N. VILENKIN COMBINATORIAL MATHEMATICS FOR RECREATION Translated from the Russian by George Yankovsky MIR PUBLISHERS Moscow‘UDC 519.1=20 First published 1972 Revised from the 1969 Russian edition Ha asraniicxom ssuKeCONTENTS 5 preface... 2. eee eee ees © 7 A bookshelf problem... ...... 40 Sanerent a renlonnenarte ores King Arthur's Round Table... . . 40 OF COMBINATORICS She's got a date... .. eee ee Superstitious cyclists... 2... g A session in telepathy... ..... . 4B Permutations with repetitions... .. | 9 General problem of derangements... . 44 Number systems........... 10 Subfactorials . pe Secret lock. ..----... 2. 41 Caravan in the deserts) 1) 48 Morse code... ........... ‘1 Merry-goround ...., 2.2... 6. Wigwag code... ss... .. ff Standing in line ata ticket office... . 48 Electronic digital computer... . 42 The problem of the two ranks... .. 54 Genetic code... .......... 43 New properties of combinations... . Sf General rules of combinatorics... . 13 uaprER IV, THE COMBINATORICS Domino problem ........... 45 oF PARTITIONS ‘The crew of a spaceship... ..... 15 ; Checkerboard problems... .. . eRe ee cS How many people don't kxow foreign Placing objects into cells... 6... 55 i 5 17 A bouquet of flowers... . cee anguages! vee ww ee - The number-of-divisors problem... . 56 The principle of inclusion and exclusion 18 icing apples 3 Where's the mistake? ..... 2... 20 ee ‘The sieve of Eratosthenes Oe eine pia ey ve Mailing photographs . 2... 2... ST CHAPTER IT, PERMUTATIONS AND COMBINATIONS Flags on masts... ......... 58 Football championship ........ 22 Total number of signals... ..... 59 Permutations without repetitions... . 22 Particle statistics... ....--. 59 A science club. ........... 22 Partitions of integers... ...... 59 Permutations of nelements . . .. . . - 23 Mailing packages... .. 1... 60 The problem of the rooks |... ... 23 General problem of postage stamps... 6 Linguistic problems. ......... 24 Combinatorial problems of information Round dance... .......... 25 theory. ....... pe ee OL Permutations with repetitions... . , 25 Entrance-exams problem... ... . . ot Anagrams . 2... 2... ee eee 26 Paying money. ........- 62 Combinations ............ 27 Buyingeandy..... 2... .- + 68 Genoese lottery ............ 29 Getting change ...........-. 64 Buying cakes... .......... 80 Partitioning integers... . tee 68 Combinations with repetitions sees. Bf Arrays of dots. . 2... 1 we + 6 The football championship again... . 32 Dual arrays » 1.1... eos or Properties of combinations... . . 33 Euler’s formula, .-..- ~~ vee 87 A particular case of the principle of CEAPTERtT Tome meronce into pectasion [ena kexeitsioe a ae 37 Wandering abouttown......... 70 Alternating sums of combinations... . 37 THe arithmetic square ss ee CHAPTER IIT. COMBINATORIAL PROBLEMS WITH Figurate numbers... ........ 74 RESTRICTIONS ‘The arithmetic triangle... ..... 72 Lions and tigers. .......... 39 The extended arithmetic triangle... . 73 Building a stairway. ......... 39 Thechessking............ %‘Tho generalized arithmetic triangle . . . Generalized arithmetic triangles and a base-m number system . . . Some properties of the numbers Cp, (k, n) ‘A checker in the corner . ‘The arithmetic pentagon... .... Geometric proof of properties of combina- tions . . . Random walks .........--- Brownian motion... ... bees ‘The queen's realm... .....- Absorbing barriers 5... 2... oe Random walks on an infinite plane . . . ‘The general problem of the rocks . . . . Symmetric arrangements ....... Two knights... CHAPTER VI. RECURRENCE RELATIONS Fibonacci numbers... 2... - An alternative prof ......... The process of successive partitions . . . Multiplying and dividing numbers . . . Problems involving polygons... .. Difficulties of a majordomo . Lucky trolleybus tickets... ... . Recurrence tables... .......- Alternative solution of the majordomo problem... ee. eee eee eee Solution of recurrence relations . . . . . Linear recurrence relations with constant coefficients 1B cs 7 8 19 8t BRLSE 87 89 1 ot 92. 93 94 96 97 98 99 100 The case of equal roots of a characteristic equation Application of the theory ‘of recurrence relations to problems of transmitting infor- mation . . A third solution to the majordomo problem CHAPTER VII. COMBINATORICS AND SERIES Dividing polynomials . Algebraic fractions and power series . . . Operations on power series . : Using power series to prove identities . . Generating functions. ........ Newton’s binomial theorem... . . . The multinomial theorem... . . _ Newton's series... 2... eu ee Extracting square roots... .. 04. Generating functions and recurrence rela- tions . . Decomposition into partial fractions . . . On a single nonlinear recurrence relation Generating functions and partitions of integers. Summary of the combinatorics of parti- tims 2... ee. Combinatorial Problems... ..... Solutions and Answers. ....... Index 2... ee 102 103 103 104 104 107 108 109 4109 14h 112 114 146 116 118 119 122, 123 152 205PREFACE Specialists in a broad range of fields have to deal with problems that involve combinations made up of letters, numbers, or any other objects. The department head in a factory has to allocate production assignments to machine-tool opera- tors, the agronomist has to decide on what crops to grow on a selected group of fields, the school principal draws up schedules of lessons, the investigating chemist analyzes relations involv- ing atoms and molecules, the linguist examines the meanings of combinations of letters in an unknown language, and so forth. The field of mathematics that studies problems of how many different combinations (subject to certain restric- tions) can be built out of a specific number of objects is called combinatorial mathematics (com- Binatorics). This branch of mathomatics has its origin in the 46th century, in the gambling games that played such a large part in high society in those times. Whole fortunes, from gold and precious stones to pedigreed horses, castles and estates, were won'or lost in a game of cards or dice. All manner of lotteries were in vogue. It is quite natural that jthe first combinatorial problems had to do mainly with gambling, such as in how many ways can a certain sum in throws of two or three dice be scored, or in how many ways is it possible to get two kings in a card game. These and other problems in games of chance gave the initial impetus to develop combinatorial mathematics and the burgeoning theory of pro- bability. One of the first to enumerate the various com- binations ‘achieved in games of dice was the Italian mathematician Tartaglia. He drew up a table illustrating the number of ways r dice can fall. It was not taken into account, however, that the same sum can be obtained in different ways (say, 1+ 34+4=4+2+4 2). In the 17th century, the French scholars Pascal and Fermat made a theoretical investigation into the problems of combinatorics. Again, the starting point was in the form of problems of games of chance, particularly the so-called prob- lem of points in determining the division of the stakes of an interrupted game of chance. This problem was posed to Pascal by his friend the Chevalier de Méré, an ardent gambler. Roughly, the problem was this: a match of coin tossing to six winning games is interrupted when one player has won five tosses and the other four tosses. How