4
Exploring Math
(sort p)=(sort q)
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
(sort p)-:(sort q)
1
The last sentence above uses -: to match the two lists, giving 1 if they agree in
every item, and 0 otherwise. This makes a comparison possible without reading
all the items that result from an equals comparison.
Exercises are commonly used by a student or teacher to test a student’s
understanding, in order to decide what best to do next. We will also use them to
suggest further exploration. A few tips on carrying out such explorations:
Before pressing the enter key, think through what the result should be;
experiments will teach much more if this rule is always followed.
On the other hand, do not hesitate to try anything you choose; the result may be
unintelligible or it may be an error message, but no serious harm can occur.
Use lists in experiments. Their results often show interesting patterns.
Do not hesitate to try things totally unknown. For example:
%:a
0 1 1.41421 1.73205 2 2.23607
This result will probably convince you that you have discovered the symbol for
the square root, and you might experiment further as follows:
roots=:%:a
roots*roots
0 1 2 3 4 5
• However, do not spend too much time on results that may be, at the moment,
beyond your powers. It may be better to defer further exploration until you
have learned some further math (such as complex numbers). For example:
%:-a
0 0j1 0j1.41421 0j1.73205 0j2 0j2.23607
• Explore a complex sentence by experimenting with its parts. For example:
i:4
_4 _3 _2 _1 0 1 2 3 4
i: 3
_3 _2 _1 0 1 2 3
Function for symmetric lists
+:3
6
>: +:3
7
>:@+:3
of) g
7
The function f@g is f atop (applied to the result
Chapter 1 Exploration
5
i.@>:@+:3
0 1 2 3 4 5 6
]3
3
Identity function
Exercises
1. What are the commonly-used names for the functions (or verbs) denoted here
by + * -
[plus times minus or addition multiplication (or product) subtraction]
2. Enter plus=:+ to assign the name plus to the addition function, and then
experiment with the following expressions:
3 plus 4 * 2
11
zero=:0
one=:1
two=:2
three=:3
four=:4
times=:*
three plus four times two
3. As illustrated by the preceding exercise, much math could be expressed in
English words without forcing students to learn the “difficult” special notation
of math. Would you prefer to stick to English words?
4. Experiment with the following editing facilities for correcting errors:
• Correct a line being entered by using the cursor keys (marked with
arrows) to move the cursor to any point, and then type or erase (using the
delete or backspace keys). The cursor need not be returned to the end of
the line before entering the line.
• Revise any line by moving the cursor up to it and pressing enter to bring it
down to the input area for editing.
Not only is it important to think through the expected result of an experiment
before executing it on the computer, but it is also a good practice to look for
patterns in any lists or tables you may see. Then verify your observations by
doing calculations by hand for short lists, and then test them more thoroughly on
the computer. For example, the list of odd numbers:
1+2*a
1 3 5 7 9 1
may be added by hand to give 36. Now add only the first five of the list, the first
four, and so on down to the first one.
Do you see a pattern in these results? If not, compare them with the following list
of squares:
(1+a)*(1+a)
1 4 9 16 25 36
6
Exploring Math
It appears that for any value of n, the sum of the first n odd numbers is simply the
square of n. This may be tested further as follows:
n=:20
a=:i.n
odds=:1+2*a
odds
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
sum=:+/
sum odds
400
n*n
400
The sum function +/ gives the sum of its arguments, but calculation of the
subtotals (the sum of the first one, the first two, etc.) would provide a more
thorough test. Thus:
sum\ odds
1 4 9 16 25 36 49 64 81 100 121 144 169 196 225 256 289 324 361
400
(1+a)*(1+a)
1 4 9 16 25 36 49 64 81 100 121 144 169 196 225 256 289 324 361
400
Hereafter we will suggest many experiments without showing the results,
expecting students to use the computer to produce them.
B. Ramble or Research
The main point of this book is to introduce a new tool for exploring math, and to
foster its use by applying it to a variety of topics. In other words, it provides a
ramble through a variety of topics rather than a systematic study of any one of
them.
Rambles through any subject can be much more rewarding, and more self-
directed, if one has a systematic knowledge of at least some aspect of it. For
example, amateur shell-collecting is more interesting to one with some knowledge
of molluscs and their classification; walks through parks are more rewarding to
one with some systematic knowledge of plant, animal, or insect life; and walks
through hills and mountains are made more interesting by a knowledge of
elementary geology.
However, any book on rambling would surely fail if stuffed with serious
digressions on the systematic study of each interesting point as it is discovered. It
is better to provide the reader with effective but unobtrusive pointers to other
sources.
Books 2 and 3 provide deeper studies of two branches of math: arithmetic and
calculus. Being that branch of math that deals with whole numbers, arithmetic is
the most elementary and accessible of subjects in math but, as treated in Book 2,
Chapter 1 Exploration
7
it also provides simple introductions to many more advanced topics, including
proofs, permutations, polynomials, logic, and sets.
These books are easy to consult because they use the same J notation. Moreover,
they incorporate more systematic introductions and discussions of the notation
itself. Further texts of this character include Reiter’s Fractals, Visualization and J
[1], and Concrete Math Companion [2].
On the other hand, treatments in conventional notation of a wide variety of topics
are more readily available in libraries. Use of them in conjunction with the
present text will require sometimes difficult translations between J and
conventional notation. However, the effort of translation is often richly repaid (as
it is in translating from one natural language to another) by deeper understanding
of the matters under discussion.
In fact, a deep appreciation of the method of exploration proposed here may best
be found in an attempt to write a companion volume to some chosen conventional
text. Some guidance in such an endeavour is provided by Concrete Math
Companion [2], published as a companion to Concrete Mathematics [3].
9
Chapter
2
What Is Math
math is the short form of mathematics, for
which the British use maths, preserving
the ugly plural form for a singular noun.
A. Relations
It is commonly thought that math is about numbers. So it is, but numbers are not
the only, nor even the most important, concern of math. It would be more accurate
to say that math is concerned with relations, and with proofs of relations.
Although the first chapter dealt only with numbers, it should be clear that the
interesting aspects were the relations between results. For example:
The first six non-negative integers
The integers in random order
The relation between multiplication and
The lists a and b are not equal
a=:i.6
b=:?.~6
b
5 1 2 4 3 0
3*a
0 3 6 9 12 15
a+a+a
0 3 6 9 12 15
(3*a)=(a+a+a)
addition
1 1 1 1 1 1
a=b
0 1 1 0 0 0
sort=:/:~
sort b
0 1 2 3 4 5
sort a
0 1 2 3 4 5
(sort a)=(sort b)
1 1 1 1 1 1
But are similar; one is a permutation of the other
10
Exploring Math
We will further illustrate this matter of relations by examples that do not concern
numbers. For example, the word 'POST' is said to be an anagram of the word
'SPOT' because the letters of 'SPOT' can be permuted to give the word 'POST'.
Thus 'SPOT' and 'POST' are similar in the sense already defined for lists. The
similarity of these words may be tested as follows:
w=:'SPOT'
x=:'POST'
sort w
OPST
sort x
OPST
(sort w)=(sort x)
1 1 1 1
Sorting w produces OPST. Is it an anagram? We will say that it is, although it is not
an English word.
You could (and should) attempt to write down all distinct anagrams of 'SPOT',
finding a surprising number of English words among them. However, this might
be rather difficult to do; in a long list of words it is easy to overlook repetitions,
and you may not even know how many anagrams to expect all together.
We will now use the anagram function A. for this purpose. Its left argument
chooses one of many permutations to apply to the list right argument. Thus:
w
SPOT
8 A. w
POST
12 A. 8 A. w
SPOT
The permutation 12 A. is the inverse of 8 A.
0 1 2 3 4 5 6 7 8 A. w
SPOT
SPTO
SOPT
SOTP
STPO
STOP
PSOT
PSTO
POST
30 A. w
|index error
| 30 A.w
The last result shows that there is a limit to the valid left argument; properly so,
since there is a limit to the number of different permutations of a list. But how
many are there? In the case of a two-item list 'AB' there are clearly only two
possibilities, the identity permutation that leaves the list unchanged, and the one
that gives 'BA'. Thus:
Chapter 2 What is Math
11
0 1 A. 'AB'
AB
BA
Write down all permutations of the list 'ABC' to convince yourself that there are
six possible permutations. Thus:
(i.6)A.'ABC'
ABC
ACB
BAC
BCA
CAB
CBA
Exercises
1. Produce all anagrams of various three-letter English words to find those
words that have the largest number of proper English words among their
anagrams.
2. Did you find any word more prolific than 'APT'?
3. Find all English words among the anagrams of 'SPOT'.
In solving the last exercise above, it was necessary to find the largest left
argument of A. permitted. This could be done by experiment. Thus:
22 A. 'SPOT'
TOSP
23 A. 'SPOT'
TOPS
24 A. 'SPOT'
|index error
| 24 A.'SPOT'
(i.24)A.'SPOT'
SPOT
SPTO
SOPT
SOTP
STOP
STOP
PSOT
PSTO
POST
POTS
PTSO
PTOS
OSPT
OSTP
OPST
OPTS
12
Exploring Math
OTSP
OTPS
TSPO
TSOP
TPSO
TPOS
TOSP
TOPS
But what is the general relation between the number of permutations and the
number of items in the list to be permuted? Although we are dealing with English
words and anagrams rather than with numbers, this is a proper mathematical
question because it concerns relations. The question can be answered in the
following steps:
In a four-letter word, the first position in an anagram can be filled in any one of
four ways.
Having filled the first position, the next can be filled from the remaining three
letters in three different ways.
The next position can be filled in two ways.
The last position can be filled in one way.
The total number of ways is the product of these, that is, four times three times
two times one.
This product over all integers up to a certain limit (4 in the present example) is so
useful that it is given its own name (factorial) and symbol (!). Thus:
!4
24
4*3*2*1
24
!0 1 2 3 4 5 6 7
1 1 2 6 24 120 720 5040
The number of items in a list is a function that is also provided with a symbol:
w3=:'APT'
#w3
3
i.!#w3
0 1 2 3 4 5
(i.!#w3)A.w3
APT
ATP
PAT
PTA
TAP
TPA
Exercises
Chapter 2 What is Math
13
4. Comment on the following experiments:
sort=:/:~
w=:'SPOT'
sort w
table=:(i.!#w)A. w
# table sort table
5. A table with more rows than columns may be displayed more compactly by
transposing it. Try the following:
transpose=:|:
transpose table
The function A. applies to lists of numbers as well as to lists of letters (words),
and when applied to lists such as i.3 and i.4 produces tables that show its
behaviour more clearly. The following experiment uses the link function (;) to
box tables and link them together for more convenient comparison:
i=:i.24
(i A. 'SPOT');(i A. 'ABCD');(i A. 0 1 2 3)
+----+----+-------+
|SPOT|ABCD|0 1 2 3|
|SPTO|ABDC|0 1 3 2|
|SOPT|ACBD|0 2 1 3|
|SOTP|ACDB|0 2 3 1|
|STPO|ADBC|0 3 1 2|
|STOP|ADCB|0 3 2 1|
|PSOT|BACD|1 0 2 3|
|PSTO|BADC|1 0 3 2|
|POST|BCAD|1 2 0 3|
|POTS|BCDA|1 2 3 0|
|PTSO|BDAC|1 3 0 2|
|PTOS|BDCA|1 3 2 0|
|OSPT|CABD|2 0 1 3|
|OSTP|CADB|2 0 3 1|
|OPST|CBAD|2 1 0 3|
|OPTS|CBDA|2 1 3 0|
|OTSP|CDAB|2 3 0 1|
|OTPS|CDBA|2 3 1 0|
|TSPO|DABC|3 0 1 2|
|TSOP|DACB|3 0 2 1|
|TPSO|DBAC|3 1 0 2|
|TPOS|DBCA|3 1 2 0|
|TOSP|DCAB|3 2 0 1|
|TOPS|DCBA|3 2 1 0|
+----+----+-------+
B. Proofs
Although proofs are an important (and many would say the essential) part of
mathematics, we will spend little time on them in this book.
In introducing his book Proofs and Refutations: The Logic of Mathematical
Discovery [4], Imre Lakatos makes the following point:
Its modest aim is to elaborate the point that informal, quasi-
empirical, mathematics does not grow through a monotonous
increase of the number of indubitably established theorems but
14
Exploring Math
through the incessant improvement of guesses [Italics added] by
speculation and criticism, by the logic of proofs and refutations.
The main point of the present book is to exploit a new tool for the exploration of
relations and patterns that can be used by both mathematicians and laymen to find
those guesses that are amenable to, and worthy of, proof. We will defer further
discussion of proofs to Chapter 9, partly to allow the reader to garner guesses that
can be used to illuminate the discussion.
We will, however, recommend the reading of Lakatos at any point. The book is
highly entertaining, instructive, and readable by any layman with the patience to
look up the meanings of a small number of words such as polyhedron, polygon,
and convex.
The following quotes from Lakatos reflect his view of the importance of
guessing:
Just send me the thereoms, then I shall find the proofs.
Chrysippus
I have had my results for a long time, but I do not yet know how I am to
arrive at them.
Gauss
If only I had the theorems! Then I should find the proofs easily enough.
Riemann
I hope that now all of you see that proofs, even though they may not
prove, certainly do help to improve our conjecture.
Lakatos
On the other hand those who, because of the usual deductive
presentation of mathematics, come to believe that the path of discovery
is from axioms and/or definitions to proofs and theorems, may
completely forget about the possibility and importance of naive
guessing.
Lakatos
Exercises
6. Read the three pages of Section C, Chapter 5, of Book 2.
C. Summary
In brief, we will interpret math in the following sense: it concerns relations, and
provides languages for expressing them, as well as for expressing transformations
on tangible representations.
For example, the first four counting numbers can be represented by the list of
symbols
1 2 3 4:
! 1 2 3 4
1 2 6 24
A transformation (or function)
*/\ 1 2 3 4
1 2 6 24
Chapter 2 What is Math
15
A second transformation
(! 1 2 3 4)=(*/\ 1 2 3 4)
1 1 1 1
Equivalent to the first
17
Chapter
3
Function Tables
The pleasures of the table
belong to all ages
Jean Anthelme Brillat-Savarin
and make it plain upon tables
that he may run that readeth it
Habakkuk
The effect of multiplication can be shown rather neatly in a succession of
products of a list as follows:
a=: i.6
0*a
0 0 0 0 0 0
1*a
0 1 2 3 4 5
2*a
0 2 4 6 8 10
However, a more perspicuous table of products with each item of a can be
prepared as follows:
a*/a
0 0 0 0 0 0
0 1 2 3 4 5
0 2 4 6 8 10
0 3 6 9 12 15
0 4 8 12 16 20
0 5 10 15 20 25
Similar tables can be prepared for other known functions. For example:
(a*/a);(a+/a);(a-/a)
+---------------+------------+----------------+
|0 0 0 0 0 0|0 1 2 3 4 5|0 _1 _2 _3 _4 _5|
|0 1 2 3 4 5|1 2 3 4 5 6|1 0 _1 _2 _3 _4|
|0 2 4 6 8 10|2 3 4 5 6 7|2 1 0 _1 _2 _3|
|0 3 6 9 12 15|3 4 5 6 7 8|3 2 1 0 _1 _2|
|0 4 8 12 16 20|4 5 6 7 8 9|4 3 2 1 0 _1|
|0 5 10 15 20 25|5 6 7 8 9 10|5 4 3 2 1 0|
+---------------+------------+----------------+
18
Exploring Math
Much can be learned from such tables. For example, the multiplication table is
symmetric, that is, each row is the same as the corresponding column, and its
transpose (|:a*/a) is the same as the table a*/a itself. This implies that the
arguments of multiplication may be exchanged without changing the product, or,
as we say, multiplication is commutative. The same may be said of addition, but
not of subtraction, which is non-commutative, as is obvious from its table. Tables
for both negative and positive arguments are even more interesting. For example,
try each of the three tables with the following symmetric argument:
i: 6
_6 _5 _4 _3 _2 _1 0 1 2 3 4 5 6
Note how the multiplication table is broken into quadrants of exclusively positive
or exclusively negative numbers by the row and column of zeros, and try to
explain why this occurs.
The symbol / in the sentence a*/a denotes an adverb, because it applies to the
verb * to produce a related verb (that is in turn used to produce a table).
It is much easier to interpret a table if it is bordered by its arguments. We will use
a second adverb called table for this purpose. For example:
b=:2 3 5 7 11
a *table b Bordered multiplication table
+-+--------------+
| | 2 3 5 7 11|
+-+--------------+
|0| 0 0 0 0 0|
|1| 2 3 5 7 11|
|2| 4 6 10 14 22|
|3| 6 9 15 21 33|
|4| 8 12 20 28 44|
|5|10 15 25 35 55|
+-+--------------+
+table~ a Bordered addition table
+-+------------+
| |0 1 2 3 4 5|
+-+------------+
|0|0 1 2 3 4 5|
|1|1 2 3 4 5 6|
|2|2 3 4 5 6 7|
|3|3 4 5 6 7 8|
|4|4 5 6 7 8 9|
|5|5 6 7 8 9 10|
+-+------------+
*table~ i:6
+--+-----------------------------------------------+
| | _6 _5 _4 _3 _2 _1 0 1 2 3 4 5 6|
+--+-----------------------------------------------+
|_6| 36 30 24 18 12 6 0 _6 _12 _18 _24 _30 _36|
|_5| 30 25 20 15 10 5 0 _5 _10 _15 _20 _25 _30|
|_4| 24 20 16 12 8 4 0 _4 _8 _12 _16 _20 _24|
|_3| 18 15 12 9 6 3 0 _3 _6 _9 _12 _15 _18|
|_2| 12 10 8 6 4 2 0 _2 _4 _6 _8 _10 _12|
|_1| 6 5 4 3 2 1 0 _1 _2 _3 _4 _5 _6|
| 0| 0 0 0 0 0 0 0 0 0 0 0 0 0|
Chapter 3 Function Tables 19
| 1| _6 _5 _4 _3 _2 _1 0 1 2 3 4 5 6|
| 2|_12 _10 _8 _6 _4 _2 0 2 4 6 8 10 12|
| 3|_18 _15 _12 _9 _6 _3 0 3 6 9 12 15 18|
| 4|_24 _20 _16 _12 _8 _4 0 4 8 12 16 20 24|
| 5|_30 _25 _20 _15 _10 _5 0 5 10 15 20 25 30|
| 6|_36 _30 _24 _18 _12 _6 0 6 12 18 24 30 36|
+--+-----------------------------------------------+
Tables also provide an interesting and effective way to explore unfamiliar functions.
