Sets, Relations and Functions

UNIT 1 SETS, RELATIONS AND FUNCTIONS

Structure
1.0 1.1
1.2

Page No.
5 5 5 9 13

1.3 1.4

Introduction Objectives Introducing Sets Operations on Sets

1.3.1 1.3.2 1.4.1 1.4.2 1.4.3 Basic Operations Properties Common to Logic and Sets Cartesian Product Relations and their types Properties of Relations

Relations

1.5 1.6 1.7

Functions
1.5.1

16 22 22

1.5.2

Types of Functions Operations on Functions

Summary Solutions / Answers

1.0

INTRODUCTION

In common parlance, we find people using the words given in the title of this unit. Do they have the same meaning in mathematics? Youll find this out by studying this unit. You will also see how basic the concept of set and function or to any area of mathematics and subjects depend on mathematics. In this unit we will begin by introducing you to various kinds of sets. You will also study operations like, union and intersection. While doing so you will see in what way Venn diagrams are a useful tool for understanding and working with sets. Next we will discuss what a relation is, and expose you to some important types of relations. You will come across while studying banking, engineering, information technology and computer science, of course mathematics. As you will see in your study of computer science, an extensive use of functions is made in problem-solving. Finally, we lead you detailed discussion of functions. Over here we particularly focus on various points of functions and fundamental operations on functions.

1.1 OBJECTIVES
After studying this unit, you should be able to: explain what a set , a relation or a function is give examples and non-examples of sets, relations and functions perform different operations on sets establish relationships between operations on sets and those on statements in logic use Venn diagrams explain the difference between a relation and a function. describe different types of relations and functions. define and perform the four basic operations on functions

1.2 INTRODUCING SETS

In our daily life we encounter collections, like the collection of coins of various countries, a collection of good students in a class, a collection of faculty members of IGNOU, etc. In the first of these examples, it is easy for anybody anywhere to tell

Basic Combinatorics

whether an object belongs to this collection or not. If we take the collection of coins of a country, then a coin will be in the collection if it is a coin of that country, not otherwise. The criterion for being a member of the collection is objective and clear. However, if we take the collection of all good students, it is very difficult to say whether a person belongs to this collection or not because the characteristic good is not very clearly defined. In this case the collection is not well-defined, while the previous collection is well-defined. Similarly, the collection of all the IGNOU students is well-defined. Definition: A set is a well-defined collection. The objects belonging to a set are called elements or members of that set. We write the elements of a set within curly brackets. For instance, consider the set A of stationary items used by Nazia. We write this as A = {pen, pencil, eraser, sharpener, paper} Another example is the set B = {Lucknow, Patna, Bhopal, Itanagar, Shillong} of the capitals of 5 states of India. Note that A and B are well-defined collections. However, the collection of short people is not well-defined, and therefore, it is not a set. Also note that the elements of a set dont have to appear similar. For example, {pen,Lucknow,4} is a set consisting of 3 clearly defined elements. As you have seen, we usually, denote sets by capital letters of the English alphabet. We usually denote the elements by small letters a,b,x,y . If x is an element of a set A, we write this as xA (read as x belongs to A). If x is not an element of A, we write this as x A (read as x does not belong to A). There are three ways of representing sets: Set-builder form, Tabular form and the pictorial representation through Venn diagrams. In the Set-builder form, or property method of representation of sets, we write between brackets { } a variable x,which stands for each of the elements of the set which have the properties p(x), and separate x and p(x) by a symbol : or | (read as such that). So the set looks like {x: p(x)} or {x | p(x)}. For instance, the set {x | x is a white flower} is the set of all white flowers, or {x: x is a natural number and 2<x<11}is the set of natural numbers lying between 2 and 11. In Tabular form, or the listing method, the elements of a set are listed one by one within the brackets { }, each separated from the other by a comma, as in the examples A and B given above. The accepted convention for writing a set by the listing method is that elements will not be repeated. For example, in the set A= {4,2,8,2,6}, 2 is repeated, which is not necessary.So we will write A ={4,2,8,6}. We shall introduce you to Venn diagrams a little later. For now, let us consider a few more sets. Definition: A set with no element is called the empty(or null, or void) set, and is denoted by or{}. For example A = {x:x is an integer between 13 and 17 which is divisible by 6}, has no element, i.e., A is the empty set. Definition: A set having a finite number of elements is called a finite set.

Sets, Relations and Functions

For example, {1,2,4,6} is a finite set because it has four elements, , the null set, is also a finite set because it has zero number of elements; the set of stars in the sky is also a finite set. Definition: A set having infinitely many elements is called an infinite set. For example, the set N of natural numbers is infinite. Similarly, Q , Z, R and C, the set of rational numbers and complex numbers, respectively, are infinite set. B= The set of all strengthness in a given plane. Now try the following exercises. E1) E2) How would you represent the set of all students who have offered the IGNOU course? Explain, with reason, whether or not i) the collection of all good teachers is a set ii) the set of points on a line is finite. E3) Represent the set of all integers by the listing method.

When we deal with several sets, we need to understand the nature of the elements of those sets, whether the elements of two given sets have some elements in common or not, and so on. These questions involve concepts, which we now define. Definition: A set A is said to be a subset of a set B if each element of A is also an element of B. In this case B is called a superset of A. If A is a subset of B, we represent this by A B. As a statement in logic we represent this situation as, A B [x A x B] B contains A or B is a superset of A is represented by B A. If A is not a subset of B, we represent this by A B. For example, if A ={4,5,6} and B ={4,5,7,8,6}, then A B. But if C = {3,4} then C B. Remark: If A B and B C, then A C. Definition: Two sets A and B are equal if every element of A belongs to B and every element of B belongs to A. We represent this by A=B. For example, if A= {1,2,3}, B ={2,3,1}, then A B and B A, so that A = B. Definition: A set A is said to be a proper subset of a set B if A is a subset of B and A and B are not equal. We represent this by A B. For example, if A ={4,5,6} and B = {4,5,7,8,6}, then A B; and if A={Java, C, C++, Cobol} and B={Java, C++}, then A B. Note: A set can have many subsets and many supersets. For example A={1,2,3,4,5}, B= {2,3,4,5,6,7}, and C= {2,3}, then for C, A and B can be used as supersets. Similarly, if X= {Ram, Rani, Sita, Gita}, Y= {Rani}, and Z= {Sita}, then Y and Z both are subsets of X. Definition: The power set of a set A is the set of all the subsets of A, and is denoted by P(A).
(1834 -1923) Fig 1: John Venn

Basic Combinatorics

Mathematically, P(A) = { x:x A}. Note that P(A) and A P(A) for all sets A. For example, if A={1}, then P(A)= { ,{1}} and if A= {1,2}, then P(A)={ , {1}, {2}, {1,2}} Similarly, if A = {1,2,3}, then P(A) = { ,{1}, {2}, {3}, {1,2}, {1,3}, {2,3}, {1,2,3}}. Definition: Any set which is a superset of all the sets under consideration is known as the universal set. This is usually denoted by , S or U. For example, if A = {1,2,3}, B = {3,4,6,9} and C = {0,1}, then we can take U = {0,1,2,3,4,5,6,7,8,9} or U= N, or U=Z as the universal set. Note that the universal set can be chosen arbitrarily for a given problem. But once chosen, it is fixed for the discussion of that problem. Theorem 1: If A is a set with n elements, then |P(A)| = 2n. Proof: We shall prove this by mathematical induction. For this, we first check if it is true for n= 1.Then assuming that it is true for n = m, we prove it for the case n = m +1. It will, then, follow that the result will be true n N. Step I: If A = 1, then P(A) = 2=21. Step II: Assume that the theorem holds for all sets A of cardinality k, i.e. if A = k, then A has 2k subsets. Step III: Now consider any set A = {x1, x2, x3,,xk, xk+1}, with k+1 elements. Consider its subset B= {x1, x2, x3,,xk }. Now B has 2k subsets, each being a subset of A. Now, take any such subset {x i1 ,x i 2 ,,x i r }of B. Then {x i1 ,x i 2 ,,x i r ,x k+1}

is a subset of A that is not a subset of B. So, for each of the 2k subsets of B, we attach xk+1 to it to get 2k more subsets of A. You can see that this covers all the subsets of A. So the number of subsets of A=2k +2k = 2.2k = 2k+1 . Hence the theorem. Now try these exercises. E4) E5) E6) E7) E8) Give two proper subsets and two supersets of the set of vowels of the English alphabet. Find the power set of the set A ={a,e,i,o,u}. For which set A, is P(A) = 1 ? If A B, is P(A) P(B)? Why? P(A) = P(B) A = B. True or false? Why?