are the stakes to be divided? It was clear that a division of 5 to 4 would not be fair. Applying methods of combinatorics, Pascal solved the problem for the general case when one player has r games left to win and the other one has s games. An alternative solution was given by Fermat. Further advances in the theory of combinations were connected with the names of Jakob Bernoulli, Leibniz and Euler. However, in these studies the main role was played by appli- cations to various games (lotto, solitaire, etc.). During recent years, combinatorial mathematics has seen extensive developments associated with greater interest in problems of discrete mathe- matics, Combinatorial methods are employedin solving transport problems, in particul scheduling; the scheduling of production facili- ties and of the sale of goods. Links have beon established between combinatorics and problems of linear programming, statistics, ete. Combinato- rial methods are used in coding and decoding and in the solution of other problems of infor- mation theory. The combinatorial approach also plays a signi- ficant role in purely mathematical problems such as the theory of groups and their representations, in the study of the foundations of geometry, nonassociative algebras, and elsewhere. In the present book, the aim has been to set forth a variety of combinatorial problems in popular form and understandable language. At the same time, an attempt is m&de to present some rather involved combinatorial problems and to give the reader an idea of the methods of recurrence relations and generating functions. The first chapter is devoted to the general rules of combinatorics, the rules of sum and product. In the second chapter we investigate permutations and combinations. This traditio- nally grade-school material is accompanied by 8 an analysis of some amusing examples. In the third chapter, a study is made of combinatorial problems in which certain restrictions are imposed on the combinations. Chapter IV considers problems involving partitions of numbers into integers and contains a description of certain geometrical methods in combinatorics. Chapter V is devoted to random-walk problems and to a variety of modifications of the arithmetic triangle, Chapter VI takes up recurrence rela- tions, and Chapter VII discusses generating functions and, in particular, the binomial for- mula. ‘The last section of the book is devoted to com- binatorial problems of which there are over 400. ‘This material has been taken from a variety of sources, including Whitworth's Choice and Chance (London, 1901), John Riordan’s An Introduction to Combinatorial Analysis (New York, 1958), an interesting book by A. M. Yaglom and I. M. Yaglom entitled Nonelementary Problems in an Elementary Ezposition (Moscow, 1954), and various collections of problems given at mathe- matical olympiads in the USSR.CHAPTER I THE GENERAL RULES OF COMBINATORICS SUPERSTITIOUS CYCLISTS “Another eight!” exclaimed the chairman of the cyclists’ club with exasperation as he examined the twisted wheel of his bicycle. “And why? All because I was given this blasted No. 008 ticket when I became a member. Now hardly a month goes by without one of the wheels smashing into a figure eight. What I need is a ticket with a different number. To avoid accusations of superstition I'll reregister all the members, lea- ving out the digit eight altogether.” The next day he changed all the tickets. Now how many members were there in the club if it is known that all three-digit numbers were used that did not contain 8? (Say, 000 was used but not 836.) To start with, let us determine the number of one-digit numbers without eight. Clearly, there are nine such numbers: 0, 1, 2, 3, 4, 5, 6, 7, 9. Now indicate all the two-digit numbers lacking the numeral 8. They can be found by taking all the ono-digit numbers that we found and writing any one of the acceptable nine digits after it. This yields nine two-digit numbers for each one-digit number, and since there are 9 one-digit numbers, we get 9 X 9= 81 two- digit numbers without any 8's: 00, 01, 02, 03, 04, 05, 06, 07, 09 40, 41, 412, 13, 44, 15, 16, 17, 19 20, 21, 22, 23, 24, 25, 26, 27, 29 30, 31, 32, 33, 34, 35, 36, 37, 39 40, 41, 42, 43, 44, 45, 46, 47, 49 50, 54, 52, 53, 54, 55, 56, 57, 59 60, 61, 62, 63, 64, 65, 66, 67, 69 70, 71, 72, 73, 74, 75, 76, 77, 79 90, 91, 92, 93, 94, 95, 96, 97, 99 Thus, there are 9*= 81 two-digit numbers without 8. Each one of these numbers can be used to adjoin any one of the nine acceptable digits to. This yields 9? x 9 = 93 = 729 three- digit numbers. Therefore, the club had 729 regi- stered cyclists. If we were to take four-digit 635 K& 2% 2% “& a numbers, there would be 9!= 6,561 numbers without any 8's. In another club, the members were still more superstitious and they even threw out the num- ber 0 as being too much like a wheel. They made do with eight digits: 1, 2, 3, 4, 5, 6, 7, 9. How ‘many members were there in the club if all ticket numbers were three-digit? The problem is similar to the first one, the sole difference being that in place of 9 digits we have 8 altogether. So the club had 8° = 512 mem- bers. PERMUTATIONS WITH REPETITIONS The problem of the cyclists belongs in the following class of problems. We have objects referring to n different varieties. They are used to make up arrangements of k objects in each or, as we shall say in the future, k-arrangements. The arrangements can contain objects of the same type, and two arrangements will be con- sidered distinct if they differ either as to objects40 ‘or as to the order of the objects. The problem is to find the total number of such arrangements. Arrangements of this type are called k-permuta- tions of n distinct things with repetitions. The number of all such arrangements is denoted by 4k* In the first problem of the cyclists, ‘the number of things (types of elements) was 9 {we took all digits except 8), and each permuta- tion (each number) contained three elements. In this caso, the number of permutations came to 43= 9%. It is natural to assume that if the number of things is n, and cach permutation contains k elements, then the number of permu- tations with repetitions isn". ‘We wish to prove that the number of k-permu- tations of n distinct objects with repetitions is equal to Akon (4) ‘We carry out the proof by means of mathema- tical induction with respect to &, which is the number of elements in a permutation with a fixed value of n. For k= 1 the answer is obvious—each permutation (with repetitions) consists of one element only, and the different permutations are obtained by taking distinct objects. But since the number of types is n, the number of permutations is n. Thus, At in accord with formula (1). Now assume that we have the proof for Ak-! = = n"1 and we consider k-permutations with repetitions. We obtain all such permutations as follows. Let us take any (k —1)-permutation (with repetitions) (a, ..., ay) and adjoin to it the element a of one of the available n objects. We get some k-permutation (a,,... +++ Ghats aq). It is now clear that out of each (k —1)-permutation we get as many k-permuta- tions as there are distinct objects, that is, n per- mutations. It is obvious that in this manner we will not miss a single k-permutation and will * A permutation of n things taken