Often, the display of a bordered function table provides a precise and easily-remembered
definition of the function. For example:
Relation
<table~ i:5
+--+--------------------------+
| |_5 _4 _3 _2 _1 0 1 2 3 4 5|
+--+--------------------------+
|_5| 0 1 1 1 1 1 1 1 1 1 1|
|_4| 0 0 1 1 1 1 1 1 1 1 1|
|_3| 0 0 0 1 1 1 1 1 1 1 1|
|_2| 0 0 0 0 1 1 1 1 1 1 1|
|_1| 0 0 0 0 0 1 1 1 1 1 1|
| 0| 0 0 0 0 0 0 1 1 1 1 1|
| 1| 0 0 0 0 0 0 0 1 1 1 1|
| 2| 0 0 0 0 0 0 0 0 1 1 1|
| 3| 0 0 0 0 0 0 0 0 0 1 1|
| 4| 0 0 0 0 0 0 0 0 0 0 1|
| 5| 0 0 0 0 0 0 0 0 0 0 0|
+--+--------------------------+
(<table~ a),.(=table~ a),.(>table~ a) Relations
+-+-----------+-+-----------+-+-----------+
| |0 1 2 3 4 5| |0 1 2 3 4 5| |0 1 2 3 4 5|
+-+-----------+-+-----------+-+-----------+
|0|0 1 1 1 1 1|0|1 0 0 0 0 0|0|0 0 0 0 0 0|
|1|0 0 1 1 1 1|1|0 1 0 0 0 0|1|1 0 0 0 0 0|
|2|0 0 0 1 1 1|2|0 0 1 0 0 0|2|1 1 0 0 0 0|
|3|0 0 0 0 1 1|3|0 0 0 1 0 0|3|1 1 1 0 0 0|
|4|0 0 0 0 0 1|4|0 0 0 0 1 0|4|1 1 1 1 0 0|
|5|0 0 0 0 0 0|5|0 0 0 0 0 1|5|1 1 1 1 1 0|
+-+-----------+-+-----------+-+-----------+
Power and “outof”
(^table~ a),.(!table~ a)
+-+-------------------+-+------------+
| |0 1 2 3 4 5| |0 1 2 3 4 5|
+-+-------------------+-+------------+
|0|1 0 0 0 0 0|0|1 1 1 1 1 1|
|1|1 1 1 1 1 1|1|0 1 2 3 4 5|
|2|1 2 4 8 16 32|2|0 0 1 3 6 10|
|3|1 3 9 27 81 243|3|0 0 0 1 4 10|
|4|1 4 16 64 256 1024|4|0 0 0 0 1 5|
|5|1 5 25 125 625 3125|5|0 0 0 0 0 1|
+-+-------------------+-+------------+
%: table~ a
+-+-----------------------------------+
| |0 1 2 3 4 5|
+-+-----------------------------------+
Roots
20
Exploring Math
|0|0 1 _ _ _ _|
|1|0 1 2 3 4 5|
|2|0 1 1.41421 1.73205 2 2.23607|
|3|0 1 1.25992 1.44225 1.5874 1.70998|
|4|0 1 1.18921 1.31607 1.41421 1.49535|
|5|0 1 1.1487 1.24573 1.31951 1.37973|
+-+-----------------------------------+
Exercises
1. Produce and examine bordered tables for the following functions:
<. >.
<: >:
%
2.
Produce and examine bordered tables for the following “commuted”
functions:
<.~ >.~
<:~ >:~
%~
3. Produce and examine bordered tables for the following Greatest Common
Divisor and Least Common Multiple functions:
+. *.
In particular, apply them to the argument 0 1 (as in +.table 0 1) and note
that with the interpretation of “true” for 1 and “false” for 0 (as was done by
Boole), they then represent the logical functions “or” and “and”
4. Explain the equality denoted by the following sentence:
(e>:/e)=(e>/e)+.(e=/e=:s 6)
5. First enter:
at=:+/~ e
mt=:*/~ e
st=:-/~ e
dt=:%/~ e
trans=:|:
Then comment on the results of the following:
at-:trans at
mt-trans mt
st+trans st
dt*trans dt
The following exercises suggest a sequence of experiments that should be tried
only after reviewing the tips on explorations given in Chapter 1:
6. a=:i.6
+:a
-:a
(+:a)-(-:a)
(+:--:)a
Exercises
Double minus half
Dmh=:+:--:
Dmh a
7. Contrast the result of the following sentence with those of Exercise 6:
+:--:a
Chapter 3 Function Tables 21
+/%#)a
8.
Average=:+/%#
Average a
Average 3 1 4 1 6
9. Re-enter the sentence (a*/a);(a+/a);(a-/a) from the beginning of this
chapter, and compare the result with the following:
a(*/;+/;-/)a
f=:*/;+/;-/
a f a
f~ a
23
Chapter
4
Grammar And Spelling
The level is low
but it has not fallen
Jacques Barzun
I can spell all the words that I use
and my grammar’s as good as my neighbour’s
W.S. Gilbert
A. Introduction
We have already made significant use of J, why trouble us now with its grammar?
On the other hand, if grammar is important, why was it not treated first?
In learning our native language we spend years at it and become quite proficient
before we even hear of grammar. However, grammar becomes important for more
advanced use of the language in clear writing and speaking. Moreover, the
teaching of grammar relies on many examples of the use of the language that
would not be familiar to a beginner.
Similarly, more advanced and independent writing in J will require knowledge of
its grammar. Moreover, we will find it helpful to refer to sentences from earlier
chapters to illustrate and motivate discussions of the grammar.
In learning a second language a student has the advantage of already appreciating
the purposes and value of language, as well as some knowledge of grammar from
her native tongue. On the other hand, one may be seriously misled by such
knowledge, and the student is sometimes best advised to forget her native
language as much as possible: one may know too many things that are not true.
The beginner in J will already know much of two relevant languages: English,
and Mathematical Notation (to be referred to as MN). The knowledge of English
grammar is very helpful, especially when we recognize certain analogies between:
• English verbs (action words) and functions such as + and - and *
• Nouns on which verbs act, and the arguments (such as 3 and 4 and 'STOP') to
which functions apply
• Pronouns such as a and b and mt used in the preceding chapter, and pronouns
such as “it” and “she” used in English
• Adverbs (such as table in the preceding chapter) that apply to verbs
(functions) to produce different, but related, verbs
24
Exploring Math
Knowledge of MN can be very helpful, particularly in providing familiarity with
numbers and symbols for common functions, and with some of the purposes of
math. On the other hand, MN can be very misleading because it shows little
concern for simple and consistent grammar. For example:
• The simple forms a+b and a*b used for some functions of two arguments is
abandoned in others, as in xn for the x^n used in J, and in (n
number of ways of choosing m things from n)
m) for m!n (the
• The rule that a function of one argument precedes its argument (as in -b and
sqrt b) is abandoned in the case of the factorial (n!). In J this is written as
!n.
• The ambivalent use of the minus sign to denote two different functions as
determined by the number of arguments provided (subtraction in a-b, and
negation in -b) is not extended to all functions as it is in J. For example, a%b
and %b denote divided by and reciprocal; a^b and ^b denote power and
exponential; and a+/b and +/b denote the addition table and sum over.
• The imposition of hierarchical rules of execution for certain functions: power
is performed before multiplication and division, which are performed before
addition and subtraction. The reasons for the development of such rules in
MN lie in the expressions used for polynomials, and will be discussed further
in the corresponding chapter.
B. The Use of Grammar
The rules of grammar determine how a sentence is to be parsed, that is, the order
in which its parts are to be interpreted or executed. In particular, these rules cover
the use of punctuation, which can make an enormous difference, as illustrated by
the following sentences:
The teacher said George was stupid
The teacher, said George, was stupid
The punctuation in J is provided by parentheses, as illustrated by the following
sentences from Chapter 2:
a=:i.6
b=:?.~6
(3*a)=(a+a+a)
1 1 1 1 1 1
3*a=a+a+a
3 0 0 0 0 0
Removal of the punctuation yields a quite different result
The parsing of a sentence does not depend on the particular word used, but only
on the class to which it belongs. Thus the English examples used above would be
parsed without change if the nouns farmer and Mary were substituted for the
nouns teacher and George. Similarly, the sentence (3*b)=(b+b+b) would parse
the same as (3*a)=(a+a+a).
The classes concerned are called the parts of speech. J has only six parts of
speech (including the punctuation provided by parentheses), all of which have
Chapter 4 Grammar and Spelling
25
been used in earlier chapters. For example, the nouns 3 and 2, and the verbs + and
* and - occur in the first three sentences in Chapter 1, and the copula =:
(analogous to the copulas is and are in English) occurs in the next.
As in English, an adverb applies to a verb to produce a related verb. Examples
occurring in Chapter 1 are:
The adverb / which inserts its argument function between items of the noun to
which it applies. For example, +/1 2 3 4 is equivalent to 1+2+3+4, and the
function +/ may therefore be called the sum function.
The adverb \ which uses its argument function to scan all prefixes of its noun
argument: +/\1 2 3 is equivalent to (+/1),(+/1 2),(+/1 2 3).
In English, the phrase “run and hide” uses the copulative conjunction “and” to
produce a new verb that is a composition of the actions described by the verbs
“run” and “hide”. In J, @: is a conjunction that applies its first argument verb to
the result of its second argument verb. For example:
a
0 1 2 3 4 5
b
5 1 2 4 3 0
a-b
_5 0 0 _1 1 5
+/a-b
0
a +/@:- b
0
sumdif=:+/@:-
1 2 3 4 5 sumdif 2 3 5 7 11
_13
Exercises
1. Search earlier chapters for further examples of the various parts of speech.
2. State the effect of the adverb ~ in the sentences a-~b and a^~b.
C. Punctuation and Other Rules
In J, a sentence can be completely punctuated so that the only grammatical rule
needed to parse it concerns the use of parentheses. For example, the area of a
rectangular field can be computed as follows:
Length=:8
Width=:6
Area=:Length*Width
Area
48
26
Exploring Math
If instead the width and the length of the roll of wire available to enclose the field
are given, the area may be computed as follows:
Roll=:32
Sides=:Roll-(Width+Width)
Length=:Sides%2
Length*Width
60
Extent available for other two sides
Area for given roll and width
The whole may be re-expressed as a single sentence punctuated as follows:
Area=:((Roll-(Width+Width))%2)*Width
Although long names such as Width and Roll can be helpful in understanding the
point of a sentence, they can also obscure its structure. Briefer (but still
mnemonic) names may be substituted:
W=:Width
P=:Roll
field
A=:((P-(W+W))%2)*W
An abbreviation for the perimeter of the
Other grammatical rules make it possible to omit some parentheses. The next rule
(after the rule for parentheses) is:
• A sentence is executed from right to left
Consequently, the phrase (P-(W+W)) may be re-written as (P-W+W). Hence:
A=:((P-W+W)%2)*W
This can be further simplified by using the fact that multiplication is
commutative:
A=:W*((P-W+W)%2)
A=:W*(P-W+W)%2
Since division is not commutative, this trick cannot be repeated, but because
division by two is equal to multiplication by one-half, we have:
A=:W*(P-W+W)*0.5
A=:W*0.5*(P-W+W)
A=:W*0.5*P-W+W
Although an unparenthesized sentence or phrase is executed from right to left, it is
easily read from left to right. To illustrate this we will use the right-to-left
execution rules to fully parenthesize the last sentence above:
A=:(W*(0.5*(P-(W+W))))
This can now be read from left to right as follows: A is W times the value of the
entire phrase that follows it, which in turn is 0.5 times the phrase that follows it,
and so on.
The foregoing example made no use of adverbs and conjunctions, and for a
sentence that does include them we need a further rule:
• Adverbs and conjunctions are applied before verbs.
For example:
Chapter 4 Grammar and Spelling
27
+/a*b is equivalent to (+/)a*b
^&3 a+b is equivalent to (^&3)a+b
A complete formal statement of the grammar of J may be found in J Dictionary
[5], which is also available on the computer by using the Help menu. This
statement of the grammar should perhaps be studied at some point, but it is
probably better to begin by reviewing familiar sentences and trying to apply the
grammatical rules to them. You might review the sentences of earlier chapters as
follows:
• Modify and simplify them, using the methods suggested in the foregoing
examples (as well as any others that occur to you).
• Try to read the resulting sentences from left to right, using English to
paraphrase them.
• Assign values to any names used in the sentences so that they may be entered
for execution. If you modify a sentence in any way that changes its meaning,
you will probably be alerted to the fact by seeing a different result upon
entering it.
The following Exercises highlight points that you might well miss in your review.
Exercises
3. Comment on the sentence a=:0,1,2,3,4,5 used in Chapter 1 to introduce the
first example of a list.
[The comma denotes a catenate verb that appends one list (or a single item) to
another. Also experiment with other forms of catenate as in:
b=:i.-6
a,b
a,.b
a,:b
a;b
Called stitch
Called laminate
Called link ]
4. Why is it possible to enter a list of numbers as in a=:0 1 2 3 4 5 as well as
by using the catenate function as in Exercise 3?
[Certain results that can be produced by functions can also be entered more
simply as constants. For example:
3-5
3+8%10
3%5
3+j.4
2,3,5,7
This sentence
this constant
is equivalent to
_2
3.8
3r5
3j4
2 3 5 7 ]
5. Read the first five pages of Part II (Grammar) of J Dictionary [5] (also
available in Help, as described in Chapter 10).
D. Spelling
The many words in English are each represented by one or more letters from a
rather small alphabet. The words (nouns, verbs, etc.) of J are each represented by
28
Exploring Math
one or more characters from an alphabet of letters and other symbols. For
example:
+ +. +: & i. A.
Every word of more than one character ends with a dot or a colon.
Any other sequence beginning with a letter and continuing with letters or digits
(but not ending with a dot or colon) is a name that may be used with a copula, as
in the following examples:
a=:i.6
plus=:+
g=:/\
Pronoun
Proverb
Pro-adverb
p3=:^&2
Proverb
The representation of numbers is illustrated by:
2 and 2.4 and 0.4
_2 and _2.4 and _0.4
A decimal point must be preceded by a negative sign or at least one digit. As
shown in Exercise 4, an r may be used in a number to denote a rational fraction
(as in 2r3 for two-thirds), and a list may be represented by a list of numbers (as in
2.3 2r3 4).
The spelling rules of J determine how words are formed from the string of
characters that comprise a sentence. They can be clarified by applying the word-
formation verb to a (quoted) sentence. For example:
;: '+/4 3 2 1*/i.6'
+-+-+-------+-+-+--+-+
|+|/|4 3 2 1|*|/|i.|6|
+-+-+-------+-+-+--+-+
It should also be noted that redundant spaces may be inserted in a sentence to
improve readability, as in a=: i. 6 instead of a=:i.6 .
29
Chapter
5
Reports
Cornelius the centurion,
a man of good report
Acts
A. Introduction
If a is a list of twelve monthly receipts for a year, then a quarter-by-month report
of the same receipts can be obtained as follows:
]qm=:4 3$ a=:1 7 4 5 2 0 6 6 9 3 5 8
1 7 4
5 2 0
6 6 9
3 5 8
The sum over the quarters is given by:
+/qm
15 20 21
A two-year report for constant receipts of 10 can be obtained by:
ten=:2 4 3$10
ten
10 10 10
10 10 10
10 10 10
10 10 10
10 10 10
10 10 10
10 10 10
10 10 10
A more realistic report can be obtained by applying the repeatable random
generator to this array:
30
Exploring Math
yqm=:?.ten
yqm
1 7 4
5 2 0
6 6 9
3 5 8
0 0 5
6 0 3
0 4 6
5 9 8
The sums over the years of this report are:
+/yqm
1 7 9
11 2 3
6 10 15
8 14 16
Because yqm has three categories or axes, we call it a rank-3 report or array. Its
rank-2 cells are the two quarter-by-month tables seen in its display, and its rank-1
cells are the eight rows (arranged, in effect, in a 2 by 4 array).
The sums over the quarters in each year are the sums over the two rank-2 cells,
yielding a 2 by 3 array (for the two years and three months in each quarter). Thus:
+/"2 yqm
15 20 21
11 13 22
Similarly, the sums over the three months in each quarter are a 2 by 4 array given
by:
+/"1 yqm
12 7 21 16
5 9 10 22
Exercises
1. Enter the foregoing expressions, and verify that they reproduce the foregoing
results.
2. The function ?. reproduced the same result because it is a repeatable random
number generator. Try the expression ?ten several times to show that the
results do not repeat.
3. Predict and verify the results of +/"3 yqm and +/"0 yqm.
4. Experiment with the box function, as in <3 4 5 and <yqm and <"2 yqm and
<"1 yqm.
5. The sentence $yqm gives the shape of the array yqm. Apply $ to other results
such as +/yqm and +/"2 yqm and +/"1 yqm.
Chapter 5 Reports
31
6. The function # gives the number of items or major cells in its argument.
Apply it to various arguments.
The expression f"k can be used to apply any function f to the rank-k cells of its
argument. For example, the mean or average function can be used as follows:
mean=:+/%#
mean 3 4 5 6
4.5
(mean;mean"2;mean"1) yqm
+-----------+--------------+-------------------------------+
|0.5 3.5 4.5| | |
|5.5 1 1.5|3.75 5 5.25| 4 2.33333 7 5.33333|
| 3 5 7.5|2.75 3.25 5.5|1.66667 3 3.33333 7.33333|
| 4 7 8| | |
+-----------+--------------+-------------------------------+
Exercises
7. Experiment with rank cases of the following functions, and state in English
the meanings of the various results:
|.
Reverse
2&|.
Rotate
#
$
Number of items
Shape
B. Transposition
Given a year-by-quarter-by-month report yqm we may want to see the receipts
displayed as a quarter-by-month-by-year report qmy. If we refer to the successive
axes (or categories) by the indices 0 1 2, we may say that qmy is to be obtained
by the transposition 1 2 0 (choosing axis 1, then axis 2, then axis 0). Thus:
qmy=:1 2 0 |: yqm
qmy;yqm;($qmy);($yqm)
+---+-----+-----+-----+
|1 0| | | |
|7 0| | | |
|4 5| | | |
| |1 7 4| | |
|5 6|5 2 0| | |
|2 0|6 6 9| | |
|0 3|3 5 8| | |
| | |4 3 2|2 4 3|
|6 0|0 0 5| | |
|6 4|6 0 3| | |
|9 6|0 4 6| | |
| |5 9 8| | |
|3 5| | | |
|5 9| | | |
|8 8| | | |
+---+-----+-----+-----+
(mean;mean"2;mean"1) qmy
+---------+---------------+-----------+
|3.75 2.75| 4 1.66667|0.5 3.5 4.5|
32
Exploring Math
| 5 3.25|2.33333 3|5.5 1 1.5|
|5.25 5.5| 7 3.33333| 3 5 7.5|
| |5.33333 7.33333| 4 7 8|
+---------+---------------+-----------+
Transpositions may also be used on higher-rank arrays, as in the following
product-by-year-by-quarter-by-month report:
pyqm=: ?. 2 2 4 3$10
ypmq=: 1 0 3 2 |: pyqm
ypmq ([;$@[;];$@])pyqm
Boxing of various ranks can also be used to clarify displays:
<"2 ypmq
+-------+-------+
|1 5 6 3|5 4 7 7|
|7 2 6 5|0 7 2 3|
|4 0 9 8|6 9 0 6|
+-------+-------+
|0 6 0 5|7 2 7 6|
|0 0 4 9|9 9 6 8|
|5 3 6 8|3 7 0 2|
+-------+-------+
<"3 ypmq
+-------+-------+
|1 5 6 3|0 6 0 5|
|7 2 6 5|0 0 4 9|
|4 0 9 8|5 3 6 8|
| | |
|5 4 7 7|7 2 7 6|
|0 7 2 3|9 9 6 8|
|6 9 0 6|3 7 0 2|
+-------+-------+
33
Chapter
6
Terminology
If this young man expresses himself in terms too deep for me,
Oh what a singularly deep young man this deep young man must be
W.S. Gilbert
Special terminology used in various branches of knowledge often imposes a
serious burden on a beginner. It may sometimes be safely dismissed as pretentious
and no better than familiar terms, but serious treatment of a topic may well
require finer distinctions than those provided by familiar language. For example,
the familiar average may sometimes be substituted for mean as defined in math
and statistics. However, mean refers not only to average (the arithmetic mean),
but also to various ways of characterizing a collection by a single number,
including the geometric mean, harmonic mean, and common mean.
Similarly, the grammatical terms adopted in J (from English) may seem
pretentious to anyone familiar with corresponding terms in math, but they make
possible significant distinctions that are not easily made in MN. We illustrate this
by a few sentences and the classification of items from them in both J and MN:
with=:&
cube=:^ with 3
commute=:~
into=:% commute
pi=:7 into 22
2 into cube a=:i.6
Noun
Pronoun
Verb or Function
Proverb
Adverb or Operator
Pro-adverb
Conjunction or Operator
Pro-conjunction
List or Vector
Table or Matrix
Report or Array
22
pi
%
cube
~
commute
&
with
a
a*/a
a+/a*/a
J MN
Constant
Variable
Function or Operator
Operator
Operator
Vector
Matrix
Array
34
Exploring Math
In the foregoing, MN makes the same distinction made by noun and pronoun in J,
but uses the terms constant and variable. The term variable may prove somewhat
misleading, particularly when used for a pronoun such as pi (for the ratio of the
circumference to the diameter of a circle), which is not expected to vary. The
following sentences may be used to clarify the choice of the word variable:
sqr=:*:
The square function in J
(sqr 0)=(0+0)
(sqr 2)=(2+2)
(sqr 0)=(0*0)
(sqr 2)=(2*2)
(sqr 3)=(3*3)
Each of these sentences express a “true” relation in the sense that each
comparison yields 1. However, the first pair are true only for the specific
arguments 0 and 2, and for no other. The last three suggest (correctly) that the
indicated relation remains true for any argument, or, as we say, the argument is
allowed to vary. This generality is commonly indicated by using a pronoun
argument, or, as stated in MN, a variable:
(sqr x)=(x*x)
In MN, the term operator (or functional) is used for both of the cases
distinguished in J by adverb and conjunction. Moreover, in MN the term operator
is also commonly used to refer to functions.