Let us conclude this section with the pictorial representation of sets. You know that the pictorial representation of any object helps in understanding the object. This is why a pictorial representation of sets, known as a Venn diagram, helps in understanding and dealing with sets.

Sets, Relations and Functions

The English priest and logician John Venn invented the Venn diagram. Through Venn diagrams we can easily visualize the abstract concept of a set and operations on sets. In this diagram, the universal set is usually represented by a rectangle and its subsets are shown as circles or other closed geometrical figures inside this rectangle. For example, A = {Lucknow, Patna, Bhopal, Itanagar, Shillong} can be represented using a Venn diagram as in Fig. 2. Here U could be any superset of A.
*Lucknow, *Patna, *Bhopal, *Itanagar, *Shillong A
Fig. 2: A Venn diagram

Now that you are familiar with basic definitions related to sets, let us discuss some basic operations that can be performed on sets. This is when we shall appeal to Venn diagrams very often, as you will see.

1.3 OPERATIONS ON SETS

Let us now study sets obtained by applying operations on sets. We will cover four operations here, namely, union of sets, intersection of sets, complement of sets and symmetric difference. While studying them you will see how useful a Venn diagram can be for proving results related to these operations. In this section we will also look at some rules that are common to operations on sets and operations on statements, which you studied in Block 1.

1.3.1 Basic Operations

In this sub-section we shall define each of the operations one by one. Definition: The union of two sets A and B is the set of all those elements which are either in A or in B or in both A and B. This set is denoted by A B, and read as A union B. Symbolically, A B= {x:x A or x B} For example, if A = {x: x is a stamp} and B = {4,5}, then A B= {x:x is a stamp or a natural number lying between 3 and 6}. And A={Ram, Mohan, Ravi} and B={Ravi, Rita, Neetu}, then A B={Ram, Mohan, Ravi, Rita, Neetu}. If A B, then A B = B, and vice versa. This can be shown immediately using a Venn diagram, as in Fig.3.(a), where A is shown as the square contained in the circle representing B. In Fig.3(b), A B is shown when A and B have some elements in common, and in Fig.3(c), we depict A B when A and B have no element in common.
U

U B A

Basic Combinatorics

(a)

(b) Fig. 3: Venn diagram for union

(c)

Definition: The intersection of sets A and B is the set of all the elements which are common to both A and B. This set is denoted by A B, and read as A intersection B. Symbolically, A B = {x :x A and x B}; For example A = {1,2,3} and B = {2,1,5,6}, then A B ={1,2}. Again if A= {1} and B = {5} then A B = { } or . Remark: For any two sets A and B, A B A A B and A B B A B. What is A B if A B? Do you agree that it is A? Let us use a Venn diagram to check this (see Fig.4(a)). If A and B have some elements in common, then the Venn diagram for A B looks like Fig 4.(b), and if A and B have no element in common, then the Venn diagram will be as in Fig.4(c).
U B A A B A A B B U

(a)

(b) Fig. 4: Venn diagram for intersection of sets

(c)

Definition: The difference of two sets A and B is the set of all those elements of A which are not elements of B. Sometimes, we call this set the relative component of B in A. It is denoted by A~B or A\B, and is read as A complement B. Symbolically, A~B = {x : x A and x B} and B ~A = {x : x B and x A} For example, if A = {4,5,6,7,8,9} and B = {3,5,2,7}, then A ~ B = {4,6,8,9} and B ~ A = {3,2}. From this example it is clear that A ~ B B ~ A. In fact, this is usually the case. So, the difference of sets is not a commutative operation. In Fig.5(a), A B, so that A~B= . In Fig.5(b) we show A~B when A B, and in Fig.5(c) we show A~B when neither A B nor B A. In Fig. 5(d), we show A~B when A and B are disjoint.

U B A B A

(b)

A~B =

10

Sets, Relations and Functions (a)

U

(b)
U

(c)
A B

(d)
B

(c)

(d)

Fig. 5: A ~ B in different situation is the shaded portion.

There is one particular difference that shows up very often, which we now define. Definition: The complement of a set A, is the set U \ A, and is denoted by A or Ac. For example, U={Physics, Chemistry, Mathematics} and A={Mathematics}, then the complement of A is A= {Physics, Chemistry}. The Venn diagram showing the complement of is the set of those elements of the universal set U which are outside A (see Fig.6).
A
U

Fig. 6: Venn diagram for A.

Definition: The symmetric difference of two sets A and B is the set of all those elements which are in A or in B, but not in both. It is denoted by A B. i.e., A B =(A~B) (B~A). Note that A B = B A, i.e. the symmetric difference is commutative.

For example A= {1,2,3,4,5} and B= {3,5,6,7}, then A~ B= {1,2,4},and B~A = {6,7} A B={A~B) (B~A) = {1,2,4,6,7}

Now you may try these exercises. E 9) Make a Venn diagram for A B for each of the situations i) A B, ii) A iii) B A and A B ; iv) A B= . B,

E10) Let A= {Math, Physics, Science}, B={Computer,Math,Chemistry}, C={Math}. Find A (B C). E11) If A = {1,2,3,4,5,6}, B={4,5,6,7,8,9},find i) A~ B, ii) B~ A, iii) A B. E12) For which sets A and B would A~B = B~A ? E13) Write a program in C to perform E 10. E14) Under what conditions can A B = A B ?

11

Basic Combinatorics

While discussing these operations, you may be wondering that they seem to satisfy properties very similar to those of propositional logic covered in Block 1 of this course.You are right! Let us look at this aspects now.

1.3.2 Properties Common to Logic and Sets

Before looking into the properties we shall first present a very useful principle to you. This will allow you to see how one property can be proved in several situations simultaneously. Duality Principle: The duality principle is very convenient for dealing with theorems about sets. Basically if any theorem is given to you, by applying the duality principle you can get another theorem (the dual of the previous one). In any statement involving the union and intersection of sets, we can get from one of the statements to the other by interchanging with and with U. For example, the dual of A (B C) is A (B C) and the dual of U =U is U = . So, for example what is true for A (B C) will be true for A (B C) too. This is why if the first property in each of the pairs below is proved the second one follows immediately. For any universal set U and subsets A, B and C of U, the following properties hold. i) Associative properties: Union: A (B C)= (A B) C Intersection: A (B C) = (A B) C

ii) Commutative properties: Union: A B= B A Intersection: A B = B A. iii) Identity: Union: A = A Intersection: A U=A. iv) Complement: Union: A A = U Intersection: A A = v) Distributive properties: Union: A (B C) = (A B) (A C) Intersection: A (B C) = (A B) (A C) De Morgans Laws: For any two sets A and B the following laws known as De Morgans laws, hold
Fig. 7: Augustus De Morgan (18061871)

1. (A B) = A B , and 2. (A B) = A B De Morgans laws can also be expressed as 1. A ~ (B C) = (A~ B) (A~ C) 2. A ~ (B C)= (A~ B) (A~ C) Each of the properties above corresponds to a related property for mathematical statements in logic (which we have covered in Unit 2 and Unit 3 of Block 1 of this course). Now try these exercises.