r at a time be symbolized as ,P, or P (my 1) yn aise Translatort. not take any one twice [if (a, ...) p14) # F (bt, ++ oy Ops) OF if ay ~ bp, then (ay, .. fees Oyaty My) Fay vey Opts By) ‘There- fore the number of k-permutations of n distinct objects with repetitions is n times greater than the number of ( —1)-permutations of the same objects with repetitions. Consequently, 4% = = ndh-1, But we already have the proof of Ak-1 = nh-1, Therefore, AR = n-nb-l nk Thus, equation (1) has been proved for all values of ke Formula (1) occurs in a wide range of problems, a few of which will occupy us in the forthcoming sections. NUMBER SYSTEMS Besides the decimal number system which we commonly use, there are other systems of nume- ration such as the binary, ternary, octal, ete. number systems. In the base-n number system we use n digits. Let us calculate how many natural numbers can be written in a baso-n system of numeration by using exactly k digits (ior the sake of convenience, we will regard 0 as a natural number). If we allow for numbers beginning with zero, then every i-digit number in a base-n number system may be regarded as a permutation, with repetitions, consisting of k digits of n distinct types. Using formula (1) wo find that the number of numbers thus repre- sented is n*. However, the natural numbers do not have representations beginning with zero. We will thus have to subtract from the value of n* the number of numbers whose base-n representation begins with zero. Discarding the first digit (zero) in these numbers, we get a (k — 4)-digit number (which possibly also begins with zero). By for- mula (1), there will be n®1 such numbers. Hence, the total number of k-digit numbers in44 a base-n system of numeration is equal to nk (n—1) nh— nb For example, in the decimal system there are 10% X 9= 9,000 four-digit numbers: out of 10,000 numbers between 0 and 9,999, we discard one thousand, namely, those from 0 to 999, There is another way to derive our formula, In a k-digit number represented in a base-n num- ber system, the first digit can be any one of the digits 1, 2,..., n—4; the socond digit and succeeding digits, any one of the digits 0,4, 2, . +4, n—4. We thus have n—1 candi- dates for position one, and n candidates for each of the remaining k—1 positions. From this it is easy to see that the desired total number is (n— 4) nk, SECRET LOCK Safes and storage lockers make use of secret locks, called combination locks, that open up only when a specific combination of numbers or letters is dialled. Suppose a dial has 12 letters and the combination has 5 letters. How many failing attempts can be made by a person who does not know the combination? Using formula (1), we find the total number of combinations to be 125 = 248,832 which means that 248,831 attempts can be fai- lures, Incidentally, safes are frequently made to give a warning signal at the first failure to dial the right combination. MORSE CODE ‘The Morse code is used in telograph communica- tions. In this code, the letters, numbers and punctuation marks are represented by dots and dashes. Some characters require only a single sign, like Z (+) whereas others use all five signs, like zero, 0 (— ———— ). ‘Why the number 5? Couldn’t we make do with a smaller number of dots and dashes, say four, to transmit all our communications? The answer is no, and the reason becomes clear if we apply the formula for the number of permutations with repetitions. From formula (1) it follows that A} = 2. In other words, using one character, it is possible to transmit only two letters (E+ and T—). Using two characters, it is possible to transmit 24 = 4 letters, three yield 2° = 8 letters and four, 24= 16, Thus, the total number of letters that can be transmitted by means of four characters is 24+44+8416=30 Taking an alphabet of 26 letters, and also nume- rals and marks of punctuation we see that symbols made up of four characters do not suffice. But if we take combinations of 5, then we get 32 sym- bols in addition to the 30 obtained from four. ‘Those 62 symbols ar quite sufficient for tele- graph communications. There is also a five-digit code for telegraph communications that makes use of five symbols for every letter. Instead of the dash-and-dot system, uso is made of alternating current flow or of sending current signals and noncurrent signals. This code yields exactly 2°= 32 com- binations, which is enough for all the letters. Numerals and punctuation utilize the same com- binations, and so a telegraph apparatus opera- ting on the five-digit code has a special device for shifting from letters to numerals and back again, WIGWAG CODE The navy has a visual signalling system by flags, called a somaphore. Each letter is repre- sented by two flags in a specific arrangement.|| INTERVAL, 3 < % Ike lar A NUMERALS FOUN ) $l. 8, I ERROR (REPEATER sever E 5} jy For the most part, an individual letter is dis played by a pattern of one flag on each side of the signal-man. But some of the letters (say H, 0, X) require both flags on one side. The reason is obvious if we examine the formula for permutations with repetitions. The point is that each flag has a total of five positions: straight down, inclined down, horizontal, incli- ned up and straight up. Since we have two flags, the total number of combinations of the basic positions comes to A? = 5?= 25. One position has to be omitted (both flags down) since it serves to separate words. This yields 24 combinations, which is not sufficient to transmit all letters of the alphabet. That is why some letters have patterns with both flags on one side. ELECTRONIC DIGITAL COMPUTER Electronic computers are capable of handling an enormous range of problems. A single machine can decipher inscriptions in an unknown language, perform the design computations for a dam, and process the trajectory data of a space vehicle. How does one account for such versatility? Mainly the reason is that all these problems reduce to computations involving numbers. How does such a machine handle so many pro- blems involving so many numerical data? How many combinations of numbers is a computer capable of processing? Let us illustrate this case with a relatively small computer called “Strela” (Arrow). This413 machine has an immediate-access memory of 2,048 storage cells, each one of which can accom- modate 43 binary digits. Each digit is either a 0 or a 4. Altogether we have 43 x 2,048 > > 87,000 distinct positions, and two ways 40 or 1) the cells can be filled. Using formula (1), we find that the “Strela” can be in any one of more than 287.0 distinct states. This enormous num- ber is far beyond the limits of our imagination. Suffice it to say that the number of neutrons packed side by side in a sphere the radius of which is equal to the distance to the most distant ‘stellar systems does not exceed 25%, If we used just one memory cell, it would take nine years for a typist pool of 100,000 wor- ers to type out all the numbers that this single cell can accommodate (we assume they work -a seven-hour day and do one 43-digit number in 40 seconds). GENETIC CODE Breaking the genetic code has been one of the most remarkable achievements of twentieth century biology. Biologists now know how gene- tic (hereditary) information is passed on to the next generation. This information is recorded