The terms list, table, and report are used in J with meanings familiar to anyone,
whereas the corresponding terms vector, matrix, and array might be known only
to specialists. The familiar use of vector is as a carrier, as in disease vector. It
might be thought that a vector “carries” its items, but the actual etymology of the
term in math is quite different, although not as bizarre as that of matrix.
New terminology should be approached by using dictionaries to learn the
etymology of terms, both old and new. For example, a verb is defined as a word
that (amongst other things) expresses an action; the corresponding word function
comes from a root meaning “to perform”.
Attention to etymology is also rewarding in every-day work. For example, the
meaning of atom appears clearly in its derivation (a[not] + tem[cut]) as something
that could not be cut.
The American Heritage Dictionary [6] presents etymology in a particularly
revealing manner: all words derived from a given root are listed together in an
appendix. This highlights surprising and insightful relations, such as that between
tree and true. As a further example, the root tem that occurs in atom also occurs in
anatomy, microtome, and tome. Incidentally, tome does not necessarily mean a
big book, but rather one of the volumes “cut” from a book, such as the 24 tomes
of the original Oxford English Dictionary.
Lewis Thomas, a noted bio-chemist, explored the pleasure and profit of
etymology in his delightful book et cetera, et cetera. [7]. It is well worth reading.
Chapter 6 Terminology
35
Exercises
1. Speculate on the possible relation between the similar-sounding words tree
and true. Then look them up in AHD [6] , and consult their common Indo-
European root in the appendix.
2. Read the entries in the Indo-European sub-dictionary of AHD for the roots ag,
ak, ar, and gene, and look up some of the words derived from them.
37
Chapter
7
Decimal and Other Number Systems
Sixty-four I hear you cry!
Ask a silly question and
get a silly answer!
Tom Lehrer
A. Introduction
To most people, the decimal representation is so familiar, and so closely
identified with “the number itself”, that it may be difficult to grasp the notion of
representation. For example, what is one to make of the assertion:
The decimal representation of 365 is 365 ?
We will use lists to clarify the discussion:
The decimal representation of 365 is 3 6 5
The octal (base-8) representation of 365 is 5 5 5
The base-value function
bv=:#.
10 bv 3 6 5
365
8 bv 5 5 5
365
The main idea of a base or radix representation is embodied in the function #.
which we will now re-express in terms of more familiar functions. Familiarity
with decimals should make it clear that the representation 3 6 5 is to be
evaluated by multiplying the first item by 100, the second by 10, and the third by
1, and summing the products. Thus:
r=:3 6 5
w=:100 10 1
r*w
300 60 5
+/r*w
365
The weights w would not be appropriate for a list of other than three items, and
the following suggests a more general expression:
38
Exploring Math
y=:1 9 9 6
base=:10
#y
4
i.-#y
3 2 1 0
base^i.-#y
1000 100 10 1
+/y*base^i.-#y
1996
z=: 3 7 1 4
+/z*8^i.-#z
1996
This is the reversal of the list i.#y
BV=:+/@:(] * [ ^ i.@:-@:#@])
10 BV 1 9 9 6
1996
Equivalent to bv=:#.
8 BV 3 7 1 4
1996
We may also define and explore specific cases of the base-value function by
combining it with various left arguments:
bv10=:10&#.
bv8=:8&#.
bv2=:2&#.
bv8 z
1996
bv2 1 0 1
5
What function will yield the representation of a given argument? In other words,
what are the functions inverse to the functions b10, b8, and b2? The adverb ^:_1
gives the inverse of a function to which it is applied. Thus:
inv=:^:_1
sqrt=:%:
sqr=:sqrt inv
sqrt i.6
0 1 1.41421 1.73205 2 2.23607
sqr sqrt i.6
0 1 2 3 4 5
bv8i=:bv8 inv
bv8i 365 1996
0 5 5 5
3 7 1 4
bv2 inv 365 1996
0 0 1 0 1 1 0 1 1 0 1
1 1 1 1 1 0 0 1 1 0 0
Chapter 7 Decimal and Other Number Systems
39
2 #. bv2 inv 365 1996
365 1996
We learn to add decimal numbers by adding the items of their representations,
and performing “carries” as required. What would the result mean if we did not
perform the carries? For example:
bv10i=:bv10 inv
]d10=:bv10i 365 1996
0 3 6 5
1 9 9 6
s10=:+/d10
s10
1 12 15 11
10#.s10
2361
365+1996
2361
d8=:bv8i 365 1996
d8
0 5 5 5
3 7 1 4
8#.+/d8
2361
It appears that the sum +/d10 does indeed represent the correct sum in base-10.
Why then do we normally perform the carries?
We could perform successive carries on the sum s10 as follows:
1 12 15 11
1 12 16 1
1 13 6 1
2 3 6 1
We first verify that 2 3 6 1 represents the correct sum:
d=:2 3 6 1
(10#.d),(10#.s10),(365+1996)
2361 2361 2361
The reason that the representation d is preferred is that its items can be simply
written side-by-side to give the normal decimal form, whereas the items of s10
would give the quite different result 1121511.
Similar remarks apply to bases other than 10.
Exercises
1. Perform the carries on the base-8 sum +/d8 (that is, 3 12 6 9)
2. Enter x=:?.4#1000 to obtain four random integers less than 1000. Then
obtain their base-10 represetations, sum them, and perform the carries
40
Exploring Math
necessary to obtain a normalized representation. Verify the correctness of the
final results.
3. Repeat Exercise 2 for bases other than 10.
4. The method for adding multi-digit decimal numbers commonly taught
requires a sequence of carries interleaved with the additions, whereas the
method used here first performs all additions, and then performs the carries.
Which is the least error-prone? Which is the easier to re-check by repeating
all or part of the process?
5. Give a clear statement (in English) of the “carrying” or “normalization”
process commonly taught. Include the case of bases other than 10, as well as
the case where a carry occurs from the leading position (thus increasing the
number of items in the list).
As suggested in the last exercise, the hand procedure for normalization can be
precisely prescribed in English. Can it also be defined as a (computer-executable)
function in J? We begin with a process on a specific argument:
Initialize the result as an empty list
Current item is last item of argument
Prefix remainder to the result list
Compute the carry to the next position
Truncate by dropping the treated item
Add carry to last item
y=:3 4 25
r=.i.0
ci=._1{.y
r=.(10|ci),r
c=.<.ci%10
y=.}:y
ci=.c+_1{.y
r=.(10|ci),r
c=.<.ci%10
y=.}:y
ci=.c+_1{.y
r=.(10|ci),r
c=.<.ci%10
y=.}:y
r
3 6 5
(10#.r),(10#.3 4 25)
365 365
The last two groups of four steps are identical, a uniformity that was achieved by
truncating the argument each time. Complete uniformity would allow the entire
process to be stated more compactly (and more generally) as a repetition or
iteration of a fixed procedure defined by the four steps. It remains to make the
first block uniform: initialize the carry to zero, and replace the first line of the
block as follows:
r=.i.c=.0
ci=.c+_1{.y
The foregoing process may now be defined as an iteration as follows:
Chapter 7 Decimal and Other Number Systems
41
NORM=: 3 : 0
r=.i.c=.0
label_loop.
if. 0<#y. do.
ci=.c+_1{.y.
r=.(10|ci),r
c=.<.ci%10
y.=.}:y.
goto_loop.
end.
r
)
NORM 3 4 25
3 6 5
In the foregoing definition:
• The right argument is denoted by y.
• The block to be iterated is delimited by do. and end.
• Repetition of the block is determined by if. followed by a condition
• The result of the function is the result of the last sentence (that is, r)
• The entire definition is terminated by a right parenthesis alone on a line
A function that works correctly on the argument that guided its definition may not
work in general, and should be thoroughly tested. For example:
NORM 10 11 12
1 2 2
(10#.NORM 10 11 12),(10#.10 11 12)
122 1122
The discrepancy clearly occurs because the carry computed in the final iteration is
not zero, and must not be ignored. To rectify this, we make the condition for
repetition depend upon a non-zero carry as well as upon a non-empty argument:
NORM=: 3 : 0
r=.i.c=.0
label_loop.
if. (c~:0)+.(0<#y.) do.
ci=.c+_1{.y.
r=.(10|ci),r
c=.<.ci%10
y.=.}:y.
goto_loop.
end.
r
)
NORM 10 11 12
1 1 2 2
42
Exploring Math
NORM 1234 5 6
1 2 3 4 5 6
The function may now be generalized to a dyadic definition in which the first
argument specifies the base used: each occurrence of 10 is replaced by x., and the
line NORM=: 3 : 0 is replaced by NORM=: 4 : 0 :
NORM=: 4 : 0
r=.i.c=.0
label_loop.
if. (c~:0)+.(0<#y.) do.
ci=.c+_1{.y.
r=.(x.|ci),r
c=.<.ci%x.
y.=.}:y.
goto_loop.
end.
r
)
(8 NORM 5 3 21);(10 NORM 10 11 12)
+-----+-------+
|5 5 5|1 1 2 2|
+-----+-------+
Finally, it will be convenient to define a function whose dyadic case is NORM and
whose monadic case is 10&NORM. Thus:
N=:(10&NORM) : NORM
(8 N 5 3 21);(N 10 11 12)
+-----+-------+
|5 5 5|1 1 2 2|
+-----+-------+
Exercises
Although the formal definition of the process carried out by N is rather
6.
involved, the hand-calculation of it is quick and trivial. Confirm this by
performing it on various lists, checking the accuracy of your work by applying the
function 10&#. to each list and its normalized form.
7. The copula =. used in the definition of NORM differs from the =: used
elsewhere. Its use localizes the assigned name so that it bears no relation to
the same name used outside the definition. Experiment with the distinction by
defining a function GNORM that is identical to NORM except for the use of global
assignment (=:) and compare the behaviour of the two functions. A name can
be erased by using 4!:55, as in 4!:55 <'c' .
B. Addition
In the example d10=: bv10i 365 1966 we have already seen how the decimal
representations of two numbers may be added to obtain a representation of the
sum; we may now obtain a standard representation by applying the function N.
Thus:
Chapter 7 Decimal and Other Number Systems
43
d10=: bv10i 365 1966
d10
0 3 6 5
1 9 9 6
+/d10
1 12 15 11
N +/d10
2 3 6 1
Exercises
8. Use bv10i to compute the table of decimal representations of the list of
numbers a,b,c, where a=:365 and b=:1996 and c=:29. From this table
compute the standard representation of the sum a+b+c.
9. Use ar=:bv10i a and br=:bv10i b and cr=:bv10i c to obtain the decimal
representations of the numbers of Exercise 1, and use them in expressions to
obtain the standard decimal representation of the sum b.
In the table produced in Ex. 8, each of the shorter lists (that is, 3 6 5 and 2 9) are
padded with zeroes on the left, a change that does not change the values of the
numbers they represent. In Ex. 9 the representations are not so padded, and the
lists of differing lengths cannot be added directly. They may be added as
illustrated below:
ar;br;cr
+-----+-------+---+
|3 6 5|1 9 9 6|2 9|
+-----+-------+---+
bv10&> ar;br;cr
365 1996 29
bv10i bv10&> ar;br;cr
0 3 6 5
1 9 9 6
0 0 2 9
N +/ bv10i bv10&> ar;br;cr
2 3 9 0
a+b+c
2390
Padding can also be provided more directly, using the fact that the simple opening
of a boxed list pads it, albeit on the wrong side:
>ar;br;cr
3 6 5 0
1 9 9 6
2 9 0 0
pad=:|."1@:(|.&>)
pad ar;br;cr
0 3 6 5
44
Exploring Math
1 9 9 6
0 0 2 9
C. Multiplication
The commonly-taught methods for addition and multiplication both interleave
carries with other computations: in multiplication, each item of the multiplier is
applied to the multiplicand and the carries are propagated to give a list of results
which are then added to lists for the other items of the multiplier, producing a
further sequence of carries. However, as in addition, the carries can all be
segregated in a final normalization. For example:
a=:365 [ b=:1996
ar=:bv10i a [ br=:bv10i b
t=:ar*/br
t
3 27 27 18
6 54 54 36
5 45 45 30
This table of products may now be summed to collect those corresponding to the
same powers of ten, that is, diagonally as follows:
s=:3,(27+6),(27+54+5),(18+54+45),(36+45),30
s
3 33 86 117 81 30
(10#.s),(a*b)
728540 728540
This may also be expressed by using the oblique adverb /., which applies its
function argument to each of the diagonals. Thus:
]s=:+//.t
3 33 86 117 81 30
Exercises
10. Carry out by hand the process defined by +//.ar*/br for various values of ar
and br, and test the correctness of the resulting products.
11. Experiment with the expression </.ar*/br to get a clear view of the
behaviour of the oblique adverb.
12. Define and test a function TIMES such that ar TIMES br gives the standard
decimal representation of the product of numbers whose decimal
representations are ar and br.
Chapter 7 Decimal and Other Number Systems
45
A clearer view of the justification for the diagonal sums used in the expression
+//.t can be obtained by producing a table of powers of ten which multiplied by
t gives products weighted by the appropriate powers of ten:
a=:365 [ b=:1996
ar=:bv10i a [ br=:bv10i b
t=:ar*/br
ea=:i.-#ar [ eb=:i.-#br
exp=:ea +/ eb
wp=:10^exp
wpt=:t*wp
wpt
300000 270000 27000 1800
60000 54000 5400 360
5000 4500 450 30
+/+/wpt
728540
a*b
728540
TIMES=:N@(+//.@(*/))
ar TIMES br
7 2 8 5 4 0
(10#.ar TIMES br),(a*b)
728540 728540
Exercises
13. Perform hand-calculations of products using the process defined by the
function TIMES, and compare its use with the commonly-taught process.
Which requires the most writing? Which is the more error-prone? Which is
the easier to re-check by re-calculation of parts of the process?
D. Subtraction
Subtraction leads to the question of representing negative arguments. We will use
lists of negative numbers, with the standard form limited (as it is for positive
arguments) to numbers whose magnitudes are less than the base. For example:
10#. _3 _6 _5
_365
10#._3 _4 _25
_365
The function bv10i=:10&#.^:_1 can be used to obtain the representation of a
negative number by applying it to the magnitude, and then multiplying the
resulting list by _1. Thus:
c=:_365
ar=: _1 * bv10i@| a
46
Exploring Math
cr
_3 _6 _5
A corresponding function for either positive or negative arguments can be
obtained by multiplying not by _1, but by the signum of the argument:
* 365 0 _365
1 0 _1
REP10=: * * 10&#.^:_1@|
REP10 _365
_3 _6 _5
REP10 365
3 6 5
With this representation of negative numbers, expressions for addition apply
equally for subtraction. For example:
a=:365
b=:1996
t=:REP10 a,b
t
0 3 6 5
1 9 9 6
-/t
_1 _6 _3 _1
(10#.-/t),(a-b)
_1631 _1631
The normalization function must be modified in the same manner:
NOR=: *@#. * NORM&|
N=:10&NOR : NOR
N 3 4 25
3 6 5
N _3 _4 _25
_3 _6 _5
Exercises
14. Read Chapter 4 of Book 2 (Arithmetic), and try some of its Exercises. Note
particularly the section on Mixed Bases.
47
Chapter
8
Recursion
re-, back + currere, to run
AHD[5]
The factorial function ! introduced in Chapter 2 was seen to be a product of the
first positive integers. Thus:
!n=:4
24
(4*!3),(4*3*!2),(4*3*2*!1),(4*3*2*1)
24 24 24 24
It would therefore appear that !n might be defined simply as n*!n-1. Such a
definition is said to be recursive, because the function being defined recurs in its
own definition. But a sequence of the form:
f n
n*f n-1
n*(n-1)*f n-2
would continue forever (through n=:0 and n=:_1 etc.), and it is clear that two
further pieces of information are required: when to stop the process, and the value
of the function for the argument at the stopping point. For the present case of the
factorial, the stopping condition could be that the argument be 1, and the stopping
value could be given by the identity function ]. The three required functions are:
p=:]*f@:<:
q=:]
r=:1&=
The complete definition may now be expressed and used as follows:
f=:p`q@.r
f 4
24
f"0 (1 2 3 4 5)
1 2 6 24 120
48
Exploring Math
In the definition of f, the conjunction ` ties the functions p and q to form a
gerund, from which the agenda conjunction selects one for execution according
to the index (0 or 1) provided by its right argument function r. Once f is defined
as above, we can experiment with p and the other functions to see some of the
workings of the definition of f:
Display the definition of p
p
] * f@:<:
p 4
24
r 4
0
r 4 3 2 1
0 0 0 1
q 1
1
Exercises
1. Compare the results of f"0(4 3 2 1 0) and !4 3 2 1 0 and redefine f so
that it agrees with ! for the argument 0.
The problem of Exercise 1 could be solved by redefining q and r as follows:
q=:>:
r=:0&=
f 0
1
However, it would seem more straightforward to define q as the constant 1 as
follows:
q=:1
f 0
│domain error
│ f 0
A problem arises because 1 is a noun, not a function, and the arguments in the
gerund p`q must both be functions. We therefore need a function that returns the
constant value 1 when applied to any argument. Such constant functions are
commonly needed, and are produced by the rank conjunction ("), used in Chapter
5 to modify a function, as in <"2. Thus:
1"0 x=:i.4
1 1 1 1
1"_ x
1
Rank 0 produces a result for each atom of x
Infinite rank gives a single result for any argument
x"1 'Now is the time'
Chapter 8 Recursion
49
0 1 2 3
The function q may therefore be redefined as follows:
q=:1"_
f"0 (4 3 2 1 0)
24 6 2 1 1
Finally, f (of rank 0) may be redefined compactly as follows:
f=:(]*f@:<:)`(1"_)@.(0&=)"0
f 4 3 2 1 0
24 6 2 1 1
As a second example of recursive definition we will define the sum of the first n
odd numbers, first met in Chapter 1:
sod=:0"_`(>:@+:@<: + sod@<:)@.*
sod 4
16
sod"0 i.6
0 1 4 9 16 25
The definition of sod may be interpreted as follows: When the argument n is 0,
then the signum on the right returns 0, choosing the leading function in the
gerund, giving a result of 0; otherwise, the result is the nth odd number (that is,
>:@:+:@:<:) plus the sum for an argument n-1 (that is, sod@<:).
Exercises
2. For convenience, certain constant functions are provided directly, without the
need for the rank operator. Experiment with the constant functions _9: and
_8: and so on through 9:. Use 1: and 0: to simplify the definitions of f and
sod above.
3. Because increment (>:) is the inverse of decrement (<:), the expression
>:@+:@<: is of the form gi@f@g, where gi is the inverse of g. We say that
this is a case of applying f under g, and denote it by f&.g. Use this fact to
simplify the definition of sod, and check the resulting behaviour.
Recursive definition essentially specifies a function in terms of the same function
applied to a simpler case, and its use can enormously simplify many definitions.
For example, the Tower of Hanoi puzzle is stated as follows:
A set of n drilled discs of different diameters stacked as a pyramid on a peg A
is to be moved one at a time to a peg C without ever placing a larger on a
smaller. A third peg B may be used as intermediary.