Fig. 7: Augustus De Morgan (1806-1871)

12

Sets, Relations and Functions

E15) Find the dual of i) A (B C) = (A B) C, and ii) (A B) (A C). E16) Draw a Venn diagram to represent A (B C). E17) Check whether (A B) C = A (B C) or not using a Venn diagram. Let us now focus on subsets of a particular kind of product of sets.

1.4 RELATIONS
Sometimes we need to establish relations between two or more sets. For example, a software development company has a set of specialists in different technology domains, or a company gets some projects to develop. Here the company needs to establish a relation between professionals and the project in which they will participate. To solve this type of problem the following concepts are required.

1.4.1 Cartesian product

Very often we deal with several sets at a time, and we need to study their combined action. For instance, combinations of a set of teachers and a set of students. In such a situation we can take a product of these sets to handle them simultaneously. To understand this product let us first consider the following definitions. Definition: An ordered pair, usually denoted by (x,y), is a pair of elements x and y of some sets. This is ordered in the sense that (x,y) (y,x) whenever x y,that is , the order of placing of the element in the pair matters. Any two ordered pairs (x,y) and ( a,b) are equal iff x = a and y = b. For example if, A= {a,b,c}and B= {x,y,z},then A B = {a,b,c} {x,y,z} = {(a,x), (a,y),(a,z),(b,x),(b,y),(b,z),(c,x),(c,y), (c,z)},and B A = {x,y,z} {a,b,c} = {(x,a),(x,b),(x,c),(y,a),(y,b),(y,c),(z,a),(z,b),(z,c)}. Now let us think about how B A can be represented geometrically? For instance what is the geometric view of {2} R ? This is the line x =2 given in Fig.8(a). Now, after seeing geometric representation of {2} R, can you tell what {1,3}{2,3} = {(1,2),(3,2),(1,3),(3,3) }looks like? You will get four points in the first quadrant, as shown in Fig.8(b).

iff is short for if and only if.

Now, you know that the Fig. 8(a): {2} R, i.e., x = 2. of numbers is commutative. Is (1,2) (2,1). So, AB BA usually.

Fig.8(b): {1,3}{2,3}

multiplication the Cartesian

product of sets also commutative? For instance, is {1}{2}={2}{1}? No, because We can extend the definition of A B to define the Cartesian product of n sets A1, A2, .,An as follows. A1 A2A3 ,An= {(x1,x2,x3,,xn)}: x1 A1, x2 A2 x3 A3 , xn An}.

13

Basic Combinatorics

The element (x1,x2,,xn) is called an n-tuple. For instance, the 3tuple (1,1,3) {1} {1,2} {2,3}. Now you may try some exercises. E18) If X= {a,b,c} and Y={1,2,3},find i) X X , ii) X Y, and iii) X . E19) Under what conditions on A and B is AB = BA ? E20) Give the geometric representation of R {2}. With what you studied in this sub-section, you now have the background to discuss relations.

1.4.2 Relations and their Types

We often speak of relations which hold between two or more objects, e.g., discrete mathematics is one of the courses in the IGNOU MCA Ist semester, Nehru wrote Freedom of India, Chennai is the capital of Tamil Nadu. These are the relations in everyday situations. In these examples some sort of connections between pairs of objects are shown, and hence they express a relation between the pairs of objects. Definition: A relation between two sets A and B is a subset of A B. Any subset of A A is a relation on the set A. For instance, if A ={1,2,3} and B = {p,q}, then the subset {(1,p),(2,q),(2,p)} is a relation on A B. And {(1,1), (2,3)} is a relation on A. Also, R= {(x,y) N N : x > y} is a relation on N, the set of natural numbers, since R N N. If R A B,. we write x Ry if and only if (x,y) R( xRy is read as x is related to y). Theorem 2: The total number of distinct relations between a finite set A and a finite set B is 2mn, where m and n are the number of elements in A and B, respectively. For example, R1 = N L, where L is set of straight lines, in this relation we can give different ordering of the straight lines. If the relation R2 = {1,2,3} { l1,l2}, then line l1 and l2 can get three different ordering. Proof: The number of elements of A B is mn. Therefore, the number of elements of the power set of A B is 2mn (See Theorem 1). Thus, A B has 2mn different subsets. Now every subset of A B is a relation from A to B, by definition. Hence the number of different relations from A to B is 2mn. As you have seen, any and every subset of AA is a relation on A. However, some relations have special properties. Let us consider these types one by one.

1.4.3 Properties of Relations

Reflexive Relations: A relation R on a set A is called a reflexive relation if (a,a) R a A. In other words, R is reflexive if every element in A is related to itself. Thus, R is not reflexive if there is at least one element a A such that (a,a) R.

14

Sets, Relations and Functions

For example, if A= {1,2,3,4}, then the relation R1= {(1,1), (2,2), (3,3), (4,4)} in A is reflexive because for x A,(x,x) R1. However, R2= {(1,1), (2,1), (4,4)} is not reflexive since 2 A, but (2,2) R2. Symmetric Relations: A relation R on a set A is called a symmetric relation if (a,b) R (b,a) R.Thus, R is symmetric if bRa holds whenever aRb holds. A relation R in a set A is not symmetric if there exist two distinct elements a, b A, such that aRb, but not bRa. For example, if L is the set of all straight lines in a plane, then the relation R in L, defined by x is parallel to y, is symmetric, since if a straight line a is parallel to a straight line b, then b is also parallel to a. Thus, (a,b) R (b,a) R. However, if R is the relation on N defined by xRy iff xy>0, then R is not symmetric, since, 42>0 but 24 0.Thus, (4,2) R but (2,4) R. Transitive Relations: A relation R on a set A is called a transitive relation if whenever (a,b) R and (b,c) R, then (a,c) R for a,b,c A . Thus, [(a,b), R,(b,c) R (a,c) R] , a,b,c A R is transitive. A relation R in a set A is not transitive if there exist elements a,b,c A, not necessarily distinct, such that (a,b) R, (b,c) R but (a,c) R. For example, if L is the set of all straight lines in a plane and R is the relation on L defined by x is parallel to y then if a is parallel to b and b is parallel to c, then a is parallel to c. Hence R is transitive. However, the relation xSyon L defined by iff x intersects y is not transitive. Also, the relation R on A, the set of all Indians, defined by xRy iff x loves y, is not a transitive relation. Equivalence Relations: A relation R on a set A is called an equivalence relation if and only if (i) (ii) R is reflexive, i.e., for all a R, (a, a) R, R is symmetric, i.e., (a, b) R (b, a) R, for all a, b A, and

(iii) R is transitive, i.e., (a, b) R and (b, c) R (a, c) R, for all a, b, c A. One of the most trivial examples of an equivalence relation is that of equality. For any elements a,b,c in a set A, (i) (ii) a = a, i.e., reflexivity a = b b = a, i.e., symmetricity

(iii) a = b and b = c a = c , i.e., transitivity. Now let us see if xRy iff x y gives an equivalence relation on R. (i) (ii) x x, i.e., (x,x) R, i.e., R is reflexive. However, 2 3 but 3 2. So, R is not symmetric.

Thus, R is not an equivalence relation. Now you may try these exercises. E 21) Let A be the set of all people on Earth. A relation R is defined on the set A by aRb if and only if a loves b for a,b A.

15

Basic Combinatorics

Examine if R is i) reflexive, ii) symmetric, iii) transitive. Now we shall study a particular kind of relation, which is very useful in mathematics, as well as in computer science, as you will soon see.