in the form of gigantic molecules of desoxyri- bonueleie acid (DNA). The various molecules of DNA differ in the order in which four nitro- genous bases are arranged: adenine, thymine, guanine, and cytosine. These bases determine ‘the order in which the proteins of the organism are built up out of roughly twenty amino acids, -each amino acid being in the form of a code made up of three nitrogenous bases. It is easy to seo where the number 3 came from. Using combinations of two bases, we could code only 4 = 16 amino acids, which is not enough. Using three bases, we get 4° = 64 combinations, which is far and away sufficient to encode some twenty amino acids, The intriguing ques- ‘tion is how nature takes advantage of so much redundant information—the number of com- binations is 64 while the number of amino acids is only one third that. A single chromosome contains several tens of millions of nitrogenous bases. The number of distinct combinations which they can form is simply horrendous—it is equal to 4%, where is the number of bases in the chromosome; go back to formula (1). A minute portion of these combinations has been sufficient to ensure the extraordinary diversity of all living nature over the entire span of life here on the planet Earth. Note of course that only a very small fraction of the theoretically possible combinations lead to viable organisms. GENERAL RULES OF COMBINATORICS As we shall soon see, combinatorial problems offer a multiplicity of types. But most of them can be solved with the aid of two basic rules: the rule of sum and therule of product. It is often possible to partition the combinations under study into several classes, one combination appearing in one and only one class. It is then clear that the total number of combinations is equal to the sum of the numbers of combinations in all classes, This assertion goes by the name of the rule of sum. It is sometimes formulated thus: If a certain object A can be chosen in m ways and another object B can be chosen in n ways, then the cholce of “either A or B” can be accomplished in m+n ways. When employing the rule of sum in this latter formulation, take care to see that no choice of A should coincide with any choice of B (or, as we put it at the beginning, that no combina- tion should appear in two classes at once). If there are such coincidences, then the rule of sum breaks down, and all we have is m-+n—k choices, where & is the number of coincidences. The second rule, called the rule of product, is somewhat more complicated. When making upcombinations of two objects, it is often known in how many ways the first object can be chosen and in how many ways the second one, the number of ways of choosing the second object being independent of how the first object was chosen. Suppose the first object can be chosen in m ways, the second in n ways. Then the pair can be chosen in mn ways, To put it differently: If an object A can be chosen in m ways and if, after every such choice, an object B can be chosen in n ways, then the choice of the pair (A, B) in that order can be accomplished in mn ways. To prove the rule of product, note that each one of the m ways of choosing A can be combined with n ways of choosing B, which brings us to mn ways of choosing the pair (A, B). The rule of product can be pictorialized as follows: Table 1 (At, Buys. (Ao, Bas), ++ (At, Bin) + + ++ (Ass Bon) Ai, «+1 Am indicate m ways of choosing A and Bit, . » »» Bin denote n ways of choosing B, provided A has been chosen the ith way. It is clear that this table contains all ways of choosing the pair (A, B) and consists of mn elements. If the choices of B are independent of how object A is chosen, then in place of Table 1 we get the following simpler table. Table 2 » (At, Bn) (At, Bi), (As, Bz), +1 (Ay Bn) (Az, By), (Az, Ba); (Am, B1), (Ams Bz), + + ey (Ams Bn) 14 It may, of course, happen that we need com- binations of more than two elements (objects). Then we arrive at the following problem. How many k-arrangements can be made if the first element can be one of ny distinct objects, the second, one of ng distinct objects, and the kth, one of nq distinct objects. Here, two arrangements are considered distinct if at least one position is occu- pled by different elements. This problem is solved in the same way as the problem of the cyclists. The first element may Fig. 4. be chosen in m, ways. Each of the chosen elements can be combined with any one of the nz types of second elements, yielding rng pairs. Each pair can be combined with any one of the n3 types of third elements, thus yielding nngng tri- ples. Continuing in this fashion, we finally get the k-tuple mnz ... my of arrangements of the desired type. In the problem of the cyclists, we had to choose three elements (the hundreds digit, the tens digit and the units digit). At each stage we could choose one of nine permissible digits, and so we got 9X 9 x 9= 729 numbers. Here is a harder problem. Signs are being made which consist of a geo- metrical figure (a circle, square, triangle or hexa- gon), a letter and a number. How many signs can be made? Asa first step we can choose a geometric figure, This can be done in four ways (we have a total of four figures). Then we can choose one of 32 let- ters (Russian alphabet), and finally, one of 10 numerals. This brings the total to 4 x 32 x X 40 = 4,280 combinations.DOMINO PROBLEM ‘A more complicated combinatorial problem is that in which the number of choices after each stage depends on the elements chosen in the preceding stages, an instance of which is the following. In how many ways can two dominoes be chosen out of 28 pieces so that they can be put together (that is, so that some number occurs on both pieces)? First choose one piece. This can be done in 28 ways. In 7 cases, the chosen piece will be a “double”, i.e. one of the type 00, 11, 22, 33, 44, 55, 66, and in 24 cases the piece will have diffe- rent nuimperg (say?"ua/ “13, ete.) ant cue aitse instance, the second piece may be chosen in 6 ways (for example, if at the first stage the piece 11 was chosen, then in the second stage, we can take one of the pieces with 01, 12, 13, 14, 15, 16). Now in the second instance, the second piece may be chosen in 12 ways (for piece 35 we can take pieces 03, 13, 23, 33, 34, 36, 05, 15, 25, 45, 55, 56). By the rule of product, we get 7X 6= 42 choices in the first case, and 21 X 12—= 252 choices in the second case. Hence, by the rule of sum, we get 42 + 252—= = 294 ways of choosing a pair. In the above discussion we took into account the order in which the pieces were chosen, so each pair of pieces appeared twice (for instance, Of and 16 the first time and 16 and 01 the second time). If the order of choice of the pieces is dis- regarded, then the number of choices is cut by one half: we have 147 choices. THE CREW OF A SPACESHIP For cases when at each stage the number of possible choices depends on which elements were chosen earlier, it is convenient to depict the process of building combinations in the form of a tree, Starting from one point, draw line 15 segments illustrating the various choices that can be made in the first stage (here, each sogment. corresponds to a single element). The choices in the second stage are made from the endpoints of each of the line segments, if the given ele- ment