The process for two discs may be expressed by the table:
50
Exploring Math
AB
AC
BC
which is to be interpreted row-by-row as follows:
Move (the top) disc from A to B
Move from A to C
Move from B to C
The case of n discs can be expressed in terms of the case of one fewer as follows:
Move n-1 discs to the intermediary peg B, then move the remaining largest disc to
C, and finally move the n-1 discs from B to C.We will use this fact to make a
recursive definition as follows:
H=:m`b@.(1&=@[)
m=:(<:@[ H 1: A. ]) , b@] , <:@[ H 2: A. ]
b=:,:@(0 2&{)@]
p=:'ABC' Pegs
1 H p
AC
2 H p
AB
AC
BC
|: 3 H p Transposed table
AACABBA
CBBCACC
Exercises
4. Use discs and pegs (or numbered cards and labelled positions on a table) to
carry out the instructions in the foregoing tables to verify that they provide
proper solutions to the Hanoi puzzle. Also enter the expression |: 3 H p and
test it as well.
5. Give an expression for the number of moves required for n discs.
6. Explain the behaviour of the definition of H, using experiments to show the
permutation provided by the function A., the selection provided by the
indexing function {, and the purpose of the monadic function ,:. Also
redefine the main function m, using indexing to perform the necessary
permutations.
7. Experiment with the function HV=: |:@H.
8. Read the definition of agenda in [5], and experiment with the use of $: for
self-reference in recursive definitions.
9. Compare the following recursively-defined function n with the first definition
of NORM in the preceding chapter:
f=:(0:,10&|) + <.@(%&10) , 0:
g=:+/@(*./\@(0&=)) }. ]
h=:*./@(10&>)
n=:n@f`g@.h
Chapter 8 Recursion
51
53
Chapter
9
Proofs
Drug thy memories, lest thou learn it,
lest thy heart be put to proof
Tennyson
A. Introduction
It is probably advisable to begin by reviewing the brief discussion of proofs at the
end of Chapter 2.
The final experiment of Chapter 1 showed a relation between the sum of the first
n odd numbers and the square of n. We will first reproduce it here:
n=:20
odds=:1+2*a=:i.n=:20
odds
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
(+/odds),(n*n)
400 400
+/\odds
1 4 9 16 25 36 49 64 81 100 121 144 169 196 225 256 289 324 361
400
(1+a)*(1+a)
1 4 9 16 25 36 49 64 81 100 121 144 169 196 225 256 289 324 361
400
But is the indicated relation true for any positive integer n? If you are already
convinced that it is, any proof may seem pointless. However, you might still ask
why it is true. The following should be helpful in answering this:
q=:1+2*i.n=:6
r=:|.q
q,:r
1 3 5 7 9 11
11 9 7 5 3 1
First six odd numbers
Odds in reverse order
54
Exploring Math
(+/q);(+/r);(q+r);(2%~q+r);(+/2%~q+r)
+--+--+-----------------+-----------+--+
|36|36|12 12 12 12 12 12|6 6 6 6 6 6|36|
+--+--+-----------------+-----------+--+
The foregoing shows the rather obvious fact that sums over a list, over the
reversed list, and over one-half of the sum of the lists all agree. But the half-sum
of the lists has a pattern whose sum is easily expressed as a product:
(2%~q+r);(n#n);(+/n#n);(n*n)
+-----------+-----------+--+--+
|6 6 6 6 6 6|6 6 6 6 6 6|36|36|
+-----------+-----------+--+--+
The last agreement (between +/n#n and n*n) is based on the fact that
multiplication is defined as repeated addition.
The foregoing attempted to show why two results were equal by exhibiting their
equivalence to other results, where the equivalence was already known or
obvious. This is perhaps the only way to answer the question why. However, the
equivalences assumed may be made clearer by laying out the steps of the
argument as a proof, that is, as a succession of equivalent statements annotated by
the justification of the equivalence of each to the one preceding it. Thus:
+/q=:1+2*i.n
+/|.q
2%~(+/q)+(+/|.q)
2%~+/(q+|.q)
+/2%~(q+|.q)
+/(n#n)
n*n
Summation is symmetric (unaffected by ordering)
Half sum of equals is an identity
Summation is symmetric
Summation distributes over division
The definition of multiplication
Such a list of supposedly equivalent sentences can be tested (for careless errors)
by assigning a suitable value to the argument n, entering them on the computer,
and comparing the results.
This putative proof has not proved anything but it has, as Lakatos would say,
broken the original conjecture into a collection of sub-conjectures, each of which
may be profitably examined. Consider the first assertion that summation is
symmetric, and gives the same result when applied to any permutation of a list.
This may be tested as follows:
q=:1+2*i.n=:6
117 A. q
1 11 9 5 7 3
_1 A. q
11 9 7 5 3 1
(+/q),(+/117 A. q),(+/_1 A. q)
36 36 36
But why is summation symmetric? We may, for example, ask whether the notion
applies to other functions, as in product over (*/), maximum over (>./), and
Chapter 9 Proofs
55
subtraction over
(-/), beginning with the following tests:
r=:|.q
(+/q),(*/q),(>./q),(-/q)
36 10395 11 _6
(+/r),(*/r),(>./r),(-/r)
36 10395 11 6
What is it about the functions +, *, and >. that make +/, */, and >./ symmetric?
The answer is that they are both associative and commutative. These matters are
examined further in Exercises, but the main point is that any conjecture may lead
to further sub-conjectures that can be identified and pursued until the reader
reaches assertions that are satisfying to him. As Lakatos shows, assertions
satisfactory for one reader (or purpose) may not be satisfactory for another.
Exercises
1. Addition is said to be associative because a sequence of additions can be
associated by parenthesizing them in any way without changing the result. For
example, +/1 2 3 4 and (1+(2+(3+4))) and ((1+2)+(3+4)) and
(1+(2+3)+4) are all equal. Test the associativity of addition by entering a
variety of equivalent expressions.
2. Repeat Exercise 1 for product and maximum.
3. The completely parenthesized form of +/q is 1+(3+(5+(7+(9+11)))), and
the corresponding form of +/117 A. q) is 1+(11+(9+(5+(7+3)))). Write a
sequence of sentences [such as 1+(3+(5+(7+(11+9))))]that uses only
associativity and commutativity to move from the first expression to the last,
and enter them all to test their equivalence.
4. Use the words Comm and Assoc to annotate your solution to Exercise 3 to
provide a formal proof of the equivalence of +/q and +/117 A. q.
5. The proof that +/q is equivalent to n*n is completely formal except for one
omission. Complete it.
Following Lakatos’s point that a formal or informal proof may suggest further
lines of inquiry, we note that the list sum q+|.q gave items with a common value.
This is, of course, a proposition that is not true for every list q, but depends upon
some property of q. What is that property?
The point is that q is an arithmetic progression; successive items increase by the
addition of a fixed constant (in this case 2). The sum of the first and last items
therefore equals the sum of the item just following the first and just preceding the
last, and so on for further pairs. This is more easily stated (and seen) by reversing
the list to bring corresponding pairs together. Thus:
q,:|.q
1 3 5 7 9 11
11 9 7 5 3 1
56
Exploring Math
+/q,:|.q
12 12 12 12 12 12
The method of proof can therefore be applied to find expressions equal to the sum
of any geometric progression. For example:
g=:i.n=:6
g,:|.g
0 1 2 3 4 5
5 4 3 2 1 0
+/g,:|.g
5 5 5 5 5 5
(n*n-1)%2
15
+/g
15
b=:4 [ s=:3 [ n=:7
h=:b+s*i.n
h
4 7 10 13 16 19 22
+/(h,:|.h)%2
13 13 13 13 13 13 13
b+(s*n-1)%2
13
n*b+(s*n-1)%2
91
+/h
91
An AP beginning at b with steps of size s
Exercises
6. Write formal proofs for each of the foregoing results.
7. Define a function f such that f b,s,n gives the mean of the arithmetic
progression beginning at b and continuing with increments s for a total of n
items.
B. Inductive Proof
An inductive proof of the equivalence of two functions proceeds by first assuming
that they are equal for some unstated value of the integer argument n, and using
that assumption (called the induction hypothesis) to prove that they are therefore
equal for the next argument n+1. It is then shown that they are indeed equal for
some specific argument n=:k. It therefore follows that they are equal for all
values k, k+1, k+2, and so on without limit. For example:
Chapter 9 Proofs
57
ssq=:+/@*:@i.@>:"0
ssq 5
55
Sum of squares of first 6 non-negative integers
ssq i.6
0 1 5 14 30 55
Using rational constants (such as 2r6 for 2%6), we then define a putative
equivalent function g, adopt the induction hypothesis that f n is equal to g n, and
use it to prove that f n+1 equals g n+1:
g=:(1r6&*)+(3r6&*@(^&2))+(2r6&*@(^&3))
ssq n+1
+/ *: i. >: n+1
(+/*:i.>:n)+(*:n+1)
(ssq n)+(*:n+1)
(g n)+(*:n+1)
Definition of ssq
(Sum of first terms) plus last term
Definition of ssq
Induction hypothesis
Definition of g
(1r6*n)+(3r6*n^2)+(2r6*n^3)+(*:n+1)
(1r6*n)+(3r6*n^2)+(2r6*n^3)+1+(2*n)+(n^2)
(1r6*n+1)+(3r6*(n+1)^2)+(2r6*(n+1)^3)
g n+1
Definition of g
The lines of the foregoing proof that are not annotated concern the use of
manipulations from elementary algebra, including the expansion of the square and
the cube of the sum n+1. The inductive proof may now be completed by showing
that the functions are equal for the argument 0.
Exercises
8. Enter n=:6, and then enter the lines of the foregoing proof to verify that they
each give the same result. It is advisable to enter such a sequence in a “text”
or “script” file, then execute it, observe the result, and return to the script file
to correct any errors and re-try. To open the script file, hold down the control
key and press n; to execute it, hold down both the control and shift and press
w; to see the result, switch to the execute window by holding down control
and pressing the tab key; return to the script window by the same action.
9. Define the function s=: +/@:i.@>: and an equivalent function t that does
not use summation. Give an inductive proof that they are equivalent.
A recursive definition of a function f provides a clear statement of the value of
f n+1 in terms of the value of f n; this fact is obviously valuable in the
construction of an inductive proof.
But how does one find a function such as g? This matter will be treated in Chapter
14. But for present use in further experiments with inductive proofs, we provide
the following methods.
The function g is an example of a polynomial, a sum of weighted powers of the
argument, the weights being 0 1r6 3r6 2r6. They may be obtained as follows:
]w=: (ssq a) %. a ^/ a=: i.5
58
Exploring Math
_2.99066e_14 0.1666667 0.5 0.3333333 _6.50591e_14
6*w
_1.7944e_13 1 3 2 _3.90354e_13
Because %. (matrix divide) produces its results by approximation, the extreme
items of 6*w are not quite zero. They can be “zeroed” by the following function,
in which the first argument specifies the tolerance in number of decimal digits:
ZERO=: ] * |@] > 10&^@-@[
8 ZERO 6*w
0 1 3 2 0
14 ZERO 6*w
_1.7944e_13 1 3 2 _3.91687e_13
For convenience in experimenting with a variety of functions, we will adopt from
Section F of Chapter 14 the conjunction FIT, so defined that n FIT f x gives the
n-item list of coefficients of a polynomial that best fits the function f at the points
x. For example:
V=:] ^/ i.@[
FIT=:2 : 'y. %. (x. & V)'
3 FIT ^
^ %. 3&V
]c=:3 FIT ^ b=:0.2*i.5
1.00238 0.9203119 0.7569838
c p. b
1.00238 1.21672 1.49162 1.82708 2.2231
^ b
1 1.2214 1.49182 1.82212 2.22554
As remarked, g is an example of a polynomial, and the coefficients produced by
FIT can (preferably after being zeroed) be used with the polynomial function p.
to produce an equivalent function. Thus:
]c=: 8 ZERO 4 FIT ssq a=:i.5
0 0.1666667 0.5 0.3333333 0
c p. i.8
0 1 5 14 30 55 91 140
g i. 8
0 1 5 14 30 55 91 140
Exercises
Chapter 9 Proofs
59
10. Study the discussion of proofs in Section D of Chapter 5 of Book 2.
11. Find a function equivalent to the sum of cubes, and construct an inductive
proof of the equivalence.
[]c=: 8 ZERO 5 FIT scubes x=:i.6]
12. For many functions, the coefficients for an equivalent or approximate
polynomial may be conveniently obtained by using the Taylor adverb t., as in
f t. i.6. Experiment with this for the functions:
]
(^&4-^&2)
^&4
(>:^4:)
(<:^4:) ^
61
Chapter
10
Tools
Without tools he is nothing,
with tools he is all
Carlyle
A. Introduction
This chapter concerns tools for exploration. They are fully treated in Burke’s J
User Manual (available on-line under the help menu in the J system), but should
themselves be explored in the manner used for math in preceding chapters.
For example, an overall view of the tools available may be obtained by dropping
the menus. This can be done by clicking the mouse on each of them, but they can
also be dropped by first pressing the alt key, then the down arrow, then the left or
right arrow to move over the menus. The alt key will roll up a menu.
With a menu dropped, use the up and down arrows to select an item, and press
enter to execute it. Alternatively, an underscored letter in an item can be entered
to execute it. Some menu items can be invoked directly (without dropping the
menu) by pressing a key (usually while holding down the control key), as
indicated to the right of the item’s name.
For example, (as shown in the help menu), the F1 key may be pressed to display
the J vocabulary, and any entry in the vocabulary may be chosen for display by
double-clicking on it with the mouse. A definition is then displayed, and may also
be printed by using Print topic in the file menu.
Exercises
1. Using items from the help menu, display and read various pages from the User
Manual, including Chapter 1.
2. Display and read a few sections from the introduction to the J dictionary
3. Read the section on grammar in the J dictionary.
62
Exploring Math
B. Editing
As remarked in Chapter 1, a previously entered line can be brought to the input
area for editing and re-entry by moving the cursor up to it and pressing enter.
Moreover, a line containing any phrase can be found by pressing Control f to
highlight the search entry box, entering the phrase in it, and pressing enter.
Repeated searches on the same phrase will find successive occurrences of it.
Pressing Control d drops a menu of previous entries; one may be selected for use
by pressing the up arrow.
C. Script Windows
Enter Control n to open a script window, enter one or more J sentences in it, and
press Control-Shift w to execute the sentences. The execution occurs in the
execution window, and can be viewed by entering Control Tab to switch back to
it.
A window may be saved as a file (under the name shown on the window) by
pressing Control s, and can be re-opened at any time by pressing Control o. It can
also be saved under any chosen name by using Save As or Save Copy As from the
file menu.
Select the item Session Manager from the User Manual, and from it select the
item Script Windows. Read the discussion of their use.
Exercises
63
Chapter
11
Coordinates and Visualization
It was their belief that, if they stared long enough at these
mystic curves and angles, red ink would turn into black.
Alva Johnson
A. Introduction
Take a sheet of graph or squared paper (ruled with equidistant vertical and
horizontal lines), choose some point of intersection as the origin to be labelled
0 0, and label vertical lines from left to right and horizontal lines from bottom to
top with symmetric integers as follows:
i: 9
_9 _8 _7 _6 _5 _4 _3 _2 _1 0 1 2 3 4 5 6 7 8 9
Any point of intersection may then be labelled by two coordinates, the first (or x)
coordinate specifying the vertical line through it, and the second (or y) coordinate
the horizontal. Such a coordinate system makes it possible to describe geometric
figures, and leads to analytic or coordinate geometry. For example:
p=:3 4
q=:9 4
r=:6 8
s=:9 7
t=:8 6.
is=:p,q,:r
rt=:p,q,:s
qd=:p,q,s,:r
pg=:p,q,s,r,:t
A single point
Isosceles triangle
Right (-angled) triangle
Quadrilateral
Pentagon
Properties of the geometric figures can be obtained from their coordinate
representations. For example:
disp=:1&|. - ]
Rotate by 1 and subtract
64
Exploring Math
disp is
6 0
_3 4
_3 _4
length=:+/&.*:"1
length p
5
length disp is
6 5 5
Displacements from vertex to vertex
Length according to Pythagoras
Length or distance from origin
Lengths of sides of isosceles triangle
heron=:%:@(*/@:(semip,semip-])) Heron’s formula for area
semip=: 2:%~+/
heron length disp rt
9
Semi-perimeter
Area of the right triangle
area=:heron@:length@:disp
area rt
9
Area function using Heron
area is
12
Area of the isosceles triangle
Exercises
1. Plot the points p through t on graph paper, and join the appropriate points by
straight lines to show the figures is through pg . Then use the base and
altitude of each triangle to compute their areas, and compare with the results
of Heron’s formula.
2. Use the AHD[6] to examine the etymology of the several terms used for
figures that differ only in the number of their sides (or angles or vertices), and
suggest a compact common terminology.
[3-gon, 4-gon, and n-gon (from polygon)]
3. A vertex may be shifted to the left by subtracting a vector with a zero final
element. Plot the following triangles, and use both base-times-altitude and
Heron’s formula to compute their areas:
rts=:p,q,:r-8 0
is=:p,q,:s-8 0
Although plotting polygons by hand may be instructive, it is also convenient to
use the computer to plot them. We begin by normalizing the coordinates of a
figure:
• sliding them to bring the lowest point to 0 0
• sizing them to no more than 1 in magnitude
• doubling and subtracting 1 to bring them between _1 and 1
•
ravelling them to form a list for use by the plotting function
slide=:] -"1 <./
size=:] %"1 >./
scale=:,@(<:@+:@size@slide)
slide is
Chapter 11 Coordinates and Visualization
65
0 0
6 0
3 4
size slide is
0 0
1 0
0.5 1
<: +: size slide is
_1 _1
1 _1
0 1
scale is
_1 _1 1 _1 0 1
The following steps introduced the necessary graphing functions, and use them to
display the isosceles triangle:
load 'graph'
gdopen'a' Opens graph window labeled 'a'. Use mouse to return focus
to J
gdpolygon scale is
gdshow''
We then superpose a red right triangle and, finally, clear the window:
255 0 0 gdpolygon scale rt Colors red, green, blue; intensity 0-255
gdshow''
gclear 'a'
A graphics window may be closed by clicking the upper right corner with the
mouse.
The functions provided by the graphics file may be displayed by entering
names_z_'' . However, they should for the moment be treated as tools, whose
internal workings may be ignored provided that their effects are sufficiently
understood.
It will be found most convenient to enter a sequence of graphics commands in a
script window (opened by entering Control n), and to execute them by using the
drop-down run menu.
66
Exploring Math
To learn more about the use of graphics, use the mouse to drop the Studio menu
in J, then click on Labs, and then on Graph Utilities.
Exercises
4. Enter the foregoing sequence of graphics sentences in a script window, and
use the “Selection” option from the run menu to execute it.
5. Display each of the polygons defined in this section in various colors; in
particular, display rt in red and (without clearing the window) -:is in green.
6. Permute the coordinates of the polygons (as in 1 A. pg), and discuss the
appearance of the resulting figures.
7. Enter rot=:^@j.@rfd@[*], and rfd=:%&180p_1, and experiment with rot by
plotting the results of the following forms:
45 rot rt
45&rot&.> rt;is;rts
8. Experiment with, and comment on, the function rotate introduced by the
graphics file.
B. Visualization
The examples of Section A illustrate the fact that the coordinate representation
and the graphic representation of figures are complementary; each provides
certain advantages. For example, the graph of Exercise 6 shows how easy it is to
distinguish an “improper” polygon (in which sides cross), a matter that would not
be easy to spot in a table of coordinates.
On the other hand, for the computation of properties such as areas, coordinates
are far superior. For the particular triangles rt and is (and even for rts and iss
plotted by hand in Exercise 3) the computation of area appears simple, but this
simplicity is deceptive, as illustrated by the rotated figure of rts in Exercise 7.
Moreover, the determinant function provides an even simpler statement of area
than does Heron’s formula, and yields additional important information. Thus:
det=:-/ . *
rt,"1 (0.5)
3 4 0.5
9 4 0.5
9 7 0.5
det rt,"1 (0.5)
9
det (1 A. rt),"1 (0.5)
_9
AREA=:det@(,"1&0.5)
AREA rt
9
Exercises
Chapter 11 Coordinates and Visualization
67
9. If you are familiar with the computation of determinants from high school,
check the foregoing results by hand.
10. The result of AREA is positive if the coordinates are in counter-clockwise order
(when plotted), and are negative if clockwise. Test this for various triangles.
11. What is the significance of a zero result from AREA?
12. Enter t=:?.7 2$10 to generate a random table of seven points. Referring to
these points by the letters A through G, determine which of the last five lie on
opposite sides of the line determined by the first two.