1.5 FUNCTIONS
A function is a special kind of relation. If we take the example of the set A of students of IGNOU, and the set B of their enrolment numbers. Now consider R = {(a,b) AxB | b is enrollment number of a },this is a relation between A and B. It is a special relation, special because to each a A ! b such that aRb. We call such a relation a function from A to B. Let us define this term formally. Definition: A function from a non-empty set A to a non-empty set B is a subset R of AxB such that for each a A a unique b B such that (a,b) R. So, this relation satisfies the following two conditions: (i) for each a A, there is some b B such that (a,b) R (ii) if (a,b) R and (a,b) R then b = b. We usually present functions as a rule associating elements of one set with another. So, let us present the definition again, with this view. Definition: Let A and B be non-empty sets. A function (or a mapping) f from A to B is a rule that assigns to each element x in A exactly one element y in B. We write this as f: A B, read it as f is a function from A to B. Note that ( i ) to each a A, f assigns an element of B; and ( ii) to each a A, f assigns only one element of B. So, for example, suppose A ={1,2,3}, B= {1,4,9,11} and f assigns to each member in A its square values. Then f is a function from A to B. But if A={1,2,3,4}, B={1,4,9,10} and f is the same rule, then f is not a function from A to B since no member of B is assigned to the element 4 in A. Note that the former example, 11 B, but there is no element in A which is assigned to 11. This does not matter. It is not necessary that every element of B be related to some element of A. Functions are not restricted to sets of numbers only. For instance, let A be the set of mothers and B be the set of human beings. Then the rule that assigns to every mother her eldest child is a function. But the rule that assigns to each mother her children is not a function because it does not relate a unique element of B to each element of A. Now, given a function, we have certain sets and terms that are associated with it. Let us give them some names. Definitions: Let f be a function from A to B. The set A is called the domain of the function f and B is called the co-domain of f. The set {f(x)| x A} is called the

range of f, and is also denoted by f(A).

Given an element x A, the unique element of B to which the function f associates, it is denoted by f (x) and is called the f-image (or image) of x or the value of the function f for x. We also say that f maps x to f(x). The element x is referred to as the pre-image of f (x).

16

Sets, Relations and Functions

For example, if A = {1,2,3,4}, B = {1,8,27,64,125}, and the rule f assigns to each member in A its cube, then f is a function from A to B. The domain of f is A, its codomain is B and its range is {1,8,27,64}. Can you tell what will be the domain and codomain for rule f : f(x) = You can see that 1x = 0, if x = 1, in this case f(x) will be undefined. Domain of f can be taken as R~{1}and codomain can be R. Remark: Each element of A has a unique image, and each element of B need not appear as the image of an element in A. Further, more than one element of A can have the same image in B. Let us look at some examples of functions, and non-functions now. i) If b is be a fixed element of B, then f: AB : f(x)= b x A is called a constant function. Note that if b=0, then f is called the zero map, and is denoted by 0. ii) f : A A : f(x) = x x A is called the identity function , and is denoted by I. iii) Consider A= {1,2,3,4}, B= {1,4,5} and the rule f which associates 11, 2 4, 35, 45. Then f is a function from A to B. A B f
1. 1. 4. 5. 6. 7. 2. 3. 4.

x ? 1 x

Fig.11: The rule f is a function

iv) The function f from R to Z, defined by the rule that f maps any real number x to the greatest integer less than or equal to x. is known as the greatest integer function or the floor function. We denote this functions action by f(x)=[x], where [x] is the greatest integer x. For example, if x = 0.6 then f(x) =[x]=0, if x= 2.3 then f(x)=[x]=2, and if x = 5, then [x]= 5 . v) Function f: RR: f(x)=xis known as the modulus (or absolute value) function, where xis the absolute value of x. For example, if x=10 then f(x)=x=10 and if x= 10, then f(x)=|x|=10. vi) Now take, A={a,b,c} and B={1,2,3,4,5}. Consider the rule f which associates a1, a 3, b2, c3. This is not a function from A to B because, elements 1 and 3 B are assigned to the same element a A. A
a b c

B f
1. 2. 3. 4. 5.

Fig.9: The rule f is not a function

17

Basic Combinatorics

vii) Consider A= {1,2,3}, B= {1,4,5,6,7} and the rule f which associates 11, 2 1, 24. Here f is not a function from A to B since no member of B is assigned to the element 3 A. A
1 2 3

B
1. 4. 5. 6. 7.

Fig.10: The rule is not a function

Now you may try these exercises. E22) Let A= {a,b,c,d}, B= {1,2,3} and R= {(a,2), (b,1), (c,2), (d,1)}. Is R a function? Why. E23) Every function is a relation. Is every relation a function? Why? E24) Consider the following pseudocode. 1. 2. 3. 4. 5. 6. read(n) while n >1 do begin if n is even then n := n div 2 else n := 2n + 1; end

Write a function of n that describes the operations performed. E25) If A= {1,2,3,4}, B = {2,3,4,5,6,7} and the rule f assigned to each member in A is f(x)= x + 1, then find the domain and range of f. Now let us discuss some types of functions.

1.5.1 Types of Functions

Here we shall look at different types of mappings.
Surjective comes from the French word sur, meaning on top of .

Onto Mapping: A mapping f: A B is said to an onto (or surjective) mapping if f(A)= B, that is, the range and co-domain coincide. In this case we say that f maps A onto B. For example, f: Z Z : f(x) = x+1, x Z, then every element y in the co-domain Z has a pre-image y1 in the domain Z. Therefore, f(Z) = Z ,and f is an onto mapping. Injective Mapping: A mapping f: A B is said to be injective (or one-one) if the images of distinct elements of A under f are distinct, i.e., if x1 x2 in A, then f(x1) f(x2) in B. This is briefly denoted by saying f is 11. For example f: R R be defined by f(x) = 2x+1, x R, then for x1 ,x2 R (x1 x2) we have f(x1) f(x2). So, f is 11. Bijective Mapping: A mapping f: A B is said to be bijective (or one-one onto,) if f is both injective and surjective, i.e., one-one as well as onto. For example, f: Z Z : f(x) = x+2, x Z is both injective and surjective. So, f is bijective.

18

Sets, Relations and Functions

There is a particular kind of bijective function that we use very often. Let us define this. Definition: A bijective mapping f : A A is said to be a permutation on the set A. Let A = {a1,a2,,an}, and f be a bijection from A onto A that maps ai to f(ai), then we write f as f=

a1 a2 . . . an . So, the identity mapping I= f(a1) f(a2)...f(an)

a1 a2...an a1 a2...an .

Now, associated with a bijective function, we get another function very naturally, which we now define. Definition: Let f: A B be a bijective mapping. Then the mapping g : B A which associates to each element b B the unique element a A, such that f(a) = b, is called the inverse mapping of the mapping f: A B. We denote this function g by f1. Note that a function f is invertible iff f 1 exists iff f is bijective. Hence, if f: AB is a one-one onto mapping, then f -1 : BA exists, and is also 1-to-1. Note the inverse of the permutation f =

a1 a2...an is the permutation b1 b2...bn

b1 b2...bn a1 a2...an .
For example, A= R~{3} and B= R~(1}, and the function f: A B is defined by f(x)=

x2 . x 3

f inverse exists.

We can see that f is a onetoone function.

To get f 1(x) the following steps are required; 1. Replace f(x) by y in the equation describing the function. You will get y=

x2 . x 3

2. Interchange x and y. In other words, replace every x by y, and vice versa. You will get x = 3. Solve for y. 4. Replace y by f- 1(x). By applying these steps we get f 1 (x)= Now try these exercises. E26) Explain why f : Z Z: f(x) = x2 is onto? Domain and range of f is Z. E27) Which of the following kind of function would you use to provide photo identity numbers? Why? i) Constant function, ii) onetoone function, and iii) identity function. E28) Find f inverse of rule f : f(x) = x33.

y2 . y3

3x 2 . x 1

19

Basic Combinatorics

Now we can see how different operations like addition, subtraction, multiplication and division can be applied on functions.