was chosen in the first stage, etc. Fig. 2. This construction yields a tree, an examination of which immediately gives the number of solu- tions to our problem. By way of illustration let us take the problem of staffing a multi-seated spaceship where the problem comes up of psychological compatibility of the crew in an extended space mission. People that are suitable in every respect when taken separately may not fit into the pattern of a crew for prolonged space exploration. Suppose our problem is to make up a crew of three: commander, engineer, and physician. We have four candida- tes, a1, ag, az and a for the commander, three candidates for the engineer (bj, bz and bs) and three for the doctor (c,, cz and ¢3). A preliminary checkup has disclosed that commander a, is psycho- logically compatible with engineers b, and bs and with doctors cy and cs, commander a, with16 engineers 6, and d, and all doctors, commander a3 with engineers b, and b, and doctors ¢, ¢s, commander a, with all engineers and with doc- tor ¢:. Also, engineer 6, does not get along with doctor 3, engineer b, with doctor e, and engi- neer b with doctor ¢. Given these conditions, in how many ways can a crew be made up? The conditions are illustrated in the tree in Fig. 2. We seo that there are only 10 permis- sible combinations (if there were no compatibi- lity restrictions, then the number of combina- tions would, by the rule of product, be 36 = =4X3X 3). CHECKERBOARD PROBLEMS Solve the following problem. In how many ways can two checkers (or draughts) (white and black) be placed on a checkerboard so that the white can take the black? Y Fig. 3. By the rules of the game, the pieces are placed on black squares and one checker takes another by jumping over it and occupying the next square (Fig. 3). If the piece reaches the last row, it beco- mes a king and can take all men on the same diagonal except those on the end squares. This problem is complicated by the fact that for different positions of white there are different numbers of positions of black in which it can be taken. For instance, if white is on square af, then there is only one position where black is under attack. Now if white is on square ¢3, then the number of desired positions for black GEG G MV 2: by v yy Yn ey by 4 VA Ve 3 4) 4: i Bb fy (by by Y (a) 0, YA Ve Gs Zz VA i Ved | (4 Z is w ey iY, w, is 4, Finally, if white has become a king on square 48, then there are 6 positions in which black is under attack by the king. It is therefore simpler to indicate for each position of white the number of possible positions of black and add the results obtained. Fig. 4a shows a checkerboard with the appropriate figuresoS indicated. Combining them, we get 87. Thus, the sought-for arrangement is possible in 87 ways. It is clear that there are exactly the same number of positions in which black can take white, but there are fewer positions in which both men can take one another. For instance, if white stands on the edge of the board, black cannot take it no matter where black is positio- ned. Therefore, assign 0 to all squares along the edges of the board. In the same manner, we find numbers that correspond to the other black squares. They are shown in Fig. 4b. Adding up these numbers, we find that the desired arran- gement is possible in 50 different ways. Finally, let us find the number of positions of white and black in which neither can take the other. The problem can be solved in the same way by placing white on each of the black squares and counting the possible ways of positioning black so that not one of the pieces could take the other. However, it is siinpler to make use of the “teakettle principle” * and reduce the prob- % The story goes that a mathematician once asked a physicist: “You have an empty teakettle and an uniit gas range. How do you go about boiling water?” «Very 259 17 lem to one that has already been solved. To do this, let us find the total number of positions for placing white and black on the board. White can be placed on any one of 32 black squares. That leaves 31 squares for black. Thus, by virtue of the rule of product, the arrangement is pos- sible in 32 X 31 = 992 ways. But these include 87 in which white can take black, and 87 in which black can take white, We have to reject 2X 87= 174 ways. However, we must take into consideration that in doing so we have rejected some of the ways twice: because white can take black and because black can take white, We have scon that there are 50 positions in which both men can take each other. Consequent- ly, the number of positions in which neither man can take the other is 92-174 4-50 = 868 HOW MANY PEOPLE DON'T KNOW FOREIGN LANGUAGES? The method applied in solving the foregoing problem can often be used in dealing with com- binatorial problems. Here is one, A research institution has a staff of 67: 47 know Spanish, 35 German and 28 both languages. How many employees at this institution know netther Spanish nor German? To solve this problem, we partition the entire staff into parts having no elements in common. The first part consists of those who know Spanish only, the second, those who know German only, the third, those who know both languages, and the fourth part, those who know neither language Simply,” replied the physicist. “Fill the teakettle, with ‘water light the gaa and put the water on to. boll.> “Right,” sald’ the mathematician, “Now solve this problem: the gas is burning and the teakettle has water Init How do you boil the waver?” “That's no problem at all,” replied the physicist. “Just put the teakettle on the range. “No,” ‘sald the ‘mathematician firmly, “You turn off ‘the ga’, pour out the water and we arrive at or Arst problem, which we know ow to solve.” Now, ‘when a new problem is reduced to an. already solved one, we speak of applying the “teakettle principle”.(Fig. 5). What we know is that Part Three con- sists of 23 persons, but since 47 know Spanish, there are 47 — 23 = 24 persons who know only Spanish, In the same way, we find that 35 — 23 = 12 scientists know only German. From this (b) Fig. 5. we conclude that the total number of persons Knowing one of these two languages is equal to 23 -+ 24-+ 12= 59, And since there are 67 in all, only 67 — 59= 8 workers remain who know neither language. In this institution, 8 know neither Spanish nor German. ‘We can write the answer in the form 8=67—(23-+24-+12) But we get 24 by subtracting 23 from 47, and 42 by subtracting 23 from 35. Therefore, 8=67—23— (47 — 23) — (35 —23) = 67-47 354.23 Here is the regularity: we subtract from the total number of scientific workers those who know Spanish and those who know German. In this 18 way, some find themselves in both lists and are “subtracted” twice; these are precisely the poly- glots with a knowledge of both languages. Adding on these, we obtain the number of persons who never studied either language. Let us complicate the problem by adding another language—French, Suppose there are 20 scientific workers with French as their foreign language, then 12 with Spanish and French, 44 with German and French, and 5 polyglots that handle all three languages. It is then clear that those who know Spanish and French (no Ger- man) number 12 — 5 = 7, whereas 11 —5= 6 know only German and French. This leaves 2 who read French (20 —7— 6 —5= 