[Enter L=:0 1 { t, and compare signs of
the areas of the triangles C,L and D,L, etc.]
13. Compute the area of the pentagon pg of Section A.
[Referring to the points by A-E, compute the three (signed)
areas A,B,:C and A,C,:D and A,D,:E and add them]
C. Plotting Multiple Figures
As illustrated by Exercise 4, different figures may be displayed together.
However, as seen from the same exercise, they are scaled independently, and
therefore do not give a satisfactory picture.
We will now rectify this by developing functions that will handle arguments of
the form rt;is;pg, and scale the whole according to the requirements of the
entire collection. It suffices to modify the functions slide, size, and scale so as
to apply to each box (that is, under (&.) open (>)), and to find the maxima and
minima after razing the argument (by applying ;). Thus:
SLIDE=:] -"1&.> <@(<./@;)
SIZE=:] %"1&.> <@(>./@;)
SCALE=:,&.>@(<:@+:&.>@SIZE@SLIDE)
We may then proceed with experiments such as the following (which plots the isosceles
triangle together with the right-triangle displaced two places up and to the right:
POLY=:gdpolygon&>
color=:0 255 0;255 0 0
gdopen''
color POLY SCALE is;2+rt
gdshow''
14. Experiment with the plotting of multiple figures, using expressions of the
form:
Exercises
(255 0 0;0 0 255) POLY SCALE rt;pg
15. Enter SCALE <rt and SCALE rt to see that only the former gives the desired
result. Define a corresponding function M that works in either case
68
Exploring Math
[M=:SCALE@ bifo=:<^:(-:>). Observe the results
of bifo (box if open) applied to rt and to <rt.]
16.Enter the definition reg=:+.@^@j.@o.@(2:*i.%]), and verify that reg 4 and
reg 6 give the coordinates of regular polygons inscribed in a unit circle. This
definitionemploys complex numbers, so do not spend time on the definition
itself at this point. Instead plot the figure reg 6 and various permutations of
it, and interpret the figures observed. Include the following sequence:
red=:<255 0 0
gdopen 'a'
red POLY SCALE < reg 6
gdshow''
gdopen 'b'
red POLY SCALE < 1 A. reg 6
gdshow''
gdopen 'c'
red POLY SCALE < /:~ reg 6
gdshow''
<@reg"0 i.6
D. Plotting Functions
This section illustrates the use of various facilities for plotting functions:
load 'plot'
plot x=:2 3 5 7 11
In this and the following plot, the horizontal axis is labeled with the default values from 0
to 4. The next plot after that uses the form x;*:x to label this axis according to the
argument x.
Chapter 11 Coordinates and Visualization
69
The alternative function PLOT=:'stick,line'&plot draws vertical "sticks" to each
point as well as the "lines" between the points. Similarly, BAR=:'stick'&plot
produces barcharts.
Enter the definitions of these functions, and experiment with them.
plot *: x NB. Plot square function
plot x;*: x NB. square Versus argument
Entering load 'graph' also makes available a function called steps that produces a
grid from one value to another in a specified number of steps. For example:
steps 2 4 10
2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4
We will give it an alternative name as follows:
grid=:steps
grid 2 4 10 NB. 2 to 4 in 10 steps
2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4
sin=:1&o.
cos=:2&o.
plot sin x=:grid 0 6 100
70
Exploring Math
plot (sin ; sin*cos) x
plot sin */~ grid 0 3 30 NB. Multiple sines
'surface'plot sin*/~grid 0 3 30
Chapter 11 Coordinates and Visualization
71
73
Chapter
12
Linear Functions
That wholly consisted
of lines like these
C.S. Calverley
A. Distributivity
The properties of commutativity and associativity introduced in Chapters 3 and 9
concerned a single function; the important property of distributivity concerns a
pair of functions. It is commonly treated as a relation between two dyadic
functions, as illustrated below:
Assign the names a and b and c
'abc'=: 3 4 5
a,b,c
3 4 5
]d=:a*(b+c)
27
]e=:(a*b)+(a*c)
27
The general equivalence of the results d and e is expressed by saying that times
distributes over addition. However, this distributivity might equally be expressed
with the sum as the left argument of times as follows:
]f=:(b+c)*a
27
]g=:(b*a)+(c*a)
27
Times also distributes over subtraction, a fact that may be illustrated as follows:
(a*(b-c));((a*b)-(a*c));((b-c)*a);((b*a)-(c*a))
+--+--+--+--+
|_3|_3|_3|_3|
+--+--+--+--+
Does division distribute over addition? It can be tested as follows:
74
Exploring Math
(a%(b+c));((a%b)+(a*c));((b+c)%a);((b%a)+(c%a))
+--------+-----+-+-+
|0.333333|15.75|3|3|
+--------+-----+-+-+
The result is conflicting; one pair agrees, and the other does not; a matter
sometimes resolved by saying that division distributes to the left, but not to the
right. It is simpler and clearer to note that the monadic function a&% does not
distribute over addition, but that the function %&a does. We will hereafter speak
only of the distributivity of monadic functions. For example, +: (double) and -:
(halve) both distribute over addition.
Exercises
1. Does %&a distribute over subtraction? Test your assertion.
2. Repeat the experiments of this section using conformable (that is, equal in
number of items) lists a, b, and c.
3. Repeat the experiments of this section using conformable tables A, B, and C.
4. Experiment with the dyadic cases of the functions f@g and g&f for various
values of the proverbs f and g (such as f=:% and g=:-), and state clearly the
effects of the conjunctions @ and &
[b f@g c is equivalent to f b g c, and
b f&g c is equivalent to(g b)f(g c)]
5. Comment on the assertion that the equivalence of f@g and g&f is a test of the
distributivity of f over g
6. Experiment with the conjunction dtest=: 12 : 'x.@y. -: y.&x.' in
testing for distributivity. Include +dtest- and b %&3 dtest + c and b 3&%
dtest + c
B. Linearity
A function that distributes over addition is said to be linear. Linear functions
prove to be important in almost every branch of applied math.
The functions L1=:*&2 and L2=:%&2 and L3=:|."1 are each linear. Thus:
a=:3 4,9 4,:9 7 [ b=:3 4,9 4,:6 8
a;b
+---+---+
|3 4|3 4|
|9 4|9 4|
|9 7|6 8|
+---+---+
L1 (a+b)
12 16
36 16
Chapter 12 Linear Functions
75
30 30
(L1 a)+(L1 b)
12 16
36 16
30 30
Such matters may be expressed more clearly and compactly as follows:
a (L1@+ ; +&L1) b
+-----+-----+
|12 16|12 16|
|36 16|36 16|
|30 30|30 30|
+-----+-----+
a (L1@+ -: +&L1) b
1
C. Linear Vector Functions
A function of rank 1 applies to each vector in its argument, and may be called a
vector function. We will use the term in a more restrictive sense: the result must
be the same shape as the argument. Thus L3=:|."1 defined in the preceding
section is a linear vector function:
d=: 4 2 1 [ e=: 2 3 5
L3 d+e
6 5 6
(L3 d)+(L3 e)
6 5 6
If f=:+/@:*"1, then the function w&f is a weighted sum, with weights specified
by the vector w. Moreover, it is linear. For example:
w=:2 0 3
w&f d
11
w*d
8 0 3
+/w*d
11
(w&f d+e),:(w&f d)+(w&f e)
30
30
Although w&f is linear, it is not a linear vector function according to our strict
definition. Such a linear vector function may be defined as follows:
x=:5 1 2
76
Exploring Math
y=:7 2 0
g=: w&f,x&f,y&f
g d
11 24 32
t=:w,x,:y
t
2 0 3
5 1 2
7 2 0
h=: t&f
h d
11 24 32
In general, if t is an n-by-n table, then t&f is a linear vector function on any
vector of n elements.
Exercises
7. Use the arguments x and y to test the assertions that each of the following
functions is linear:
x=:2 7 1 8
y=:3 1 4 2
L4=:+/\
L5=:L4*L4
L6=:L4^:_1
[L5 is not linear. L6 illustrates the fact that the inverse of a linear function is
linear. L4 gives subtotals, and L6 gives differences: try L4 L6 x and L6 L4 x
to test the assertion that they are inverse functions.]
D. Inner Product
Applied to the sum (+/) and product (*), the dot conjunction produces the matrix
product function that is (for the arguments used in the preceding section)
equivalent to the function f defined there:
The space before the dot is essential
Using w and d and t from the preceding section
mp=:+/ . *
w mp d
11
t &mp d
11 24 32
For any square matrix m (that is, =/$m), the function mp&m is a linear vector
function. For example:
m=:5-~?.4 4$10
L=:m&mp"1
x=:2 7 1 8
Chapter 12 Linear Functions
77
y=:3 1 4 2
m;(L x);(L y);(L x+y);(L x)+(L y)
+----------+------------+-------------+-------------+-------------+
|_4 2 _1 0| | | | |
|_3 _5 1 1|5 _32 18 _37|_14 _8 16 _18|_9 _40 34 _55|_9 _40 34 _55|
| 4 _2 0 3| | | | |
|_5 _5 0 1| | | | |
+----------+------------+-------------+-------------+-------------+
Conversely, for any linear function, a matrix m that represents it (in the function
L=:m&mp"1) can be obtained by simply transposing the table produced by
applying the function to the appropriate identity matrix. For example:
(];L;|:@:L;m"_) I=:=i.4
+-------+-----------+----------+----------+
|1 0 0 0|_4 _3 4 _5|_4 2 _1 0|_4 2 _1 0|
|0 1 0 0| 2 _5 _2 _5|_3 _5 1 1|_3 _5 1 1|
|0 0 1 0|_1 1 0 0| 4 _2 0 3| 4 _2 0 3|
|0 0 0 1| 0 1 3 1|_5 _5 0 1|_5 _5 0 1|
+-------+-----------+----------+----------+
(|:L I)&mp x
5 _32 18 _37
L x
5 _32 18 _37
Exercises
8. Using the result of L4 x from Exercise 7, try to determine by hand the value
of the matrix m such that m&mp x gives the same result. Compare your result
with
|:@:L4 I=:=i.4.
9. Compare the results of the function m&mp derived in Exercise 8 with the result
of L4 when applied to the argument z=:2 7 1 8 2 8.
10. Repeat Exercise 8 for the function L6.
11. Repeat Exercise 8 for the function L3=:|."1.
The error produced in Exercise 9 illustrates the fact that the domain of the matrix
product representation of a linear function is restricted to arguments of a specific
number of items, even though the linear function from which it is derived has a
wider domain.
E. Why The Name “Linear”?
Why is a function that distributes over addition called linear? We will attempt to
answer this by applying an arbitrary linear vector function to geometric figures,
beginning with the right-angled and isosceles triangles of Chapter 11:
is=:3 4,9 4,:6 8
rt=:3 4,9 4,:9 7
]m=:?.2 2$10
1 7
4 5
78
Exploring Math
mp=:+/ . *
L=:m&mp"1
rt;L rt
+---+-----+
|3 4|31 32|
|9 4|37 56|
|9 7|58 71|
+---+-----+
is;L is
+---+-----+
|3 4|31 32|
|9 4|37 56|
|6 8|62 64|
+---+-----+
We may plot these resulting triangles (by hand or by the methods of Chapter 11)
to try to assess the effects of the linear function. Is the right-angle of rt retained?
Do the two equal sides of is remain equal? Is the order of the vertices reversed?
We may also apply the function AREA of Chapter 11 to compare the areas:
AREA=:det@(,"1&0.5)
det=:-/ . *
(AREA L rt)%(AREA rt)
_23
(AREA L is)%(AREA is)
_23
The areas of the two triangles appear to be multiplied by the same factor. In fact,
the area transformation produced by a function m&mp is the determinant of m :
det=:-/ . *
det m
_23
We now consider three points on a line, that is, a degenerate triangle having zero
area:
a=:3 4
b=:5 13
]deg=: a,b,:(a%4)+(3*b%4)
3 4
5 13
4.5 10.75
AREA deg
0
AREA L deg
0
Chapter 12 Linear Functions
79
This result suggests (correctly) that a linear function transforms a line into a line,
a fact that suggests the use of the term linear for it.
A point in three-dimensional space can be represented by a three-element vector
such as p=:3 1 5. A linear function on such a point must, of course, be
represented by a 3-by-3 matrix m. Moreover, a tetrahedron may be represented by
a 4-by-3 table, and the function AREA may be modified to give its volume as
follows:
VOL=:det@(,"1&1r6)
Exercises
12. Use a tetrahedron (such as tet=:0 0 0,0 0 1,0 1 0,:1 0 0) whose
volume is easily computed to test the behaviour of the function VOL.
13. Use a permutation of the vertices of tet to show that VOL gives the signed
volume of its argument much as AREA does. State the condition for a non-
negative volume.
[Try VOL 1 A. tet. The volume is non-negative if the vertices of the “base”
triangle are in counter-clockwise order when viewed from the leading vertex]
14. Use expressions analogous to those used for the area of a triangle to
investigate the volume transformation effected by a linear function on a
tetrahedron.
15. Define a degenerate tetrahedron (in which the four points are co-planar) to
illustrate the fact that a linear function on it yields a co-planar result.
81
Chapter
13
Representations of Functions
No computation
without representation
Adin Falkoff
A. Introduction
A family of monadic functions is commonly represented by a single dyadic
function, a particular member of the family being obtained by bonding a
parameter. As an example, consider the permutation or anagram function
introduced in Chapter 2:
a=:'ABCDE'
2 A. a
ABDCE
f=:2&A.
f a
ABDCE
A family may also be represented in several ways, using different dyadic
functions. For example:
0 1 4 3{a
ABED
p=:0 1 3 2 4
p{a
ABDCE
p&{ a
ABDCE
]b=:0;1;3 2;4
+-+-+---+-+
|0|1|3 2|4|
+-+-+---+-+
b C. a
ABDCE
The indexing or from function
A permutation vector (a permutation of i.5)
A monadic permutation function
The cycle function C.
82
Exploring Math
b&C. a
ABDCE
A monadic permutation function
Since different representations have different uses, it is important to provide
transformations from one to the other. The monadic cases of A. and C. provide
such transformations:
A. p
2
(A. p) A. a
ABDCE
]b=: C. p
+-+-+---+-+
|0|1|3 2|4|
+-+-+---+-+
C. b
0 1 3 2 4
The behaviour of these various representations of permutations can be studied by
using random permutations generated by the function ?.~. For example:
]q=:?.~9
7 1 3 2 6 4 0 5 8
A. q
288918
(A.q)A. i.9
7 1 3 2 6 4 0 5 8
C. q
+-+---+---------+-+
|1|3 2|7 5 4 6 0|8|
+-+---+---------+-+
Exercises
1. Generate a table of all permutations of order 4.
[(i.!4)A. i.4]
2. Use the example of q=:?.~9 and C.q to illustrate the scheme used in the cycle
representation of permutations.
[The third box of C.q signifies that item 5 moves to position 7, item 4 to
position 5, item 6 to 4, item 0 to 6, and item 7 to 0. Moreover, item 8 moves to
8 (and therefore remains fixed). Use the help menu for discussion of
permutations in the introduction to the dictionary, the vocabulary, and the
phrase book.]
3. Is a permutation a linear function? If it is, produce the matrix m that represents
it in the expression m&(mp=:+/ . *) .
[m=:q=/i.#q]
Chapter 13 Representations of Functions
83
85
Chapter
14
Polynomials
A. Coefficients Representation
A function that is a multiple of a non-negative integral power of its argument is
called a monomial. In MN it is written in the form 3x2, yielding the value 12 if the
argument x has the value 2.
A sum of monomials is called a polynomial, and is written in MN in the form
2x0+4x1+3x2+x3, having the value 30 if x is 2.
A direct translation to J would read as (2*x^0)+(4*x^1)+(3*x^2)+x^3. The
numerous parentheses required suggest a reason for the precedence rules adopted
in MN (power before multiplication before addition); they are precisely the rules
that permit the polynomial to be expressed without parentheses.
Exercises
1. Write a parenthesis-free J expression for the foregoing polynomial, then
assign the value 2 to x and enter the expression to test its validity.
Use the results of Exercise 1 to define a function py so that 2 4 3 1 py x
2.
yields the value of the polynomial for any single argument x.
[py=:+/@([ * ] ^ i.@#@[)]
3. Use py to define a function poly so that it applies to each element of a list x,
and test it by using it with the arguments 2 4 3 1 and i.8.
[poly=:py"1 0]
4. Comment on the function 2 4 3 1&poly.
[The dyadic function poly represents a family of polynomials, 2 4 3 1&poly
is a specific member of this family. The elements of the list 2 4 3 1 are
called coefficients, and poly is said to be a coefficients representation of
polynomials.]
86
Exploring Math
5. The dyadic case of the primitive function p. is a coefficients representation of
polynomials. Experiment with the expression c p. x for various values of c
and x.
6. Experiment with tc p./ x, where tc is a table of coefficients and x is a list.
B. Roots Representation
The product */x-r is said to be a polynomial expressed in terms of the list of
roots r. It is called a polynomial because any such function can also be expressed
in a coefficients representation. Thus:
x=:4
r=:2 3 5
x-r
2 1 _1
*/x-r
_2
_30 31 _10 1 p. x
_2
The monadic case of p. applied to the boxed roots yields the coefficients of the
other representation:
p. <r
_30 31 _10 1
Exercises
7. Define a “polynomial in terms of roots” function pir such that r&pir x
evaluates a polynomial with roots r for the argument x.
8. Why are the elements of the list r in the function r&pir called roots?
[Each of the elements is a zero or root of the function in the sense that it
yields a zero result. For example, enter pir=: */@(]-[)"1 0 and r&pir r]
9. Every function of the form r&pir can be represented in the form c&p.. Is the
converse true?
Try to define a list s such that s&pir is equivalent to d&p., where d=:2*p.<r.
Then look at the result of p. d and of (p. d) p. x=:i.8 ]
10. Discuss the result of p. d.
[The dyadic function p. covers both the coefficients and roots representations.
If the left argument is open (not boxed), it is treated as a list of coefficients. If
it is boxed (and contains two items), the last item is the boxed list of roots,
and the first is the boxed “multiplier”. If it contains a single item b, it is
equivalent to 1;b (that is, a multiplier of 1).]
Chapter 14 Polynomials
87
C. Versatility
The polynomial is a most important function in math. This importance stems from
its versatility, which in turn stems from a few simple properties.
The discussion of these properties leads to a number of topics not yet discussed,
such as complex numbers, derivatives, power series, and transcendental functions
(including the exponential (^), sine (1&o.), and cosine (2&o.)). Even if you are
unfamiliar with such matters, you will probably find it fruitful and interesting to
use this section as an introduction to them, always remembering the injunction of
Chapter 1: do not spend too much time on matters that may be, at the moment,
beyond your powers.
In presenting the properties of polynomials we will use the following in
examples:
c=:1 3 3 1
d=:2 1 0 4
s=:c+d
p=:+//.c*/d
c;d;s;p
+-------+-------+-------+---------------+
|1 3 3 1|2 1 0 4|3 4 3 5|2 7 9 9 13 12 4|
+-------+-------+-------+---------------+
• The sum (or difference) of two polynomials is itself a polynomial. For
example, the polynomial f=:c&p.+d&p. is equivalent to the polynomial
g=:(c+d)&p..
• The product of polynomials is a polynomial: c&p.*d&p. equals p&p.
• Polynomials can be used to approximate a wide variety of important
functions. A power series is a polynomial whose coefficients are each
expressible as a function of its index. For example, the reciprocal factorial
function expc=:%@! specifies the power series approximation to the
exponential funtion. Thus:
expc=:%@!
]e8=:expc i.8
1 1 0.5 0.1666667 0.04166667 0.008333333
e8&p. i.4
1 2.71667 7.26667 18.4
^i.4
1 2.71828 7.38906 20.0855
• The derivative (that is, the rate of change or slope of the tangent to the graph)
of a polynomial is itself a polynomial. For example, the derivative of c&p. is
(1 }. c * i.#c)&p.