1.5.2 Operations on Functions

If given whose domains ranges are subsets of the real numbers, we define the function f+g by (f + g)(x) to be the function whose value at x is the sum of f(x) and g(x). Symbolically, (f + g)(x) = f(x) + g(x). This is called pointwise addition of f and g. The domain of f+g is the intersection of the domains of f and g since to compute (f + g)(x) it is necessary and sufficient to compute both f(x) and g(x). Other operations on functions are defined similarly:

(fg)(x) = f(x)g(x) (pointwise multiplication) f p (x) = (f(x))p for any real exponent p with the domain of f p consisting of those points for which the p-th power of f(x) makes sense. (f/g)(x) = f(x)/g(x), for g(x) 0 (pointwise multiplication)

For example, if f(x) = 3 sin (x) and g(x) = x2, then (f+g)(x) = 3 sin(x) + x2 (fg)(x) = 3 sin(x)*x2 (f-g)(x) = 3 sin(x) x2 (f/g)(x) = 3 sin(x) / x2 The domains of both f and g are all real numbers, but the domain of f/g is { x | x 0}. Now let us consider two functions f and g from A= {1,2} to, B= {1,2,3,4}, where f={(1,1),(2,4)}. Let g be defined by the rule g(x)= x2 where the domain of g is the set {1,2}. Here both have the same domain. Since f and g assign the same image to each element in the domain, they have the same effect throughout. This is why we treat them as the same, or equal. Definition: If f and g are two functions defined on the same domain A and if f(a)= g(a) for every a A, then the functions f and g are equal, i.e., f = g. For example f(x) = x2+5, where x is a real number, and g(x) = x2+5, where x is a complex number. Then the function f is not equal to the function g since they have different domains although f(x)= x2+5= g(x x R). By this example we can conclude that even if f(a)= g(a), f and g may not be the same. So far, the operations you have seen are the same as those for member systems. However, there is yet another operation on functions which we now define. Definition: Let f and g be the operation of combining two functions by applying them one after the other. That is, the composition of f(x) and g(x), denoted by, fog . For example, consider f : R R : f(x) = (x3 + 2x )3 . We can write it as the composition of g and h, where the value of f(x) can be obtained by first calculating x3 + 2x and then taking its third power . We can write g for first or inside function g(x) = x3 + 2x. We write h for the second function : h(x) = x3 . The use of the variable x is irrelevant, we could as well write h(y) = y3 for y R. We can see that goh(x) = g(x3 + 2x) = (x3 + 2x)3 = f(x). In general (f g) (g f). For example, if, f(x) = x2 and g(x) = x+1, then (f g)(x) = (x + 1)2 and (g f)(x) = x2+1. Here we can see that f g g f. Let us see another example, where f(x) = x2, g(x) = x +1, h(x) = x3

20

Sets, Relations and Functions

Then, f (g h) (x) = (x3 +1)2 and (f g) h (x) = (x3+1)2. Here we can see f (g h) = (f g) h. Now let us see how you can get product of two permutations f and g of the same set, Let f=

a2... an a1 and f(a1) f(a2) f(an) a2.... an a1 a2... an a1 g = g(a1) g(a2)... g(an) . Then fg= f[g(a1)] f[g(a2)]... f[g(an)] is itself a
permutation. For example if, f=

1 2 3 1 2 3 , g= 2 1 3 then 2 3 1 1 2 3 1 2 3 fg= 1 3 2 and gf= 3 2 1

Note that f g g f. Thus the multiplication of permutations is not commutative in general.

However, the multiplication of permutations is associative. For example, if f=

1 2 3 4 1 3 4 2 , g=
A = {1,2,3,4}, then fg=

1 2 3 4 1 2 3 4 2 3 4 1 , h= 4 2 1 3 be the permutations on

1 2 3 4 1 2 3 4 1 2 3 4 ,gf= , gh= 3 4 2 1 , 2 4 1 3 3 4 2 1 1 2 3 4 ,(fg)h= 2 3 4 1 1 2 3 4 2 3 4 1 .

f(gh)=

Here we can see the multiplication of permutation is commutative. Now try these exercises. E 29) Let f(x) = 1/x and g(x) = x3 + 2. Find the following functions, where x R. i) (f + g)(x) ii) (f - g)(x) iii) (fg)(x) iv) (f/g)(x) E30) Let f(x) = x + 1 x 1 and g(x) = x3 x R. Define the following functions. Also give their domains. i) (f + g) ii) (f g) iii) (fg) iv) (f/g) v) (f g) With this we have come to the end of this unit. Let us now summaries what we have covered in this unit.

21

Basic Combinatorics

1.6 SUMMARY
In this unit we have covered the following points: 1. We introduced basic concepts related to sets and different ways of representing them. 2. We worked at different operations on sets and there Venn diagram representations. 3. We explored some properties common to operations on sets and logical statements. 4. In the process we also documented the duality principle. 5. We defined relations as a Cartesian product of sets and looked at several examples and type of relations. 6. We defined a function as a particular kind of relation. Then we studied different types of functions as well as basic operations on functions. In the process we considered permutations and their product.

1.7 SOLUTIONS / ANSWERS

E1) E2) A={x:x is a student of IGNOU.} i) The collection of all good teachers is not a set because this collection is not well-defined. The characteristic good cannot be measured objectively. ii) The set of points on a line is not finite because infinitely many points make a straight line. E3) E4) {, 3, 2, 1,0,1,2,3,}. Set of vowels of English alphabet V={a,e,i,o,u}. Two subsets of set V are V1={a,e}, and V2={i,o}. Two supersets of V are V3={a,b,c,,z}, and V4={a,c,d,e,i,o,u,,z}. Powerset of A={a,e,i,o,u} is { ,{a},{e},{i},{o},{u},{a,e},{a,i},{a,o},{a,u},{e,i},{e,o},{e,u},{i,o},{i,u}, {o,u},{a,e,i},{a,i,o},{a,o,u},{e,i,o},{i,o,u},{i,o,u},{a,e,i,o},{e,i,o,u},{a,e,i,o,u} . E6) E7) E8) For empty set A={ } or , P(A)=1 If A B, then P(A) P(B) because every subset of A is a subset of B. If P(A) =P(B) then A P(A)= P(B)= A B. Similarly, B A. Therefore A = B.

E5)

E9)

22

Sets, Relations and Functions

Fig.12: The Shaded portion is A B

You can try them for the other situations. We are showing in Fig. 12 for the second situation. E10) A (B C) = {Math, Physics, Science}= A. E11) i) A ~B ={1,2,3}

ii) B ~A ={7,8,9} iii) A B ={1,2,3,7,8,9} E12) Only if A and B are . E13) Write separate functions to find A~B, B~A and A B with passing sets A and B as argument, return the resultant set. E14) A B can be equal to A B if either A B or B A. E15) i) Dual of A (B C) = (A B) C is A (B C) = (A B) C. ii) Dual of (A B) (A C) is (A B) (A C). E16)

A (B C)

Fig.13: The lined portion represents A (B C)

E17)

C (A B) C (A (B C) C

Fig.14(a)

Fig.14(b)

Shaded area in Fig.14 (a) and Fig.14(b) are not same so (A B) C is not equal to (A (B C). E18) i) X X={(a,a),(a,b),(a,c),(b,b),(b,c),(c,c)}.

ii) X Y={(a,1), (b,1), (c,1), (a,2), (b,2), (c,2), (a,3), (b,3), (c,3)}. iii) X = .