2). These wor- kers are among the 8 persons who have no com- mand of Spanish or German. Hence, the number of researchers here who cannot work in any one of the three languages is 8 — 2= 6. We can write down the answer as follows: 6==8—2=67—47— 35-4 23— (20—7—6— 5) 7 — 47 — 35+ 23— 20+ (12—5) 4+ (11-5) + 5 =67—47— 3520-4 23-4124 1—5 Now the law is perfectly clear. From the total number of workers we first subtract those who have a knowledge of one foreign language (and, possibly, other languages). In this process, some of them are “subtracted” twice since they know two languages. Therefore, we add the numbers 23, 12, and 11 which indicate how many employees read two languages (and, possibly, a third lan- guage). But those with three languages are rejected three times and then added three times. And since they have to be subtracted, we reject 5. THE PRINCIPLE OF INCLUSION AND EXCLUSION The foregoing examples enable us to state a general law. Let there be N objects some of which have properties a4, 2, ++, Gn. Each object can cither have none of these19 properties, or one or several. Let us de- note by NW (aj, aj, +++ o) the number of objects having properties ay, aj, .. 5 ay (and, possibly, certain other properties). If we wish to emphasize that only objects devoid of a certain property are taken, this property will be indicated by a dash. Say, N (aa20,) indicates the number of objects having proper- ties aq and a2, but devoid of property a, (the question of the other properties remains open). The number of objects not possessing a single ‘one of the indicated properties is denoted, by this rule, by N (aja, ... ay). The general law consists in the fact that N (jog «ef, = N—N (a4) —N (2)... — = (cin) +N (o404)-+N (cute) +--+ EN (@y¢tn) + +++ +N (Gtn10n)—N (ceyg0t3) — FN (Gn-20n-1%m) + ++ + + (=A)"N (atcte on) (2) Here, the algebraic sum is extended to all com- binations of properties a, a2, a (without regard for their order), the + sign heing used if the number of properties accounted for is even and the — sign if that number is odd. For inst- ance, N (ayogaects) enters with a+ sign while NV (cigaucqo) enters with a — sign. Formula (2) defines what is known as the principle of inclusion and exclusion (the inclusion and exclusion formula)’ first, all objects with at least one of the proper- ties a4, a2, .+ +; Om are excluded, then all objects having at least two of these properties are included, then those having at least three are excluded, ete. Let us prove formula (2). The proof is by induction on the number of properties. For one property the formula is obvious. Each object either has this property or does not. Therefore, N(@)=N—=N (a) Now suppose that formula (2) has been pro- ved for the case when the number of properties is equal to n—4: N (ae... fy) = NN (cy) — EN (G4ct2) + 6. EN (Gtn-2%m—1) — =N (xpetq09) — 62 = (Cen ttn2in-t) + ee (—A)POEN (aryete « nt) (3) —N Gna) By hypothesis, this formula holds true for any collection. In particular, it is true for the collec- tion V (@) of elements having the property ap. For this collection, formula (3) takes the form N (ijetg «+. Oth, 40tn) = N (en) —N (ct4tm) — TN (@nt0n) +N (cereta0tn) + «2 + EN (Cin20n-10%m) — N (cegotgetgetn) — eee b (AEN (eee. Gtn-icin) (4) (the restriction is indicated that in each case we take only those objects having the property a). Let us subtract (4) from (3). On the right we get what we desire: the right-hand side of formula (2). The left-hand member is the diffe- rence N (ahah. hy) —N (ia. 0h 40m) (8) But N (aja, ... ons) is the number of objects got possessing the properties a4, cay...) pat and, possibly, possessing the property a). Now, N (ajay... Gn-t Gm) denotes the number of objects that do not possess the properties G4, Gay ++ +y On-t Dut definitely possess the property a. This means that the difference (5) is just equal to the number of objects that do not possess a single one of the properties (a, ca, .. + +41 Gy-ty On)» In other words, N (cheesy) IN (ag. th gn) = =N (ajog... 0h yah) Thus, after subtraction, the left-hand member of formula (2) also appears on the left. The for- mula is thus proved for the case when the number of properties is equal to n. Thus, the relation (2) holds true for n proper- ties if it holds for n —1 properties (for n= 4 oeit has already been proved). It follows that the telation holds for any set of properties. Formula (2) may be represented symbolically as follows: N(@B'...0)=N (1a) (1B)... (1-0) (6) After removing the brackets, the product Naf ... A has to be written as N (aB.. . 2). For example, in place of NV afb we write V (aba). WHERE'S THE MISTAKE? 5 The class president reports as follows: “Our class has 45 students, of which 25 are boys. ‘Thirty students get marks of “good” and “excel lent” (of this number, 16 are boys). A total of 28 students go in for athletics, of which number 18 are boys and 17 are students with marks of “good” and “excellent”. Fifteen boys get good and excellent marks and also go in for athletics.” ‘A couple of days later, the home-room teacher, who, ironically, was the mathematics instructor, summoned the class president and stated that his report contained an error. Let us try to find out what the mistake was. To do this, we com- pute the number of girls who do not engage ia athletics and get passing marks (3’s) or even failing marks (2's). Denote membership in the set of boys by 4, students getting good marks (4 s and 5's) by a and those who go in for athle- tics by as. We find WV (ajaza1). By hypothesis, we have N (a4) =25, N (a) =30, N (as) =28, N (exa2) = 16, N (@4e4s) = 18, N (centes) =17, N (oegotgots) = 15 By the principle of inclusion and exclusion, we find that NN (xjexgees) = 45 —25—30—28 16-4 18-4 17— —15=—2 But the answer can only be positive! So the report definitely contains an inconsistency and is incorrect. 20 THE SIEVE OF ERATOSTHENES eres One of the greatest mysteries of mathematics is the distribution of the prime numbers among all the natural numbers. There are cases where two primes occur every other number (say, 17 and 19, 29 and 31), and then there are cases where a million composite numbers occur in an unbroken sequence. Today, mathematicians know a good deal about how many primes occur among the first N natural numbers. In these computations, there is a very useful method that goes back to the ancient Greek scholar Eratosthe- nes (third century B.C.) who lived in Alexan- dria. Eratosthenes was a versatile man engaged in a wide range of problems. He excelled in mathe- matics, astronomy and many other fields. True, such breadth led to a certain superficiality. Contemporaries referred somewhat ironically to Eratosthenes as “Beta”—always second best (the second mathematician after Euclid, the second astronomer after Hipparchus, ete.). In mathematics, Eratosthenes wanted to know how to find all the prime numbers from among the natural numbers up to V (Eratosthenes con- sidered 1 to be prime; today, mathematicians consider 1 to be a special type not belonging either to the primes or to composite numbers). Here is how he went about it. First of all, he crossed out all numbers divisible by 2 (excluding the number 2 itself). Then he took the first of the remaining numbers (namely, 3). It is clear that this number is prime. Then he crossed out all numbers