• The integral (or anti-derivative) of a polynomial is itself a polynomial. For
example the integral of c&p. is (0 , c % 1+i.#c)&p. .
• The composition (c&p.)@(d&p.) is a polynomial.
Exercises
88
Exploring Math
11. Experiment with the foregoing examples.
12. Define and use plus=:+ and times=:+//.@(*/) and der=:1: }. ] * i.@#
and int=:0: , ] % 1: + i.@# . Comment on their behaviour.
[der@int is an identity function. The function plus fails for arguments that
do not have the same number of items. Try the function plus=:+/@,: and
examine how the laminate function pads a shorter argument with (non-
significant) trailing zeros]
13. Explain the reason for the diagonal sums (produced by /.) used in the
function times.
[See the multiplication of decimal numbers in Section C of Chapter 7]
The Taylor series adverb t. produces a function that gives the coefficients of a
power series. For example:
c&p. t. i.8
1 3 3 1 0 0 0 0
(c&p.*d&p.) t. i.8
2 7 9 9 13 12 4 0
^ t. i. 7
1 1 0.5 0.1666667 0.04166667 0.008333333 0.001388889
sin=:1&o.
cos=:2&o.
]sc=:sin t. i.8
0 1 0 _0.1666667 0 0.008333333 0 _0.0001984127
]cc=:cos t. i.8
1 0 _0.5 0 0.04166667 0 _0.001388889 0
The power series for an ordinary polynomial (that is, one with a finite list of
coefficients) ends with (non-significant) zeros, but the series for a transcendental
function continues with non-zero terms. However, the coefficients for the
exponential, sine, and cosine diminish rapidly in magnitude. This rapid decline
accounts for the utility of a small number of terms in approximating functions.
Exercises
14. Predict and confirm the results of ((cos*cos)+(sin*sin))t. i.8
15. Repeat Ex 14 for ((cc times cc)plus(sc times sc))t. i.8
16. Repeat Ex 14 for (^t.i.8)times(^@-t.i.8)
17. The function h=:(1 2 3&p.)@(4 3&p.) is a polynomial. Determine its
coefficients by hand, and confirm the result by entering h t. i.8.
18. Read Section 9D (Expansion) of Book 2.
If f and g are polynomials, then (f*g)%g is equivalent to f. On the other hand,
division for an arbitrary pair (such as f%g) may be not a polynomial, but a power
series. For example:
Chapter 14 Polynomials
89
f=:1&p.
g=:1 _1 _1&p.
(f%g) t. i. 8
1 1 2 3 5 8 13 21
The foregoing Taylor series may be surprising: it is the Fibonacci series, in which
each item is the sum of the two preceding it. This matter is discussed in Concrete
Mathematics [3], and in Concrete Math Companion [2].
D. Parity
A function E is said to be even if E is equivalent to E@-, that is, E x equals E-x for
any argument x. Graphically this implies that the graph of an even function is
reflected in the vertical axis.
A function O is odd if O is equivalent to -O@-, that is, O x equals -O-x for any x.
Consequently, the graph of an odd function is reflected in the origin.
Exercises
19. What is the parity (odd or even) of each of the functions sine and cosine?
20. Enter sin t. i.8 and cos t. i.8 and comment on the power series of odd
and even functions.
[The coefficients of all odd powers of an even function are zero, and conversely.]
21. What are the parities of the products of an even function with an even; an
even function with an odd; an odd with an odd? Test your assertions.
22. What is the parity of the exponential function?
The exponential is an example of a function that is neither odd nor even.
However, any function can be expressed as the sum of two functions, an odd part
and an even part. For example:
opex=:2: %~ ^ - ^@-
epex=:2: %~ ^ + ^@-
(^,opex,epex,:opex+epex) i.8
1 2.71828 7.38906 20.0855 54.5982 148.413 403.429 1096.63
0 1.1752 3.62686 10.0179 27.2899 74.2032 201.713 548.316
1 1.54308 3.7622 10.0677 27.3082 74.2099 201.716 548.317
1 2.71828 7.38906 20.0855 54.5982 148.413 403.429 1096.63
The odd and even parts of a function may be functions of interest in their own
right. In the present case, opex and epex are the hyperbolic sine and hyperbolic
cosine (often abbreviated as sinh and cosh); denoted in J as illustrated below:
90
Exploring Math
(5&o.,:6&o.)i.8
0 1.1752 3.62686 10.0179 27.2899 74.2032 201.713 548.316
1 1.54308 3.7622 10.0677 27.3082 74.2099 201.716 548.317
The adverbs O=: .:- and E=: ..- produce the odd and even parts of functions to
which they are applied. For example, ^O is equivalent to opex and ^E is
equivalent to epex.
Exercises
23. Compare the coefficients ^t.i.8 and opex t.i.8 and epex t.i.8 .
24. Comment on the functions cos E and cos O and sin E and sin O .
25. The function j. multiplies its argument by 0j1, the “imaginary” square root
of negative _1. Comment on the even function ^@j.E .
[^@j.E is the cosine. Try entering ^@j. t. i.8 and ^@j.E t. i.8]
E. Linearity
Since (c+d)p. x equals (c p. x)+(d p. x), it appears that a polynomial is in
some sense a linear function of its coefficients. We will now consider a series of
examples to clarify this vague statement, producing the matrix that represents the
linear function, and a simple expression for it as a power table:
mp=:+/ . *
c=:1 3 3 1 [ d=:2 1 0 4 [ x=:1 2 3 4
(c p. x);(c&p. x);(p.&x c)
+-----------+-----------+-----------+
|8 27 64 125|8 27 64 125|8 27 64 125|
+-----------+-----------+-----------+
I=:=i.4
I
1 0 0 0
0 1 0 0
0 0 1 0
0 0 0 1
]m=:|:p.&x"1 I
p.&x
1 1 1 1
1 2 4 8
1 3 9 27
1 4 16 64
m mp c
8 27 64 125
m&mp c
8 27 64 125
The matrix that represents the linear function
The matrix m that represents the desired linear function of the coefficients looks
like a power table, and may be so expressed in terms of the argument x and its
indices as follows:
]e=:i.#x
0 1 2 3
x ^/ e
1 1 1 1
1 2 4 8
1 3 9 27
1 4 16 64
Chapter 14 Polynomials
91
The table m is called the (Complete) Vandermonde matrix of x. A Vandermonde
function may be defined and used as follows:
V=:] ^/ i.@[
(#x) V x
1 1 1 1
1 2 4 8
1 3 9 27
1 4 16 64
]y=:((#x) V x)&mp c
Vandermonde
8 27 64 125
f=:c&p.
f x
8 27 64 125
CV=:# V ]
]y=:(CV x)&mp c
8 27 64 125
The Vandermonde function
Vandermonde matrix for x
Linear function in terms of
Complete Vandermonde function
The complete Vandermonde matrix is square and invertible. Its inverse provides
the inverse linear function, which may be used to determine the coefficients of a
polynomial that represents the function as illustrated below:
%. CV x
4 _6 4 _1
_4.33333 9.5 _7 1.83333
1.5 _4 3.5 _1
_0.1666667 0.5 _0.5 0.1666667
(%.CV x)&mp y
1 3 3 1
(%.CV x)&mp f x
1 3 3 1
f
1 3 3 1&p.
The inverse linear function applied to y
Using the fact that y is f x
Show the definition of f
Exercises
26. Use the foregoing discussion as a model for experimenting with
Vandermonde matrices for various values of the arguments x and c, and
comment on the results.
[The linear function (CV x)&mp applies only to arguments
92
Exploring Math
that have the same number of items as does x.]
27. Use x=:10%~i.10 and y=:(sin=:1&o.) x to obtain coefficients c such that
c&p. agrees with sin for the arguments x. Use expressions of the form (c&p.
; sin) 5 2$x to show the comparison clearly.
28. Test the use of c&p. to approximate sin by evaluating (c&p.-sin)z for other
arguments such as z=:0.65 0.8 and z=:i.5 2 and comment on the results.
[The approximation is good in the range covered by x
(0-0.9), but may be very bad for arguments outside this range.]
F. Polynomial Approximations
Sections C and E have presented two methods of approximating a function f by a
polynomial. The first used the Taylor series f t. i. n, and the second the
complete Vandermonde matrix CV x to fit the function exactly at the points f x.
We will first compare their results for the example treated in Exercise 27:
'`sin mp CV'=: (1&o.)`(mp=:+/ . *)`(# (V=:] ^/ i.@[) ])
x=:10%~i.10
tc=:sin t. i. # x
vc=:(%.CV x) mp sin x
((sin-tc&p.);(sin-vc&p.)) 5 2$x
+-------------------------+------------------------+
| 0 0|_2.18587e_13 1.5066e_12|
|_5.27356e_16 _4.43534e_14|_4.92267e_12 1.02958e_11|
|_1.04966e_12 _1.22129e_11|_1.40716e_11 1.53755e_11|
|_9.06789e_11 _4.9381e_10|_4.78062e_12 1.69003e_11|
| _2.14316e_9 _7.82095e_9| 3.04181e_11 6.7725e_11|
+-------------------------+------------------------+
((sin-tc&p.);(sin-vc&p.)) 5 2$x+0.1
+-------------------------+------------------------+
| 0 _5.27356e_16| 1.5066e_12 _4.92267e_12|
|_4.43534e_14 _1.04966e_12|1.02958e_11 _1.40716e_11|
|_1.22129e_11 _9.06789e_11|1.53755e_11 _4.78062e_12|
| _4.9381e_10 _2.14316e_9|1.69003e_11 3.04181e_11|
| _7.82095e_9 _2.48923e_8| 6.7725e_11 1.05946e_8|
+-------------------------+------------------------+
((sin-tc&p.);(sin-vc&p.)) 5 2$-x
+-----------------------+-------------------------+
| 0 0|_2.18587e_13 _1.07342e_8|
|5.27356e_16 4.43534e_14| _9.83991e_8 _5.0716e_7|
|1.04966e_12 1.22129e_11| _1.92828e_6 _6.01583e_6|
|9.06789e_11 4.9381e_10| _1.62857e_5 _3.95772e_5|
| 2.14316e_9 7.82095e_9| _8.83031e_5 _0.000183771|
+-----------------------+-------------------------+
The first panel above shows that vc provides the better approximation at the very
points on which it was determined; the second panel shows that this better
performance persists for other points in the range spanned by them; and the third
shows that the Taylor series generally performs better at points (that is, -x)
outside the range.
Chapter 14 Polynomials
93
Exercises
29. Will the use of a larger number of terms in a polynomial approximation
improve its fidelity? Experiment to test the matter.
[Not necessarily. Although the higher-order elements of the coefficients f t.
i.n may decrease rapidly, the power of the argument by which each is
multiplied in the polynomial evaluation may rapidly increase. The resulting
product produced (to limited precision) may introduce large “round-off”
errors.]
We will now develop a polynomial of lower degree that provides a “least-squares
best fit” to the values f x. With a left argument less than #x the Vandermonde
function V produces non-square power tables as illustrated below:
x=:1 2 3 4
(1&V;2&V;3&V;4&V;CV) x
+-+---+------+---------+---------+
|1|1 1|1 1 1|1 1 1 1|1 1 1 1|
|1|1 2|1 2 4|1 2 4 8|1 2 4 8|
|1|1 3|1 3 9|1 3 9 27|1 3 9 27|
|1|1 4|1 4 16|1 4 16 64|1 4 16 64|
+-+---+------+---------+---------+
Although these matrices are not square, they may be used with the generalized
inverse function denoted by %. as illustrated below:
%.3 V x
2.25 _0.75 _1.25 0.75
_1.55 1.15 1.35 _0.95
0.25 _0.25 _0.25 0.25
%.2 V x
1 0.5 0 _0.5
_0.3 _0.1 0.1 0.3
f=:(c=:1 3 3 1)&p.
]vc3=:(%. 3 V x) mp f x
11.5 _13.7 10.5
vc3 p. x
8.3 26.1 64.9 124.7
c p. x
8 27 64 125
The matrix product (%. 3 V x) mp f x used above can be written more simply
as a “matrix divide”, by using the dyadic case of the function %. . Thus:
(f x)%.(3 V x)
11.5 _13.7 10.5
Finally we define a conjunction FIT such that n FIT f x gives an n-element list
of coefficients that fits the function f at the points x. Thus:
94
Exploring Math
FIT=:2 : 'y. %. (x. & V)'
3 FIT ^
^ %. 3&V
c=:3 FIT ^ y=:0.1*i.7
c p. y
1.00133 1.10388 1.22004 1.3498 1.49317 1.65015 1.82073
^y
1 1.10517 1.2214 1.34986 1.49182 1.64872 1.82212
f
1 3 3 1&p.
d=:3 FIT f x
d p. x
8.3 26.1 64.9 124.7
f x
8 27 64 125
Exercises
30. Experiment with the conjunction FIT for various values of its parameters.
Include the example used at the beginning of this section, and compare the fit
provided by the coefficients tc with that provided by the five-element list
tc5=:5 FIT sin x.
95
Chapter
15
Arithmetic
A. Introduction
As remarked in Chapter 1, arithmetic is that branch of mathematics that deals
with whole numbers. As treated in Book 2, it includes topics such as
permutations, polynomials, and logic. These are usually considered to be
advanced topics, to be treated only after the introduction of fractions, irrational
numbers, and even complex numbers. What are the potential advantages of
extending the treatment of arithmetic in this manner?
•
It may serve to defer the treatment of fractions until the student has matured
through experience gained in many interesting uses of whole numbers. How
many cooks fear the use of fractions involved in dividing a recipe? Is 2/3
really a number since it cannot be written in decimal, although 3/4 can? And
how many question the point of complex numbers whose mechanics are often
elaborated long before any of their interesting uses can be shown?
• Although polynomials may be of little practical use when limited to integer
arguments, notions such as the product of coefficients (+//.c*/d) remain
meaningful and interesting. Indeed they provide useful insights into the
products of multi-digit numbers, as shown in Chapters 7 and 14.
B. Insidious Inverses
The familiar counting numbers may be defined as follows: there is a first (denoted
by 1), and a successor function (denoted by >:). Thus:
>:1
2
>:2
3
>:>:>:1
4
96
Exploring Math
An inverse predecessor function (denoted by <:) undoes the work of the
successor. Thus:
<:4
3
<:>:3
3
However, <: is not a proper inverse, because its application to the first counting
number cannot yield a counting number. Thus:
<:1
0
<:0
_1
<:_1
_2
In other words, the introduction of a seemingly-innocent inverse has broadened
the class of counting numbers to define the class of integers, which includes zero
and negative numbers. The introduction of the further classes of rationals,
irrationals, and complex numbers can be viewed in a similar light.
Exercises
1. Illustrate the fact that the successor and predecessor are proper inverses on the
domain of integers. Include examples of the powers <:^:n and >:^:n for both
positive and negative values of n.
2. Same and illustrate the use of a function that has a proper inverse on some
domain.
[On the domain of permutation vectors (permutations of the
integers i.n), the grade (/:) is its own proper inverse.]
3. Experiment with some of the inverse pairs listed in the definition of the power
conjunction (^:) in the J dictionary [5].
4. Read the discussion in the first three pages of Book 2.
5. Study Section 2 I (Identity Elements and Infinities) of Book 2.
C. Rational Numbers
The multiplication of two integers yields an integer. Moreover, division is inverse
to multiplication in the sense illustrated below:
_2*8
_16
(_2*8)%8
_2
Chapter 15 Arithmetic
97
More precisely, if i is an integer, then the functions *&i and %&i are inverse:
i=:8
*&i _2
_16
%&i *&i _2
_2
Again, %&i is not a proper inverse because it may lead out of the class of integers,
producing the class of rationals. For example:
%&i _2
_0.25
Exercises
6. Illustrate the fact that the rationals are closed under multiplication and
division.
D. Irrational Numbers
The square function is closed on the rationals, and the square root (%:) provides
an inverse. For example:
]r=:3%5
0.6
*:r
0.36
%: *: r
0.6
Again %: is not a proper inverse, because there is no rational whose square is 2,
and the result is to introduce a further class of irrationals. Because there is at least
one rational between any pair of distinct rationals (their average), it might seem
impossible that there could be any numbers that are not rational. However, the
school of Pythagoras produced a rather straightforward argument to show that the
square root of 2 (the length of the hypotenuse of a right-triangle with sides of unit
length) is not a rational.
E. Complex Numbers
Because there is no rational whose square is negative, the square root applied to a
negative argument leads to the further class of complex numbers. Thus:
98
Exploring Math
%:_1
0j1
]a=:i.6
0 1 2 3 4 5
%:a
0 1 1.41421 1.73205 2 2.23607
%:-a
0 0j1 0j1.41421 0j1.73205 0j2 0j2.23607
Taken together with the rationals, these imaginary square roots of negative
numbers form the class of complex numbers, closed under square root as well as
under addition, subtraction, multiplication, and division.
Exercises
7. Read Section 9F (Real and Complex Numbers) of Book 2.
8. Read Chapter 7 (Permutations) of Book 2.
9. Read Chapter 8 (Classification and Sets) of Book 2.
99
Chapter
16
Complex Numbers
A. Introduction
The following tables illustrate some of the consequences of adding the imaginary
square root of minus one to the number system:
Bordered table adverb
adapted from Ch. 3
Addition table
Multiplication table
T=:1 : '[by]over x./' ~
by=:[:":' '&;@,.@[,.]
over=:({.;}.)@":@,
]i=:%:_1
0j1
]c=:(i.4),i*i.4
0 1 2 3 0 0j1 0j2 0j3
+T c
+---+-------------------------------+
| | 0 1 2 3 0 0j1 0j2 0j3|
+---+-------------------------------+
| 0| 0 1 2 3 0 0j1 0j2 0j3|
| 1| 1 2 3 4 1 1j1 1j2 1j3|
| 2| 2 3 4 5 2 2j1 2j2 2j3|
| 3| 3 4 5 6 3 3j1 3j2 3j3|
| 0| 0 1 2 3 0 0j1 0j2 0j3|
|0j1|0j1 1j1 2j1 3j1 0j1 0j2 0j3 0j4|
|0j2|0j2 1j2 2j2 3j2 0j2 0j3 0j4 0j5|
|0j3|0j3 1j3 2j3 3j3 0j3 0j4 0j5 0j6|
+---+-------------------------------+
*T c
+---+---------------------------+
| |0 1 2 3 0 0j1 0j2 0j3|
+---+---------------------------+
| 0|0 0 0 0 0 0 0 0|
| 1|0 1 2 3 0 0j1 0j2 0j3|
| 2|0 2 4 6 0 0j2 0j4 0j6|
| 3|0 3 6 9 0 0j3 0j6 0j9|
| 0|0 0 0 0 0 0 0 0|
|0j1|0 0j1 0j2 0j3 0 _1 _2 _3|
|0j2|0 0j2 0j4 0j6 0 _2 _4 _6|
|0j3|0 0j3 0j6 0j9 0 _3 _6 _9|
+---+---------------------------+
100
Exploring Math
%:T 1 2 3 0j1 0j2 0j3 Roots
+---+---------------------------------------------------------------------------------------------+
| |1 2 3 0j1 0j2 0j3|
+---+---------------------------------------------------------------------------------------------+
| 1|1 2 3 0j1 0j2 0j3|
| 2|1 1.41421 1.73205 0.707107j0.707107 1j1 1.22474j1.22474|
| 3|1 1.25992 1.44225 0.866025j0.5 1.09112j0.629961 1.24902j0.721125|
|0j1|1 0.769239j_0.638961 0.454832j_0.890577 4.81048 3.70041j_3.07371 2.18796j_4.2841|
|0j2|1 0.940542j_0.339677 0.852887j_0.522096 2.19328 2.06287j_0.745007 1.87062j_1.1451|
|0j3|1 0.973427j_0.228999 0.933693j_0.358074 1.68809 1.64323j_0.386571 1.57616j_0.604461|
+---+---------------------------------------------------------------------------------------------+
Exercises
1. Comment on the foregoing tables, including the two-part representation that
appears to be used for each complex number.
2. Enter |@+T c and comment on the results.
3. Study the tables for other functions such as -, %, and ^ (and perhaps even +.
and *. and ^. and |).