23

Basic Combinatorics

E19) A B = B A iff A = B. E20) The geometric diagram for R{2} will be the line parallel to Y axis. See Fig.15.

Fig.15: y=2

E21) i)

For a A, aRa is reflexive because every one loves herself or himself.

ii) R is not symmetric because if a loves b then b need not love a, i.e., aRb does not always imply bRa. Thus R is not symmetric. iii) R is not transitive, because if a loves b and b loves c then a need not love c; i.e., if aRb and bRc, aRc need not be. Thus, R is not transitive. Hence, R is reflexive but is neither symmetric nor transitive. E22) R is a function because each element of A is assigned to a unique element of B. E23) Not every relation is a function. For example, this relation does not satisfy the property that, a) Each element of A must have assigned one element in B. b) If a A is assigned b B and a A is assigned b B then b = b . That is why relations those who dont satisfy above properties are not a function E24) We can see that the code has no effect on the value of n 0.In the While loop, the value of n is halved whenever it is even. If n becomes odd before reaching 1, the second part of the while loop is invoked, and n remains odd and increases forever. 0 if n = 0. This shows that f : NN is the function defined by f(n)= 1, if n is a powerof 2, undefinedotherwise E25) The domain of f is {1,2,3,4} and range of f is {2,3,4,5}. E26) Function f(x) = x2 is one-to-one because for every value of x, x2 will be a number that is different for different x. Hence, f(x) = x2 is one-one mapping. E27) One-to-one function will be used for providing identity card number, because each person must have unique identity numbers. E28) Step 1: y = x33 Step 2: x = y33 Step 3: y =
3

x+3
3

Step 4: f 1(x) = E29) i) (f+g) (x)=

x+3.

1 + x3 +2 x 1 ii) (fg) (x)= (x3 +2) x 1 iii) (f.g) (x)= (x3 +2) x

24

Sets, Relations and Functions

iv) (f/g) (x)=

1 x R. x( x + 2)
3

E30) i)

(f+g) (x) =

x + 1 +x3 x 1 x + 1 x3 x 1 x + 1 x3 x 1 x + 1 /x3 x 1, x 0
x 3 + 1 x 1 .

ii) (f-g) (x) = iii) (f.g) (x) = iv) (f/g) (x) =

v) (f g)(x) = f(x3) =

25

Some More Counting Principles

UNIT 3 SOME MORE COUNTING PRINCIPLES

Structure
3.0 3.1 3.2 3.3 3.4 Introduction Objectives Pigeonhole Principle Inclusion-Exclusion Principle Applications of Inclusion Exclusion
3.4.1 3.4.2 3.4.3 Application to Surjective Functions Application to Probability Application to Derangements

Page No.
47 47 47 51 54

3.6 3.7

Summary Solutions/Answers

57 57

3.0 INTRODUCTION
In this unit, we continue our discussion of the previous unit on combinatorial techniques. We particularly focus on two principles of counting the pigeonhole principle and the principle of inclusion-exclusion. In Sec. 3.2 you will see how obvious the pigeonhole principle is. Its proof is very simple, and amazingly, it has several useful applications. We shall also include some of these in this section. In Sec. 3.3, we focus on the principle (or formula) of inclusion-exclusion. As you will see, this principle tells us how many elements do not fit into any of n categories. We prove this result and also give a generalisation. Following this, in Sec. 3.4 we give several important applications of inclusion-exclusion. We shall continue our discussion on combinatorial techniques in the next unit.

3.1 OBJECTIVES
After studying this unit, you should be able to: prove the pigeonhole principle, and state the generalised pigeonhole principle; identify situations in which these principles apply, and solve related problems; prove the principle of inclusion-exclusion; apply inclusion-exclusion for counting the number of surjective functions, derangements and for finding discrete probability.

3.2 PIGEONHOLE PRINCIPLE

Let us start with considering a situation where we have 10 boxes and 11 objects to be placed in them. Wouldnt you agree that regardless of the way the objects are placed in the two boxes at least one box will have more than one object in it? On the face of it, this seems obvious. This is actually an application of the pigeonhole principle, which we now state. Theorem 1 (The Pigeonhole Principle): Let there be n boxes and (n+1) objects. Then, under any assignment of objects to the boxes, there will always be a box with more than one object in it. This can be reworded as: if m pigeons occupy n pigeonholes, where m > n, then there is at least one pigeonhole with two or more pigeons in it.

47

Basic Combinatorics

Proof: Let us label the n pigeonholes 1, 2, , n, and the m pigeons p1, p2, , pm. Now, beginning with p1, we assign one each of these pigeons the holes numbered 1, , n, respectively. Under this assignment, each hole has one pigeon, but there are still (mn) pigeons left. So, in whichever way we place these pigeons, at least one hole will have more than one pigeon in it. This completes the proof! This result appears very trivial, but has many applications. For example, using it you can show that: if 8 people are picked in any way from a group, at least 2 of them will have been born on the same weekday. in any group of 13 people, at least two are born in the same month. Let us consider some examples of its application, in detail. Example 1: Assuming that friendship is mutual, show that in any group of people we can always find two persons with the same number of friends in the group. Solution: If there are n persons in the group, then let the number of friends the ith person has be f(i), i = 1,, n. Clearly, f(i) can take values only between 0 and (n 1). If some f(i) is 0, it means that the ith person does not have any friends in the group. In this case, no person can be friends with all the other (n 1) people. So, no f(i) can be (n 1). So, only one of the values 0 or (n 1) can be present among the f(i)s. So, the n f(i)s can take only (n 1) distinct values. Therefore, by the pigeonhole principle, two f(i)s must be equal. Then the corresponding is have the same number of friends in the group. *** Example 2: Suppose 5 points are chosen at random within or on the boundary of an equilateral triangle of side 1 metre. Show that we can find two points at a distance of at most metre. Solution: Divide the triangle into four equilateral triangles of side m by joining the midpoints of the sides by three line segments (see Fig. 2). These four triangles may now be considered as boxes and the five points as objects. By the pigeonhole principle, at least one of these smaller triangles will have two points in or on it. Clearly, the distance between these two points is at most metre.
Fig. 2

Fig. 1 The pigeonhole principle

*** Example 3: Given any ten different positive integers less than 107, show that there will be two disjoint subsets with the same sum. Solution: The highest numbers we could be given would be 97, 98, , 106, which add up to 1015. So, consider pigeonholes marked 0, 1, 2,, 1015. The set of 10 positive integers have 210 = 1024 subsets. Put a subset in the pigeonhole marked with the sum of the numbers in the set. The 1024 subsets have to be put in 1016 pigeonholes. So, some pigeonhole will have more than one subset with the same sum. Now, note that two subsets that we get with the same sum, may not be disjoint. But, by dropping the common elements in them, we are left with disjoint subsets with the same sum. *** Here are some related exercises for you to do.

48

Some More Counting Principles

E1)

If 10 points are chosen in an equilateral triangle of side 3 cms., show that we can find two points at a distance of at most 1 cm. On 11 occasions a pair of persons from a group of 5 was called for a function. Show that some pair of persons must have attended the function at least twice. Four persons were found in a queue, independently, on 25 occasions. Show that at least on two occasions they must have been in the queue in the same order.

E2)

E3)

As you know, mathematics develops through a process of generalisation. You know that the principle is valid for n+1 objects and n boxes. It is natural to ask: what if we have, say, 4n+1 objects and 4 boxes? Can we prove a similar principle? In fact, we can, as given below. Theorem 2 (The Generalized Pigeonhole Principle): If nm + 1 objects are distributed among m boxes, then at least one box will contain more than n objects. This can be reworded as: Let k and n be positive integers. If k balls are put into n boxes, then some box contains at least [k/n] + 1 balls, where [x] denotes the greatest integer less than x. Proof: We prove this by contradiction (see Unit 2, Block 1). Suppose all the m boxes have at most n objects in them. Then the total number of objects is at most nm, a contradiction. Hence, the theorem. Applying this result, we see, for example, that suppose 479 students are enrolled in 479 the course Discrete Mathematics, consisting of 6 units. Then, at least + 1 = 80 6 students are studying the same unit at a given point of time.