divisible by 3 that follow it. The first remaining number is 5. Then he deleted all successive numbers divisible by 5, and so on. The numbers that remain after all deletions are Primes. Since writing was done on wax tablets in those days and the crossing out was done by Punching the figures, a tablet took on the appea- Tanee of a sieve. Whence the method of Eratosthe- nes for finding prime numbers came to be known as the “sieve of Eratosthenes”.Let us compute the number of numbers that remain in the first hundred after deleting by the Eratosthenes method, dividing by 2, 3, and 5. In other words: how many numbers are there among the first hundred that are not divisible by 2, 3 or 5? This problem can be solved by the principle of inclusion and exclusion. Denote by a the property of a number to be divisible by 2, by ag the property of divisibility by 3, and by aa the property of divisibility by 5. Then cya signifies that a number is divisible by 6, cys moans that it is divisible by 10, and cats, that it is divisible by 15. Finally, aa2¢3 signifies that a number is divisible by 30. Our task is to find the numbers from 4 to 100 that are not divisible either -by 2, 3 or 5, that is, such as do not possess a single one of the pro- perties a4, ct, a3. By formula (2) we have NN (aejogers) = 100—N (cx) —V (eta) —N (2s) + EN (ctgeta) +N (gets) +N (e202) —N (cxrcraces) However, in, order to find the numbers up to NV that are divisible by n, we have to divide N 24 by n and take the integral part of the resulting quotient. Therefore, N (es) =50, N (cz) = 33, N (ces) =20, N (ayo) = 16, IN (cestg) = 10, (ceg0ta) = 6, N (cycegeg) = 3 and, hence, N (cxjexgors) = 26 ‘Thus, 26 numbers between 4 and 100 are not divisible by 2, 3, or 5. These numbers remain after the first three stages in the Eratosthenes process. In addition, we will also have the num- bers 2, 3, and 5 themselves, There will thus be 32 numbers left altogether. After three stages in the elimination process of the sieve of Eratosthenes there will remain 269 numbers out of the first thousand. This fol- lows from the fact that in this case NV (4) =500, N (2) =333, N (3) =200, N (cio) = 166, N (cty%g) = 100, V (age) = N= (a4¢a043) = 33CHAPTER IL PERMUTATIONS AND COMBINATIONS We have considered some general rules for solving combinatorial problems of a broad variety of types. However, as in geometry, it is not always convenient to reduce the solution of problems to axioms; it is preferable to take advantage of theorems. It is often better to use ready-made formulas instead of solving by gene- ral rules, The point is that certain types of problems occur much more often than others. The combination types that occur in such pro- blems are called permutations and combina- tions, Special formulas have been derived for such combinatorial problems. One such formula is already familiar: at the beginning of Chapter I it was demonstrated that the number of k-per- mutations of n distinct objects with repetitions is equal to n, Let us now find out how many such permutations can be formed if we do not allow for repetitions, that is to say, if all the elements in the permutations are distinct. First let us examine the following problem. FOOTBALL CHAMPIONSHIP In the USSR football champtonship, Group One of Class A consists of 17 teams that aspire to gold, silver and bronze medals. In how many ways can the medals be distributed? This problem is solved by means of the rule of product. Any one of the 47 teams can get the gold medal. We thus have 17 possibilities. But if some one team gets the gold medal, then there remain 16 aspirants for the silver medal, since the same team cannot get both medals. After first place with the gold medal has been settled, there remain 16 possibilities for the silver medal. In the same way, the bronze medal can goto any one of the remaining 15 teams after the gold and silver medals have been awar- ded. By the rule of product we find that the medals can be distributed in 4,080 ways (17 X 16 x 15 = 4,080). 22 PERMUTATIONS WITHOUT REPETITIONS The above problem belongs to the class of combinatorial problems concerning permutations without repetitions. The general statement of the problem is this: There are n distinct objects. How many k-arran- gements can be made out of them? Two arran- gements are considered distinct if they differ even by a single element or have a different order. Such arrangements are termed permutations without repetitions and are symbolized by Ak. In building &-permutations out of n distinct things without repetitions, we have to make k choices. To start, we can choose any one of the n objects available. Once this choice has been made, the second step is to choose from the n—1 remaining objects, since the choice already made cannot be repeated (recall that in contrast to the case of permutations with repetitions there is only one thing of each type here). In the same way, for the third step we have a choice of n—2 free objects, at the fourth stage we have n—3 objects, +, at the th stoge we have n—k-+1 objects. Therefore, by the rule of product we find that the number of k-arrangements of n objects without repetitions is expressed as follows: Aban(n—1)... (n—k+1) (fy A SCIENCE CLUB Let us apply this newly derived formula to the following problem. A science club consists of 25 members. The task at hand ts to elect a presi- dent, a vice president, scientific secretary and treasurer. In how many ways can this choice be made if each member of the club can occupy only one post at a time? Here we have to find the number of permuta- tions (without repetitions) of 25 elements taken four at a time. The point is that there is a diffe- rence in who heads the club and what persons23 occupy the subsidiary posts (say, a choice of Ivanov as president, Tatarinov as vice presi- dent, Timoshenko as scientific secretary and Alekseyey as treasurer differs from Timoshenko as president, Ivanov as vice president, Tatarinov as scientific secretary and Alekseyev as treasu- rer). The answer is therefore contained in the formula Aly =25 x 24 x 23 x 22 = 303,600 PERMUTATIONS OF N ELEMENTS When constructing permutations of n ele- ments without repetitions taking k at a time, we obtained arrangements that differed both as to elements and as to their order. But if we take arrangements involving all n elements, then they can differ solely in the order of the ele- ments, Such arrangements are called permu- tations of n elements, or, briefly, n-permuta- tions. In other words, n-permutations are permuta- tions of n elements without repetitions in which all elements participate. We can also say that permutations of n elements are all possible rearrangements, each one of which contains all the elements once and all of which differ solely in the order of the elements. The number of n-permutations is denoted by P,. The formula for P, is readily obtained from the formula for the number of permutations without repetitions, Namely, Py=AR=n(n—1)...2%4 2) ‘Thus, to find out how many permutations there are of n elements, multiply together all the natu- ral numbers from 1 to n. This product is denoted by nl (read: n-factorial),-We thus have Prama1x2x... Xm We agree that 4 4. In the future we will encounter 0!, Though it might seem that O! should be equal to zero, we agree to consider 0! equal to unity (0! = 1). The point is that factorials have the obvious property n(n—)! This equality holds true for n > 4. It is natu- ral to define 0! so that it should remain true for the caso n = 4 as well, that isso that {1= 1 x Ol. But then we have to set 0! = 1. It will also be noted that formula (1), the number of permutations without repetitions, can be written as at 1 A= Gl 8) Indeed, in (8) all the factors (1, 2, 3, . 