Two-part representations for individual numbers are not uncommon:
• The result of 36%4 is represented as 9.25, using an integer part and a
fractional part joined by a dot.
• The result of 23*10^5 can also be represented as 23e5, using a factor and an
exponent joined by the letter e.
• The rational 2%3 can be represented as 2r3, using a numerator and
denominator joined by the letter r.
• Two pi cubed (2*(o.1)^3) can be represented as 2p3 using a factor and an
exponent joined by the letter p.
• The complex number 3+4*%:_1 is represented as 3j4, using a real part and an
imaginary part joined by the letter j.
• Further cases may be found in the discussion of constants in the J dictionary.
The monadic function | used in the table |@+T a is called magnitude; it yields
the square root of the sum of the squares of the real and imaginary parts of an
argument. When applied to a real (non-complex) number it is sometimes called
the absolute value.
Functions defined on real numbers are extended to complex numbers without
change, except that they apply to the new element %:_1 according to the normal
rules. The extended functions can therefore be examined in terms of elementary
algebra.
B. Addition
The sum of complex numbers can be analyzed in terms of their real and imaginary
components as follows:
i=:%:_1
ar=:5 [ ai=:2 [ br=:3 [ bi=:4
(a=:ar+i*ai),(b=:br+i*bi)
5j2 3j4
The following sequence of identities shows that the components of a sum are the
sums of the components:
Chapter 16 Complex Numbers
101
a+b
(ar+i*ai)+(br+i*bi)
ar+br+(i*ai)+(i*bi)
(ar+br) + (i*(ai+bi))
Definitions of a and b
Addition is associative and commutative
Multiplication by i (that is, i&*) distributes
over +
Exercises
4. Enter the foregoing sequence and check that each of the sentences yield the
same result.
5. Write and enter a corresponding sequence for multiplication.
C. Multiplication
In discussing multiplication we will use further functions, illustrated as follows:
a=:5j2 [ b=:3j4
]ca=:+. a
5 2
]cb=:+. b
3 4
]ab=:+. a,b
5 2
3 4
j. 4
0j4
3 j. 4
3j4
j./cb
3j4
j./+.b
3j4
102
Exploring Math
Multiplication is analyzed in the following sequence of identities:
a*b
(j./ca)*(j./cb)
(ar+j.ai)*(br+j.bi)
(ar*(br+j.bi))+((j.ai)*(br+j.bi))
(ar*br)+(ar*j.bi)+((j.ai)*br)+((j.ai)*j.bi)
(ar*br)+(ar*j.bi)+((j.ai)*br)-(ai*bi)
((ar*br)-(ai*bi))+(ar*j.bi)+((j.ai)*br)
((ar*br)-(ai*bi))+(j.(ar*bi)+(ai*br))
Exercises
6. Express the result of the foregoing sequence in English.
[The real part of a product is the difference of the product of the component
lists; the imaginary part is the sum of the real part of each multiplied by the
imaginary part of the other.]
7. Re-express the final sentence of the sequence in terms of the table ab=:+.a,b
.
[(-/*/ab)+ (j.+/*/0 1|."0 1 ab)]
The function +. produces a two-element vector representation of a complex
argument in terms of its real and imaginary components. If we plot the point
whose coordinates are given by +. and draw a line to it from the origin we see the
possibility of another two-element representation in terms of the length of the line
and its angle. This is called a polar representation, and is given by the function
*.. Thus:
*.b
5 0.9272952
Angle in radian units rather than degrees
|b
5
Magnitude (also called absolute value
for real arguments)
Multiplication is easily expressed in terms of the polar representation: the
magnitude is the product of the magnitudes, and the angle is the sum of the
angles. For example:
*. a,b,a*b
5.38516 0.3805064
5 0.9272952
26.9258 1.3078
*/|a,b
26.9258
(+.,*.)a,b,a*b
5 2
3 4
7 26
5.38516 0.3805064
5 0.9272952
26.9258 1.3078
Both representations
Chapter 16 Complex Numbers
103
The measure of an angle in radians is the length of arc measured on a circle of
radius one unit; consequently one-half pi radians is a right-angle, and therefore
equivalent to 90 degrees, and pi radians is a “straight” angle of 180 degrees. Since
the constant 180p_1 is 180 multiplied by the reciprocal of pi, the conversions
between radians and degrees may be expressed as follows:
rfd=:1r180p1&*
Radians from degrees
dfr=:180p_1&*
Degrees from radians
rfd 0 45 90 180
0 0.7853982 1.5708 3.14159
dfr rfd 0 45 90 180
0 45 90 180
pid=:({.,dfr@{:)"1@*.
Polar representation in degrees
pid a,b,0j1,1j1,_1j0
5.38516 21.8014
5 53.1301
1 90
1.41421 45
1 180
D. Powers and Roots
We will illustrate the use of powers and roots by developing a function to give the
coordinates of regular polygons:
Second (square) root of _1
First four powers of second root of _1
Coordinates of 4-sided polygon (square)
Cube root of _1
Coordinates of hexagon
2%:_1
0j1
(2%:_1)^i.4
1 0j1 _1 0j_1
+.(2%:_1)^i.4
1 0
0 1
_1 0
0 _1
3%:_1
0.5j0.8660254
+.(3%:_1)^i.6
1 0
0.5 0.8660254
_0.5 0.8660254
_1 1.22461e_16
_0.5 _0.8660254
0.5 _0.8660254
reg=:+.@((-:%:_1:)^i.)"0 Function for regular polygons
104
Exploring Math
Boxed polygons of 3-6 sides
<@reg 3 4 5 6
+--------------+-----+-------------------+--------------+
| | | 1 0| 1 0|
| 1 0| 1 0| 0.309017 0.951057| 0.5 0.866025|
|_0.5 0.866025| 0 1|_0.809017 0.587785|_0.5 0.866025|
|_0.5 _0.866025|_1 0|_0.809017 _0.587785| _1 0|
| | 0 _1| 0.309017 _0.951057|_0.5 _0.866025|
| | | | 0.5 _0.866025|
+--------------+-----+-------------------+--------------+
Compare the function reg with that used in Chapter 11, and use the plotting
functions of that chapter in the following Exercises.
Exercises
8. Plot the figures (reg 4);(2*reg 4) in contrasting colors.
9. Use the function rot of Chapter 11 to plot rotated figures.
E. Division
Since %&b (division by b) is the inverse of *&b (multiplication by b), division is
easily expressed in a polar representation: the magnitude is the quotient of the
magnitudes, and the angle is the difference of the angles. For example:
a%b
0.92j_0.56
*.a,b,a%b
5.38516 0.3805064
5 0.9272952
1.07703 _0.5467888
%/|a,b
1.07703
A complex number may be normalized by dividing it by its magnitude, yielding a
complex number with magnitude 1. For example:
b%5
0.6j0.8
|b%5
1
norm=: ]%|
]nb=:norm b
0.6j0.8
Since a normalized number can be restored by multiplying its norm by a real
number, it is often convenient to work in terms of normalized numbers, and then
multiply results by appropriate real scale factors.
The reciprocal of a normalized number is simply related to the number itself. For
example:
Chapter 16 Complex Numbers
105
%nb
0.6j_0.8
+nb
0.6j_0.8
b*+b
25
%:b*+b
5
The monad + is called the conjugate; it reverses
the sign of the imaginary part
The product with the conjugate is a real number; the
magnitude is its square root.
We have yet to examine division in terms of the real/imaginary representation.
This may be approached by noting that a%b is equivalent to a*%b (that is,
multiplication by the reciprocal). Since we already have expressions for the
product and the reciprocal, the overall result can be obtained by simple, but
perhaps tedious, algebra.
107
Chapter
17
Calculus
A. Secant Slope
If a function f is plotted over a range of arguments that includes x and y, then the
straight line through the points x,f x and y,f y is called a secant line, and the
quotient of the differences (f y)-(f x) and y-x is called its slope. This slope
gives the approximate rate of change of the function in the vicinity of x and y. For
example:
f=:*:
’x y’=:1 3
rise=:(f y)-(f x)
run=:y-x
]slope=:rise%run
4
The secant slope may be expressed in a function that uses the run as the left
argument, and in an adverb that may be applied to any function:
ss=:(f@+-f@])%[
2 ss 1
4
SS=:1 : '(x.@+-x.@])%['"0
2 f SS 1
4
2 ^&3 SS 1
13
Secant slope of cube with run of 2 at 1
]r=:10^-i.6
1 0.1 0.01 0.001 0.0001 1e_5
Slopes of cube for various runs and points of
x=:i.7
r ^&3 SS/ x
application
1 7 19 37 61 91 127
0.01 3.31 12.61 27.91 49.21 76.51 109.81
0.0001 3.0301 12.0601 27.0901 48.1201 75.1501 108.18
1e_6 3.003 12.006 27.009 48.012 75.015 108.018
1e_8 3.0003 12.0006 27.0009 48.0012 75.0015 108.002
1e_10 3.00003 12.0001 27.0001 48.0001 75.0001 108
108
Exploring Math
B. Derivative
As the run decreases in size, the slope appears to be approaching a limit, which
we may interpret as the derivative, the slope of the tangent at the point x.
However, a zero value for the run gives only the meaningless ratio of 0 divided by
0:
0 ^&3 SS/ x
0 0 0 0 0 0 0
For the case of the cube, this derivative may be obtained exactly, because the
cube of x+r is (x^3)+(3*(x^2)*r)+(3*x*r^2)+(r^3), and the rise (found by
subtracting x^3) is(3*(x^2)*r)+(3*x*r^2)+(r^3). Dividing this by the run
gives (3*x^2)+(3*x*r)+(r^2). Setting r to zero in this expression gives 3*x^2,
the derivative of the cube at the point x.
The function for the derivative of the cube may therefore be expressed and used
as follows:
d3=:3:*^&2
d3 x
0 3 12 27 48 75 108
This result may be compared with the final row of the table of secant
slopes.Similar analysis for other powers yields d4=:4:*^&3 for the derivative of
^&4, d5=:5:*^&4 for the derivative of ^&5, and so on. We define a corresponding
adverb for the derivative of any power:
D=:1 : 'x.&*@(]^x."_-1:)'
2 D x
0 2 4 6 8 10 12
3 D x
0 3 12 27 48 75 108
4 D x
0 4 32 108 256 500 864
None of this constitutes a proof that the derivatives of all powers follow this
pattern, but it does suggest an induction hypothesis for a recursive proof. This
matter is treated in Book 3.
Exercises
1. If f=:^&3 is the cube and g=:5:*f is five times the cube, what is the
derivative of g?
[Five times the derivative of f, that is, 5:*3:*^&2. Since any secant slope of g
is five times the slope of f, the same is true of the limiting value, that is, the
derivative.]
2. If h=:2:*^&4, what is the derivative of the sum s=:g+h?
[The sum of the derivatives of g and h,
that is, (5:*3:*^&2)+(2:*4:*^&3)]
Chapter 17 Calculus
109
3. If C=:3 1 4 2"_ and E=:0 1 2 3"_ are constant functions, then
t=:+/@(C*]^E) is a weighted sum of powers. What is its derivative?
[der=:+/@(C*E*]^E-1:)"0. Try der 1 2 3 4 5]
C. Polynomials
The preceding Exercises developed the fact that the derivative of a weighted sum
of powers is itself such a sum, with the exponents decreased by 1. Since a
polynomial is a weighted sum of powers, its derivative is also a polynomial, of
degree one less. The derivative of c&p. is d&p., where the coefficients d are
obtained from c by applying the following function:
dc=:1:}.]*i.@#
For example:
c=:6 5 4 3
]i.#c
0 1 2 3
c*i.#c
0 5 8 9
1}.c*i.#c
5 8 9
]d=:dc c
5 8 9
(c&p. ,: d&p.) x=:i.7
6 18 56 138 282 506 828
5 22 57 110 181 270 377
Exercises
4. Use the fact that the polynomial 0 0 0 1&p.is equivalent to the cube function
to compare the use of the function dc with the derivatives of powers obtained
in the preceding section.
[Compare (dc 0 0 0 1)&p. x=:i.7 with 3 D x ]
5. Comment on the polynomial (dc dc c)&p..
[This is the second derivative of c&p., that is, the rate of change of the rate of
change. For example, if c&p. gives the position of a vehicle, then (dc c)&p.
gives its speed, and (dc dc c)&p. gives its acceleration.]
110
Exploring Math
D. Differential Equations
Many important functions are simply related to their rates of change, their first or
second derivatives. For example, capital invested at compound interest increases
at a rate proportional to its value, and the exponential or growth function
(denoted by ^) increases at a rate equal to itself. In other words, the exponential is
equal to its derivative.
Is there a polynomial with this property? Clearly not, since the derivative of a
polynomial is of lower degree, possessing one less term. However, it is possible
to define a power series having the desired property.
6. Try to develop a rule or function to generate the coefficients of a power series
Exercises
that equals its derivative.
[Hint: Begin with the coefficients c=:1 1 1 1 1 1,
and apply the function dc to it.]
Pursuing the idea suggested in the exercise we have:
c=:1 1 1 1 1 1
dc c
1 2 3 4 5
Since the second element of the derivative dc c is twice the value of the
corresponding element of c, we replace the third element by one-half its value to
compensate:
c=:1 1 1r2 1 1 1
dc c
1 1 3 4 5
Since the third element of dc c is now six times its required value of one-half, we
replace the fourth element of c by 1r6, and so on:
dc c=:1 1 1r2 1r6 1 1
1 1 0.5 4 5
dc c=:1 1 1r2 1r6 1r24 1
1 1 0.5 0.1666667 5
dc c=:1 1 1r2 1r6 1r24 1r120
1 1 0.5 0.1666667 0.04166667
dc dc c
1 1 0.5 0.1666667
It should now be clear that the coefficients are the reciprocal factorials:
]c=:%!i.6
1 1 0.5 0.1666667 0.04166667 0.008333333
Chapter 17 Calculus
111
dc c
1 1 0.5 0.1666667 0.04166667
ce=:%@!@i.
ce 6
1 1 0.5 0.1666667 0.04166667 0.008333333
Coefficients for exponential
(ce 10) p. x=:i.4
1 2.71828 7.38871 20.0634
^x
1 2.71828 7.38906 20.0855
Ten-term approximation to exponential
We have, in effect, defined the exponential as that function which satisfies (i.e., is
the solution of) an equation that requires it to equal its own derivative. We may
write such equations more clearly in terms of the following derivative adverb:
D=:("0)(D.1)
^&3 D
3&*@(^&2)"0
^&3 D x=:i.6
0 3 12 27 48 75
The scalar first derivative adverb
The derivative of the cube
Applied to an argument
^D x
argument
1 2.71828 7.38906 20.0855 54.5982 148.413
Derivative of the exponential applied to
(^ = ^D) x
1 1 1 1 1 1
Test of the differential equation satisfied by ^
The hyperbolic sine (5&o.) and the hyperbolic cosine (6&o.) introduced in
Chapter 14 both satisfy a similar equation, but one that involves the second
derivative:
(5&o. = 5&o. D D) x
1 1 1 1 1 1
(6&o. = 6&o. D D) x
1 1 1 1 1 1
Sinh equals its second derivative
Cosh equals its second derivative
(1&o. = -@(1&o. D D)) x Sin is minus its second derivative
1 1 1 1 1 1
(2&o. = -@(2&o. D D)) x Cos is minus its second derivative
1 1 1 1 1 1
Exercises
7. Use the differential equation satisfied by the hyperbolic cosine together with
the approach suggested in Exercise 6 to develop a power series for it.
[coshc=:ce*0:=2:|i.. Use the Taylor series
112
Exploring Math
6&o. t. i.6 to confirm this solution]
8. Use Taylor series as guides in defining functions to generate power series for
the hyperbolic sine, cosine, and sine.
9. Experiment with the weighted Taylor coefficients adverb t:for each of the
functions treated in Exercises 6-8, study the patterns produced, and state its
definition.
10. Predict and confirm the result of ^@- t: i. 10.
11. Study and experiment with the table of derivatives given in Sec. B, Chapter 2
of Book 3.
E. The Exponential Family
In Chapter 13 we introduced odd and even adverbs that produced the odd and
even parts of functions to which they were applied. Moreover, we saw that the
odd part of the exponential was equivalent to the hyperbolic sine, and that the
even part was equivalent to the hyperbolic cosine. Thus:
O=: .:-
E=: ..-
(^O , ^E , ^ ,: ^O+^E) x=:i.6
0 1.1752 3.62686 10.0179 27.2899 74.2032
1 1.54308 3.7622 10.0677 27.3082 74.2099
1 2.71828 7.38906 20.0855 54.5982 148.413
1 2.71828 7.38906 20.0855 54.5982 148.413
(^O t. ,: ^E t.)x
^
0 1 0 0.1666667 0 0.008333333
1 0 0.5 0 0.04166667 0
Coefficients of odd and even parts of
(5&o.t. ,: 6&o.t.)x
cosine
0 1 0 0.1666667 0 0.008333333
1 0 0.5 0 0.04166667 0
Coefficients of hyperbolic sine and
Weighted Taylor coefficients
(^O t: ,: ^E t:)x
0 1 0 1 0 1
1 0 1 0 1 0
(5&o.t: ,: 6&o.t:)x
0 1 0 1 0 1
1 0 1 0 1 0
If j. is applied to the argument of the hyperbolic sine (to make it imaginary) , the
odd positions of the coefficients of the resulting function 6&o.@j. are unaffected,
because they are all zero. Moreover, those in each fourth place are multiplied by
_1 (that is the fourth power of j.1). The function 6&o.@j. is therefore equivalent
to the cosine. Thus:
6&o.@j. t. x
1 0 _0.5 0 0.04166667 0
Chapter 17 Calculus
113
2&o. t. x
1 0 _0.5 0 0.04166667 0
The sine may also be similarly expressed in terms of the hyperbolic sine.
Moreover all four of these functions can be expressed directly in terms of the
exponential, using only the function j. and the odd and even adverbs.
Finally, the real and imaginary parts of the function ^@j. are the cosine and sine
respectively. For example:
(+.^@j. x) ; ((cos ,. sin) x)
+-------------------+-------------------+
| 1 0| 1 0|
| 0.540302 0.841471| 0.540302 0.841471|
|_0.416147 0.909297|_0.416147 0.909297|
|_0.989992 0.14112|_0.989992 0.14112|
|_0.653644 _0.756802|_0.653644 _0.756802|
| 0.283662 _0.958924| 0.283662 _0.958924|
+-------------------+-------------------+
Exercises
12. Study the plot of sine versus cosine in Section 9J of Book 2.
13. See Chapters 3 (Vector Calculus) and 4 (Difference Calculus) of Book 3.
115
Chapter
18
Inverses and Equations
A. Inverse Functions
The many scattered references to “inverse” in the index suggests the ubiquity of
the notion in math. The general reason for its importance appears in the following
example: if we use heat=:*&4@*: to compute the output of an electric heater as a
function of the voltage applied, we will commonly need the inverse
volts=:%:@(%&4) to determine what voltage would be required to produce a
desired amount of heat. Thus:
heat=:*&4@*:
volts=:%:@(%&4)
(];heat;volts@heat) i.5
+---------+------------+---------+
|0 1 2 3 4|0 4 16 36 64|0 1 2 3 4|
+---------+------------+---------+
A method for obtaining the inverse of a composition of two functions may be seen
in the following example:
cff=:m@s
m=:100r180&*
s=:-&32
cff temp=:_40 32 212
_40 0 100
Celsius from Fahrenheit
Multiply by conversion factor
Subtract a conversion constant
Celsius for equal, freezing, boiling points
Inverse of m
Inverse of s
im=:m I=:^:_1
is=:s I
m s temp
_40 0 100
im m s temp
_72 0 180
is im m s temp
_40 32 212
ffc=:is@im
ffc cff temp
_40 32 212
116
Exploring Math
cff ffc temp
_40 32 212
In general, if several functions are applied one after the other, the inverse is
obtained by applying their inverses in reverse order.
Exercises
1. Define the adverb FI=: f. ^:_1 (fix and invert) and predict and confirm the
results of applying it to each of the following functions:
cff@ffc
is@im
m@s
cff
2. Repeat Exercise 1 for the following functions (perhaps using the simpler
I=:^:_1 instead of FI):
^
^.