Let us consider a few more examples of the application of this principle.

Example 4: Show that in any group of 30 people, we can always find 5 people who were born on the same day of the week. 30 Solution: 30 people can be assigned to 7 days of the week. Then at least + 1 = 5 7 of them must have been born on the same day.

***
Example 5: 20 cards, numbered from 1 to 20, are placed face down on a table. 12 cards are selected one at a time and turned over. If two of the cards add up to 21, the player loses. Is it possible to win this game? Solution: The pairs that can add up to 21 are (1, 20), (2, 19),, (10,11). So, there are 10 such pairs. In turning 12 cards, at least one of these pairs will be included. Therefore, the player will lose.

***
Example 6: Show that every sequence of n2 + 1 distinct integers includes either an increasing subsequence of n + 1 numbers or a decreasing subsequence of n + 1 numbers. Solution: Let the sequence be a1, a2,, a n 2 +1 . Suppose there is no increasing

subsequence of n + 1 numbers. For each of these aks, let s(k) be the length of the longest increasing subsequence beginning at ak. Since all n2+1 of the s(k)s are

49

Basic Combinatorics

n 2 + 1 between 1 and n, at least + 1 = n + 1 of these numbers are the same. (The n s(k)s are the objects and the numbers from 1 to n are the boxes.) Now, if i < j and s(i) = s(j), then ai > aj. Otherwise ai followed by the longest increasing subsequence starting at aj would be an increasing subsequence of length s(j) + 1 starting at ai. This is a contradiction, since s(i) = s(j). Therefore, all the n + 1 integers ak, for which s(k) = m, must form a decreasing subsequence of length at least n + 1. ***
Example 7: Take n integers, not necessarily distinct. Show that the sum of some of these numbers is a multiple of n. Solution: Let S(m) be the sum of the first m of these numbers. If for some r and m, r < m, S(m) S(r) is divisible by n, then ar+1 + ar+2 + + am is a multiple of n. This also means that S(r) and S(m) leave the same remainder when divided by n. So, if we cannot find such pairs m and r, then it means that the n numbers S(1), S(2),, S(n) leave different remainders when divided by n. But there are only n possible remainders, viz., 0, 1, 2,, (n 1). So, one of these numbers must leave a remainder of 0. This means that one of the S(i) s is divisible by n. This completes the proof.

In fact, in this example we have proved that one of the sums of consecutive terms is divisible by n. *** You may like to try some exercises now. E4) If any set of 11 integers is chosen from 1, ,20, show that we can find among them two numbers such that one divides the other. E5) If 100 balls are placed in 15 boxes, show that two of the boxes must have the same number of balls. E6) If a1, a2,, an is a permutation of 1, 2,, n and n is odd, show that the product (a1 1) (a2 2) (an n) must be even. There are several corollaries to Theorem 2. We shall present one of them here.
Theorem 3: If a finite set S is partitioned into s subsets, then at least one of the

subsets has

S k

or more elements.

Proof: Let A1, , Ak be a partition of S. (This means that Ai Aj = for i j and

S = A1A2Ak.) Then the average value of Ai is So the largest Ai has at least

S 1 [ A1 + ... + Ak ] = . k k

S k

elements.

A consequence of this result is the following theorem.

Theorem 4: Consider a function f: S T, where S and T are finite sets satisfying S > r T . Then at least one of the sets f1(t), tT, has more than r elements.

(f1(t) denotes the inverse image of the set {t}, i.e., f1(t) = {x S : f(x) = t}.)
50

Some More Counting Principles

Proof: The family {f1(t): t T} partitions S into k sets with k T . By Theorem 3,

some set in this family, say f1(t), has at least

S k

members. Since

S k

S T

>r

by our hypothesis, f1(t) has more than r elements.

Corollary: If f: S T and S > T , then f is not injective.

Proof: Putting r = 1 in Theorem 4, we see that at least one of the sets f1(t) has more than one element.

We conclude this section with some more extensions of the pigeonhole principle.
Theorem 5: Suppose we put an infinity of objects in a finite number of boxes. Then at least one box must have an infinity of objects. Proof: If every box contains only a finite number of objects, then the total number of objects must be finite. Hence the theorem. Theorem 6 (A generalisation of Theorem 3): Let A1, A2,, Ak be subsets of a finite set S such that each element of S is in at least t of the sets Ai. Then the average

number of elements in the Ais is at least t. Ai may overlap.)

S k

. (Note that, in this statement, the sets

We leave the proof to you to do, and give you some related exercises now. E7) Every positive integer is given one of the seven colours in VIBGYOR. Show that at least one of the colours must have been used infinitely many times. E8) Let A be a fixed 10-element subset of {1, 2,, 50}. Show that A possesses two different 5-element subsets, the sum of whose elements are equal. E9) The positive integers are grouped into 100 sets. Show that at least one of the sets has an infinity of even numbers. Is it necessary that at least one set should have an infinity of even numbers and an infinity of odd numbers? Let us now consider another very important counting principle.

3.3 INCLUSION-EXCLUSION PRINCIPLE

Let us begin with considering the following situation: In a sports club with 54 members, 34 play tennis, 22 play golf, and 10 play both. There are 11 members who play handball, of whom 6 play tennis also, 4 play golf also, and 2 play both tennis and golf. How many play none of the three sports? To answer this, let S represent the set of all members of the club. Let T represent the set of tennis playing members, G represent the set of golf playing members, and H represent the set of handball playing members. Let us represent the number of elements in A by A . Consider the number S T G H . Is this the answer to the problem? No, because those who are in T as well as G have been subtracted twice. To compensate for this double subtraction, we may now consider the number S T G H + T G + G H + H T . Is this the answer? No, because those playing all the three games have been subtracted thrice and then added thrice. But those members have to be totally excluded also. Hence, we now consider the number

51

Basic Combinatorics

S T G H + T G + G H + H T T G H . This is the correct answer. This reduces to 54 34 22 11 + 10 + 6 + 4 2 = 5. To solve this problem we have made inclusions and exclusions alternately to arrive at the correct answer. This is a simple case of the principle of inclusion and exclusion. It is also known as the sieve principle because we subject the objects to sieves of a progressively finer mesh to arrive at a certain grading. Let us state and prove this principle now.
_

A , or A , denotes the complement of the set A

Theorem 7 (The inclusion-exclusion formula): Let A1, A2,, An be n sets in a universal set U consisting of N elements. Let Sk denote the sum of the sizes of all the sets formed by intersecting k of the Ais at a time. Then the number of elements in none of the sets A1, A2,, An is given by
A1 A 2 ... A n = N S1 + S 2 S3 + ... + (1) k S k + ... + ( 1) n S n .
_ _ _

RHS is short for righthand side.

Proof: The proof is on the same lines of the counting argument given in the sports club example at the beginning of this section. If an element is in none of the Ais, then it should be counted only once, as part of N in the RHS of the formula above. It is not counted in any of the Sks since it is in none of the Ais.