4 m— A) enter both the numerator and the denominator. Cancelling, we find that AX=n(n—1,.., - +. (1 —k + 1), in accordance with formula (1). THE PROBLEM OF THE ROOKS How many ways are there of placing 8 rooks ona chessboard so that they do not take each other? It is clear that only one rook can occupy each rank (row) and file (column). Let us take one such position and use a, to denote the number of the ‘occupied square on the first rank, a, to denote the position on the second rank, . . ., ag the position on the eighth rank. Then (a1, a2, - « «, a3) will be a certain permutation of the numbers 4, 2, ..., 8 (it is clear that among the numbers 4, az, ++, ag there is no pair alike, for then two rooks would be occupying the same file). Conversely, if a1, dz, «+, ay is a certain per- mutation of the numbers 1, 2,..., 8, then to it corresponds a certain nontaking arrangement of the rooks. Fig. 6 shows one arrangement of the rooks corresponding to the permutation 75461328, Thus, the number of desired arrangements of rooks is equal to the number of permutations of the numbers 1, 2,..., 8,24 which is Ps. But Py=81=1.2-3-4-5-6-7-8— 40,320 ‘These rooks can therefore be positioned as requi- red in 40,320 ways. The same proof is used to demonstrate that there are n\ ways of positioning n nontaking rooks on an n by n chessboard. a Fig. 6. ‘The result would be quite different if the rooks differed in some way, say, by colour or by some other label. Suppose the rooks are numbered. ‘Then from each arrangement of non-numbered rooks wo get nl arrangemonts of numbered rooks. They are obtained by permuting the n rooks in all possible ways for the samo occupied squares. There would then be (nl)? ways of positioning nontaking rooks. We also arrive at the same conclusion by applying the rule of product directly. The first rook can be placed on any one of n* squares. Crossing out the rank and file of this rook, we have an n— 1 by n—1 board left (n — 4 ranks and n —1 files) with a total of (n — 4) * squares. This means that the second rook can be placed in (n—4)® ways. In the same manner, the third rook can be positioned in (n — 2)? ways, etc. Altogether we have ne (nt)... 2=(al)? ways of placing the rooks. LINGUISTIC PROBLEMS One of the problems of linguistics is decipher- ing inscriptions in unknown languages. Suppose we havea text written with the aid of 26 unknown signs, each one depicting one of 26 sounds. In how many ways can the sounds be correlated with the written symbols? Let us arrange the written symbols in a certain order. Then each correlation will yield a certain permutation of sounds. Using 26 sounds, we can form P2g= 26! permutations. This number is roughly equal to 4 x 10%, It is clearly impossible to verify all these possibilities. It is even beyond the capabilities of an electronic computer. Attempts are therefore made to reduce the number of possibilities. It is frequently possible to separate the vowels from the consonants (one more often encounters vowel-consonant combina- tions than vowel-vowel or consonant-consonant combinations; observing combinations of sym- bols that occur more often than others, it is possible to separate the vowels from the conso- nants). Suppose we have been able to find 7 sym- bols for vowels and 19 symbols for consonants: Let us now find out how many times the number of possibilities has been reduced. The 7 vowels can be permuted in 7! ways and the 19 consonants in 191 ways. The total number of combinations is equal to 7! x 491. Which means the work has been reduced 261/7! x 19! = 650,000 times. This is of course a gain, but 7! x 49! is still an enormous number. The next step is to compute the frequency of ‘occurrence of the individual symbols. Comparing this frequency with that of letters in related languages, one can conjecture as to the signifi- cance of certain symbols. Other symbols are found by comparing the given text with the same text in another language (the kings of old delighted in describing their heroic deeds in several languages). Suppose that in this way 4 vowels and 13 con- sonants have een identified. What are the remain-25 ing possibilities? Obviously, 31 - 6! = 4,320. Now this number of combinations can be handled by electronic computers. Cryptologists meet similar difficulties in de- coding operations. ROUND DANCE Seven girls form for a round dance, In how many different ways can they stand tn the circle? If they stood in one place, the result would be 7!= 5,040 permutations. But since the dancing girls circle all the time, their position relative to surrounding objects is not essential, the only important thing being their mutual arrangement. For this reason, we consider the permutations appearing as the dancers move round the circle to be the same. However, from each permutation we can obtain another six by means of rotation, so the number 5,040 has to be divided by 7. This yields 5,040: 7= 720 distinct permutations of girls in a round dance. Fig. 7. Generally, if we consider permutations of n objects in a circular arrangement and regard as identical those positions that go into one another in a rotation, then the number of distinet permu- tations is equal to (n— 1)!. Now let's see how many necklaces we can make out of 7 distinct beads. By analogy with the problem that was just solved, it might be imagined that the number of distinct necklaces is just 720. But a necklace can also be turned upside down in addition to being moved round the cirele (Fig. 7). So the answer is 720 : 2— 360. PERMUTATIONS WITH REPETITIONS Up to now we have been permuting objects that are pairwise distinct. But if some of the objects being permuted are identical, then wo obtain fewer permutations, some of them coincid- ing with one another. Let us illustrate with the word “part”; permut- ing, we obtain 24 different permutations: part rapt pra rtpa patr ptar ratp rtap prat rpat rpta prta trap tpar aptr artp tarp atrp atpr —tapr aprt arpt trpa_tpra If we take the word “papa” instead, then in all the permutations written out above we will have to replace “r” by “p” and “t” by “a”. Some of our 24 permutations will be the same, For example, in the first row of permutations, we get one word, “papa”. All four permutations in the second row give “paap”. Generally, all 24 per- mutations split up into foursomes, in which the replacements (“r” by “p”, “t” by “a”) all yield the same result. In the table, these permutations stand in one row. Therefore, the number of dis- tinct permutations that can be produced using the word “papa” is 24:4= 6. They are Papa, Paap, ppaa, apap, aapp, appa Stated generally, we have the following pro- blem: There are k elements of different types. How many permutations can be made out of n, elements of the first type, nz elements of the second type, . . ..ny elements of the kth type? The number of elements in each permutation is equal to n= my-+ net... -+tnq. Thus, if