^@^.
(^*^@-)
[The last function gives a domain error, because ^*^@- is a constant
function (giving 1 for any argument), and a constant function cannot have
an inverse.]
3. Repeat Exercise 1 for the following functions:
*:
%:
^&2
^&3
^&3@%
4. Although *:2 and *:_2 both yield 4, the “inverse” function %: yields only 2
when applied to 4. Comment on this matter.
B. Monotonic Functions
A (strictly) monotonic function is one that tends in the same direction as its
argument increases. A graph of such a function f (as, for example, f=:^) provides
a visualization of its inverse as follows: at any point y on the vertical axis draw a
horizontal line to intersect the graph of f, and from the point of intersection draw
a vertical line to intersect the horizontal axis at x. Then y is f x, and conversely x
is f^:_1 y.
A similar treatment of a non-monotonic function can illuminate the matter raised
in Exercise 4: the square function f=:*: graphed on a domain that includes both
negative and positive arguments is seen to be an even function, and a horizontal
line through a point such as y=:4 intersects the graph in two points, giving two
possible values for the inverse.
Only a strictly monotonic function can have a proper inverse, but a non-
monotonic function may have a useful inverse when restricted to a principal
domain in which it is monotonic. In the case of the square, the non-negative real
numbers provide such a principal domain, and the inverse *:^:_1 provides the
inverse on it.
An inverse for arguments not in a principal domain is often easily obtained from
the inverse on the principal domain. In the case of the square it is simply
-@(*:^:_1).
Any periodic function (such as the sine or cosine) cannot be monotonic, but may
be when restricted to a suitable domain.
Chapter 18 Inverses and Equations
117
Exercises
5. Define a function pn that gives both positive and negative inverses of the
square function, and test it on the argument x=:0 1 4 9 16 25.
[pn=:(],.-)@%: (Or use ,: or ; instead of ,..)]
6. Experiment with the functions n&o. and their inverses (-n)&o. for integer
values of n from 0 to 8. Which of the inverses have restricted principal
domains?
7. What are the limits of the principal domains of _1&o. and _2&o.?
[Apply them to the argument _1 1]
C. Under
I=:^:_1
idr=:10&#.
dr=: idr I
dr x=:213
2 1 3
idr dr x
213
(];dr;idr@dr) x
+---+-----+---+
|213|2 1 3|213|
+---+-----+---+
az=:,&0
az dr x
2 1 3 0
idr az dr x
2130
x*10
2130
Inverse of decimal representation; i.e., decimal value
Decimal representation
Append zero
Decimal value with appended zero
The foregoing elaborates the familiar idea that a number can be multiplied by ten
by appending a zero to its decimal representation. The full expression may be
paraphrased in English as “Obtain the decimal representation, append a zero, then
evaluate the resulting list in decimal (that is, apply the function inverse to the
decimal representation). It illustrates the form f^:_1 g f that occurs so often that
it is also provided by the conjunction &. as follows:
idr@az@dr x
2130
az&.dr x
2130
118
Exploring Math
The general idea is that f&.g applies f under g, in the sense that g “prepares” the
argument for the function f, and the “preparation” is finally undone. For example:
Double under natural logarithm
Is equivalent to squaring
+:&.^. y=:4
16
*:y
16
+:&.(10&^.) y=:4
16
-:&.(10&^.) y=:4
2
%:y
2
8. Paraphrase the foregoing expressions in detail.
Exercises
[+:&.^. takes the natural logarithm of its argument,
doubles it, and applies the exponential (inverse log).]
9. Experiment with the expressions 6+&.(10&^.)3 and 6-&.(10&^.)3 and
comment on the results.
[The first multiplies its arguments by adding their base-10 logarithms and
applying the anti-log (that is, ten-to-the-power); the second uses subtraction to
obtain the quotient. The dyadic case of the function f&.g is similar to the
monadic, but applies the “preparation” function g to each of the arguments]
10. Define the function saf=:</\ (suppress after first) and experiment with the
expressions saf b and saf&.|. b for various values of the Boolean list b,
such as b=: 0 0 1 0 1 1 0. Comment on the results.
[saf suppresses all ones after the first in a Boolean list; saf&.|.
suppresses all before the last by first reversing the list, and again
reversing the resulting list after applying the function saf.]
D. Equations
A function such as f=:3 _4 1&p. may not have a known inverse, but we can
obtain the inverse of a given argument such as y=: 6 by solving the equation y=f
x; that is, by finding a value x that satisfies the indicated relation.
If we know values a and b such that f is monotonic in the interval from a to b,
and if y lies in the interval from f a to f b, then a suitable solution x can be
obtained by simple repeated approximations: take the average of a and b; consider
the intervals bounded by it and each of them; and choose as a new interval the one
whose function values still embrace the argument y.
Chapter 18 Inverses and Equations
119
See Section C of Chapter 7 of Book 3 for an executable definition of the
foregoing bisection method, and Sections D and E for the faster Newton and
Kerner methods that employ derivatives.
The many uses of equations and their solutions in math can mostly be seen as
limited means of obtaining inverse functions.
120
Exploring Math
121
Chapter
19
Readings
A. Introduction
Reading any math text can serve as a stimulus to further exploration, whatever
notation it may be expressed in. Those, such as Book 2 and Book 3, that are
expressed in J are particularly accessible to users of this book. We will here
discuss other books of this type that are easily available because they can be
conveniently displayed on the screen (by using the Help menu), and because
selections from them can be printed (using the resulting Print menu) for study.
We will here present a few examples from two such books, J Phrases and J
Dictionary.
B. Phrases
After printing the Table of Contents and displaying and reading the first page of
the book of J Phrases to learn the conventions used, you may choose any chapter
for further exploration. Some, such as Chapter 2 (Primitive Notions) and Chapter
8 (Numbers), will provide further elaboration of matters already treated in earlier
chapters here. Others, such as Chapters 12 and 13 (Finance and Data) enter new
territory.
Chapter 16 (Extended Topics) provides an entree to a wide variety of topics
addressed by three authors: C. Burke, D.B. McIntyre, and C. Reiter.
C. Sample Topics
This section of J Dictionary and Introduction provides brief treatments of a
variety of topics. You might begin with the discussion of Classification and Sets
(Sections 8-11), and continue with Directed Graphs and Closure (Sections 20-21).
The discussion of polynomials (Sections 23-28) covers some material already
treated here in Chapter 14, but also includes matters such as explicit functions for
Newton’s and Kerner’s methods for finding roots, as well as stopes that
generalize the notions of falling factorials and rising factorials.
122
Exploring Math
D. Vocabulary and Definitions
Begin by printing out the Vocabulary. Then with the vocabulary displayed, click
the mouse on any definition, such as Self-Classify . Equal in the upper left corner.
A study of the defintion will probably provide all the information you need
concerning the conventions used. If not, display the page of the dictionary headed
by III. Definitions for details of them.
123
References
1. Reiter, Clifford A., Fractals, Visualization, and J, Second Edition, Jsoftware,
2000.
2. Iverson, Kenneth E., Concrete Math Companion, Jsoftware, 1995.
3. Graham, Ronald L., Donald E. Knuth, Oren Patashnik, Concrete
Mathematics, Addison Wesley, 1989.
4. Lakatos, Imre, Proofs and Refutations: the logic of mathematical discovery,
Cambridge University Press, 1976
5. Hui, Roger K.W., and Kenneth E. Iverson, J Dictionary, Jsoftware, 1998.
This text is available on-line in the J system, as discussed in Chapter 10.
6. American Heritage Dictionary of the English Language, Houghton-Mifflin.
(Any edition that includes the appendix of Indo-European roots.)
7. Thomas, Lewis, et cetera, et cetera : Notes of a Word-Watcher, Little, Brown
and Company, 1990.
125
Index
absolute value, 102, 104
base-value, 37, 38
addition, 1, 5, 9, 18, 24, 44, 46, 56, 57, 75, 76,
bisection, 121
80, 87, 100
Adverbs, 23, 27
agenda, 50, 53
alphabet, 28
ambivalent, 24
anagram, 10, 12, 83
and, 20
appendices, 7
approximating functions, 90
area transformation, 80
arguments, 6, 18, 23, 24, 31, 34, 38, 46, 50, 69,
78, 79, 87, 90, 94, 97, 104, 109
arithmetic, 7, 33, 57, 58, 97
Arithmetic, 97
arithmetic progression, 57
array, 29, 30, 34
assign, 5, 87
associative, 57, 103
associativity, 57, 75
atom, 34, 50
atop, 5
average, 31, 33, 99
axes, 30, 31
axioms, 14
axis, 31, 91
backspace, 5
base-8, 37, 40
block, 40, 41
Boole, 20
Boolean, 120
Bordered, 18, 101
box, 13, 30, 64, 69, 70, 84
boxed roots, 88
Boxing, 32
by, 24, 101
calculus, 7
Calculus, 109
carries, 34, 39, 40, 44
catenate, 27
Celsius, 117
Classification, 122
Closure, 122
Coefficients Representation, 87
Colors, 67
commutative, 18, 26, 57, 103
commutativity, 57, 75
commuted, 20
companion volume, 7
Comparisons, 2
complex numbers, 4, 70, 89, 97, 98, 99, 100, 102
Complex Numbers, 99, 101
composition, 25, 89, 117
Concrete Math Companion, 7, 91
126
Exploring Math
conjunction, 7, 25, 33, 34, 50, 60, 76, 78, 96, 98
dfr, 105
conjunctions, 27, 76
diagonally, 44
constant, 27, 29, 34, 50, 51, 57, 105, 111
Difference Calculus, 115
constant function, 118
constants, 27, 59, 102
conventional notation, 7
Differential Equations, 112
Directed Graphs, 122
Displacements, 66
conventions, 123
distributes, 56, 75, 76, 80, 103
coordinate geometry, 65
coordinate system, 65
Coordinates, 65
copula, 25, 28, 42
Distributivity, 75
divided by, 24, 110
Division, 106
dropping the menus, 63
copulative conjunction, 25
Editing, 64
cosine, 89, 90, 91, 92, 113, 114, 115
English, 5, 10, 11, 12, 23, 24, 25, 27, 28, 31, 33,
counting numbers, 97, 98
cube, 3, 33, 59, 109, 110, 111, 113
cursor, 5, 64
Data, 122
Decimal, 37
decimal point, 28
Decimal representation, 119
deductive, 14
Definitions, 123
degenerate triangle, 80
degrees, 104, 105
delete, 5
denominator, 102
34, 40, 104, 125
Equal, 123
equals, 2, 3, 4, 56, 57, 59, 89, 91, 92, 112, 113
Equations, 120
erase, 5
etymology, 34, 66
even, 1, 3, 9, 10, 14, 18, 23, 68, 79, 91, 92, 97,
102, 114, 115
executable, 121
execution, 24, 26, 27, 50, 64
exploration, 1, 2, 4, 7, 14, 63
exponent, 102
exponential, 24, 89, 90, 91, 112, 113, 114, 115
Exponential Family, 114
derivative, 89, 110, 111, 112, 113
Extended Topics, 122
Derivative, 110
dervatives, 89
determinant, 68, 80
determinants, 69
factor, 80, 102
factorial, 12, 19, 24, 49, 89
Fahrenheit, 117
falling factorials, 122
false, 20
Fibonacci series, 91
Finance, 122
Fractals, 7, 125
Index
127
improve, 14, 28, 95
indexing, 52, 83
Indo-European, 35, 125
induction hypothesis, 59, 110
fractional part, 102
induction hypothesis), 59
function, 2, 3, 5, 6, 10, 12, 13, 19, 24, 25, 27, 30,
31, 34, 37, 38, 40, 41, 42, 43, 44, 45, 46, 47,
49, 50, 51, 52, 53, 58, 59, 60, 61, 66, 68, 75,
76, 77, 78, 79, 80, 81, 83, 84, 85, 87, 88, 89,
90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 102,
104, 105, 106, 109, 110, 111, 112, 113, 114,
115
INDUCTIVE PROOF, 59
Infinite rank, 50
Inner Product, 78
integer part, 102
Function Tables, 17
generalized inverse, 95
geometric figures, 65, 80
geometric progression, 58
gerund, 50, 51
gopen, 67, 70
gpolygon, 67
grammar, 23, 24, 27, 63
Grammar, 23, 24
graph, 65, 66, 67, 68, 89, 91
Greatest Common Divisor, 20
guesses, 14
Help, 122
help menu, 63, 84
heron, 66
Heron’s formula, 66, 68
hierarchical rules, 24
hyperbolic, 92, 113, 114, 115
identities, 103, 104
identity, 11, 49, 56, 79, 90
Identity, 5, 98
imaginary, 92, 100, 101, 102, 104, 107, 114, 115
integers, 3, 9, 12, 40, 49, 59, 65, 98, 99
interpreted, 24, 51, 52
intervals, 120
inverse, 10, 38, 51, 78, 93, 95, 97, 98, 99, 106
Inverse, 117
Inverses, 117
INVERSES, 97
Irrational Numbers, 99
irrationals, 98, 99
iteration, 40, 41
J, 2, 7, 23, 24, 25, 27, 28, 33, 34, 40, 63, 64, 67,
87, 92, 98, 102, 125
J Introduction and Dictionary, 122
J Phrases, 122
Kerner, 121
Kerner’s, 122
Lakatos, 14, 56, 57, 125
laminate, 27, 90
languages, 14
Least Common Multiple, 20
left to right, 26, 27, 65
length, 26, 66, 99, 104, 105
128
Exploring Math
Lewis Thomas, 34
Linear functions, 76
Linear Functions, 75
noun, 9, 25, 34, 50
nouns, 24, 25, 28
Nouns, 23
Linear Vector Functions, 77
number of items, 12, 31, 40, 76, 79, 90, 94
Linearity, 76, 92
link, 13, 27
list, 2, 3, 5, 6, 10, 11, 12, 17, 18, 27, 28, 29, 34,
38, 40, 42, 43, 44, 46, 56, 57, 60, 87, 88, 90,
96
logic, 7, 14, 97, 125
Number of items, 31
Numbers, 122
numerator, 102
oblique, 44, 45
octal, 37
magnitude, 46, 90, 102, 104, 106, 107
odd numbers, 3, 5, 6, 51, 55
major cells, 31
operator, 34, 51
math, 1, 2, 4, 5, 6, 7, 9, 24, 33, 34, 63, 76, 89
or, 20
Math, 9
mathematical, 12, 125
matrix, 34, 60, 78, 79, 81, 85, 92, 93, 94, 96
matrix product, 78
mean, 31, 32, 33, 34, 39, 58
Mixed Bases, 47
MN, 23, 24, 33, 34, 87
monomial, 87
Monotonic, 118
multiplication, 1, 5, 9, 17, 18, 24, 26, 44, 56, 87,
90, 98, 99, 100, 103, 106, 107
Multiplication, 44, 103
native language, 23
natural logarithm, 120
negative numbers, with the standard form limited
(as it is for positive arguments) to numbers, 46
Newton, 121
Newton’s, 122
normalization, 40, 44, 47
normalized number, 107
origin, 65, 66, 91, 104
over, 101
Padding, 44
Parity, 91
parse, 24, 25
parsed, 24
pattern, 6, 56, 110
patterns, 4, 5, 14, 114
pentagon, 69
perform, 34, 39, 40, 52
perimeter, 26, 66
periodic function, 118
permutation, 9, 10, 11, 52, 56, 81, 83, 84, 85, 98
permutations, 7, 10, 11, 12, 52, 70, 84, 97, 98
permuted, 10, 12
permuting, 3
Phrases, 122
pi, 33, 34, 102, 105
Plotting, 69
Index
129
PLOTTING, 70
polar representation, 104, 106
polygons, 66, 68, 70, 105, 106
radian units, 104
radians, 105
Ramble, 6
polynomial, 60, 61, 87, 88, 89, 90, 91, 92, 93, 94,
random, 3
95, 111, 112
Polynomial Approximations, 94
polynomials, 7, 24, 87, 88, 89, 91, 97, 122
Polynomials, 87
POLYNOMIALS, 111
power, 1, 3, 24, 87, 89, 90, 91, 92, 93, 95, 98,
110, 112, 113, 114
Power, 19
power series, 89, 112
random generator, 29
rank conjunction, 50
rank-k, 31
ranks, 32
rate of change, 89, 109, 111
Rational Numbers, 98
rationals, 98, 99, 100
Readings, 122
POWERS AND ROOTS, 105
reciprocal, 24, 89, 105, 107, 112
predecessor, 97, 98
Primitive Notions, 122
principal domain, 118
Pro-adverb, 28, 33
Pronoun, 28, 33
Pronouns, 23
Recursion, 49
recursive proof, 110
Refutations, 14, 125
reg, 70, 105, 106
relation, 1, 9, 12, 34, 35, 42, 55, 75
relations, 9, 10, 12, 14, 34
proof, 14, 55, 56, 57, 58, 59, 60, 61, 110
proofs, 7, 9, 13, 14, 55, 58, 60, 61
Relations, 9
repeatable, 3
Proofs, 13, 55
proper inverse, 98, 99
proposition, 57
Proverb, 28, 33
punctuation, 24
Punctuation, 25
Pythagoras, 66, 99
quadrant, 66
quadrants, 18
quotient, 106, 109
repeated approximations, 120
report, 29, 30, 31, 32, 34
Reports, 29
representation, 28, 37, 38, 39, 40, 43, 45, 46, 68,
79, 83, 84, 87, 88, 102, 104, 105, 106, 107
Representations Of Functions, 83
Research, 6
reversal, 38
Reverse, 31
rfd, 68, 105
130
Exploring Math
right parenthesis, 41
right to left, 26
rising factorials, 122
Roots Representation, 88
Roots table, 102
Rotate, 31, 66
stopping value, 49
subtotals, 6, 78
Subtraction, 46
successor, 97, 98
sum, 6, 24, 25, 29, 39, 40, 43, 51, 55, 56, 57, 58,
59, 60, 61, 75, 77, 78, 87, 89, 91, 102, 103,
104, 110, 111
SAMPLE TOPICS, 122
sum function, 6, 25
Save As, 64
scan, 25
script, 59, 64, 68
Script Windows, 64
secant line, 109
Secant Slope, 109
second derivative, 113
selection, 52
Self-Classify, 123
sets, 7
Sets, 122
Shape, 31
signum, 46, 51
similar, 3, 9, 10, 35, 98, 113
sine, 89, 90, 91, 92, 113, 114, 115
slope, 89, 109, 110
solving, 120
sort, 3, 4, 9, 10, 13
specific arguments, 34
Spelling, 23, 28
square, 3, 118
squares, 6, 59, 95, 102
stitch, 27
stopping condition, 49
symmetric, 4, 18, 56, 57, 65
table, 13, 17, 18, 19, 24, 34, 43, 44, 45, 52, 68,
69, 78, 79, 81, 84, 88, 92, 93, 101, 102, 104,
110, 114
TABLE, 18, 19, 20, 23
tables, 5, 13, 17, 18, 20, 30, 52, 76, 95, 101, 102
tangent, 89, 110
tangible representations, 14
Taylor adverb, 61
Taylor series, 90, 91, 94, 95, 113, 114
Taylor series adverb, 90
Terminology, 33
tetrahedron, 81
the reciprocal factorials, 112
ties, 50
Tools, 63
Tower of Hanoi, 51
transcendental functions, 89
transformations, 14
translating, 7
transpose, 13, 18
TRANSPOSITION, 31
tree, 34, 35
true, 20, 23, 34, 35, 55, 57, 88, 110
Two-part representations, 102
Index
131
Under, 119
under open, 69
Vandermonde matrices, 94
Visualization, 7, 65, 68, 125
Vocabulary, 123
volume, 7, 81
Vandermonde matrix, 93, 94
weighted sum, 77, 111
variable, 34
Weighted Taylor coefficients, 114
vector, 34, 66, 77, 78, 79, 80, 81, 83, 104
whole numbers, 3, 7, 97
Vector Calculus, 115
verbs, 5, 23, 25, 27, 28
Versatility, 89
Width, 25, 26
window, 59, 64, 67, 68
word-formation, 28