Next, an element in exactly one Ai, say Ar, is counted once in N, and once in S1, and in none of the other Sks. So the net count is 11 = 0. Finally, take an element x in exactly m of the Ais. This is counted once in N, m times in S1, C(m, 2) times in S2 (since x is in C(m, 2) intersections AiAj),, C(m, k) times in Sk for k m. x is not counted in any Sk for k > m. So the net count of x in the RHS of the formula is 1 C(m,1) + C(m,2) + (1)k C(m, k) + + (1)m C(m, m) = 0, by Identity 2 in Sec. 2.5. So the only elements that have a net count of 1 in the RHS are those in A i . The
i =1 n _

rest have a net count of 0. Hence the formula. From this result, we immediately get the following one.
Corollary: Given the situation of Theorem 7,
A1 A 2 ... A n = S1 S 2 + ... + ( 1) k 1 S k + ... + (1) n 1 S n .

Why dont you try and prove this result? (see E 10.) What the inclusion-exclusion principle tells us is that to calculate the size of A1A2An, calculate the size of all possible intersections of the sets A1, A2,,An. Add the results obtained by intersecting an odd number of the sets, and then subtract the results obtained by intersecting an even number of the sets. Therefore, this principle is ideally suited to situations in which i) we just want the size of A1A2 An, not a listing of its elements, and ii) multiple intersections are fairly easy to count. Now let us consider some examples in which Theorem 7 is applied.
Example 8: How many ways are there to distribute r distinct objects into five (distinct) boxes with

i) at least one empty box? ii) no empty box (r 5)?

52

Some More Counting Principles

Solution: Let U be all possible distributions of r distinct objects into five boxes. Let Ai denote the set of possible distributions with the ith box being empty.

i) Then the required number of distributions with at least one empty box is A1 A 2 ... A 5 . We have N = 5r. Also, A i = (5 1) r , the number of distributions in which the objects are put into one of the remaining four boxes. Similarly, A i A j = (5 2) r , and so forth. Thus, by the corollary above, we

have A1 ... A 5 = S1 S 2 + S3 S 4 + S5

= C(5,1)4 r C(5,2)3 r + C(5,3)2 r C(5,4)1r + 0 ii) A 1 A 2 ... A 5 = 5 r C(5,1)4 r + C(5,2)3 r C(5,3)2 r + C(5,4)1r , by Theorem 7. ***
Example 9: How many solutions are there to the equation x + y + z + w = 20, where x, y, z, w are positive integers such that x 6, y 7, z 8, w 9? Solution: To use inclusion-exclusion, we let the objects be the solutions (in positive integers) of the given equation. A solution is in A1 if x > 6, in A2 if y > 7, in A3 if
_ _ _

z > 8, and in A4 if w > 9. Then what we need is A1 A 2 A 3 A 4 .

Now, to find the total number of positive solutions to the given equation, we rewrite it as x1+y1+z1+w1 = 16, where x1 = x+1, y1 = y+1, z1 = z+1, w1 = w+1. Any nonnegative solution of this equation will be a positive solution of the given equation. So, the number of positive solutions is N = C(16+41, 16) (see Example 11 of Unit 2) = C(19, 3). Similarly, A1 = C(13,3), A 2 = C(12,3), A 3 = C(11,3), A 4 = C(10,3), A1 A 2 = C(6,3), A 2 A 3 = C(5,3), and so on. Note that for a solution to be in 3 or more Ai s, the sum of the respective variables would exceed 20, which is not possible. By inclusion- exclusion, we obtain A1 A 2 A 3 A 4 = C(19, 3) C(13, 3) C(12, 3) C(11, 3) C(10, 3) + C(6, 3) + C(5, 3) + C(4, 3) + C(4, 3) + C(3, 3) = 217. *** Now you may try the following exercises. E10) Prove the corollary to Theorem 7. E11) How many numbers from 0 to 999 are not divisible by either 5 or 7? Let us now consider applications of the inclusion-exclusion principle to some specific problem types.
_ _ _ _

53

Basic Combinatorics

3.4 APPLICATIONS OF INCLUSION-EXCLUSION

In this section we shall consider three broad kinds of applications for counting the number of surjective functions, finding probability and finding the number of derangements.

3.4.1 Application to Surjective Functions

Let us first recall that a function f : S T is called surjective (or onto) if f(S) = T, that is, if for every tT sS such that f(s) = t. Now let us prove a very useful result regarding the number of such functions.
Theorem 8: The number of functions from an m-element set onto a k-element set is
i =0

(1) i C(k , i)(k i) m , where 1 k m.

Proof: We will use the inclusion-exclusion principle to prove this. For this, we define the objects to be all the functions (not just the onto functions) from M, an melement set, to K, a k-element set. For these objects, we will define Ai to be the set of all f: M K for which the ith element of K is not in f(M). Then what we want is

A1 .... A k . Now, the total number of functions from M to K is km. Also, the number of mappings that exclude a specific set of i elements in K is (k i)m, and there are C(k, i) such sets. Therefore, A i = (k 1) m , A i A j = (k 2) m , and so on. Now, applying Theorem 7, we get
A1 ... A k = k m C(k ,1)(k 1) m + C(k ,2)(k 2) m ... + (1) k 1 C(k , k 1)1m
_ _

Hence the result. For example, the number of functions from a five-element set onto a three-element set are (1) i C(k , i)(k i) m for m = 5 and k = 3, that is, 35 3.25 + 3.15 = 150.
i =0 k

Why dont you try some exercises now? E12) Eight people enter an elevator. At each of four floors it stops, and at least one person leaves the elevator. After four floors the elevator is empty. In how many ways can this happen? E13) How many six-digit numbers contain exactly three different digits? Now we look at another application.

3.4.2 Application to Probability

An important application of the principle of inclusion-exclusion is used in probability. We have the following theorem.
Theorem 9: Suppose A1,A2,, An are n events in a probability space. Then

54

Some More Counting Principles

P(A1 A2 An) = (1) r +1

r =1

1i1 <i 2 <...<i r n

P(A i1 A i 2 ... A i r )

Proof: Let us begin by observing that A1 A2 An means that at least one of the events A1, A2,, An occurs. Now, let the ith property be that an outcome belongs to

the event Ai. By De Morgans law, A1 A 2 ... A n is the complement of A1 A2 An. But the principle of inclusion-exclusion gives

A 1 A 2 ... A n = N (1) r
the total number of outcomes.

1 i1 < i 2 <...< i r n

Ai1 Ai2 ... Air , where N is

Now, we divide throughout by N and note that P( A1 A2 ... An ) = 1 P( A1 A 2 ... A n ), to get the result. Let us consider an example of the use of this result.
Example 12: Find the probability of a student in a college studying Japanese, given the following data:
_ _ _

All students have to study at least one language out of Hindi, Spanish and Japanese. 65 study Hindi, 45 study Spanish and 42 study Japanese. Further, 20 study Hindi and Spanish, 25 study Hindi and Japanese, 15 study Spanish and Japanese, and 8 study all 3 languages.
Solution: The total number of students is H S J , where H, S and J denote the

number of students studying Hindi, Spanish and Japanese, respectively. By the inclusion-exclusion principle, H S J = H+S+JH SH JS J+H S J = 65 + 45 + 42 20 25 15 + 8 = 100 Therefore, the required probability is

J 100

= 0.42.

*** You could do the following exercises now. E14) What is the probability that a 13-card hand has at least one card in each suit? E15) What is the probability that a number between 1 and 10,000 is divisible by neither 2, 3, 5 nor 7? Let us now come to the use of inclusion-exclusion for counting the number of a particular kind of permutation.

3.4.3 Application to Derangements

As you know, a permutation of a set is an arrangement of the elements of a set. So, for example, a rearrangement 1 1, 2 2, 4 3, 3 4 is a permutation of the 4-element set {1, 2, 3, 4}. In this permutation, the position of the elements 1 and 2 are fixed, but the positions of 3 and 4 have been interchanged. Now consider the permutation 1 4, 2 1, 3 2, 4 3, of {1, 2, 3, 4}, in which the position of

55

Basic Combinatorics

every element has been changed. This is an example of a derangement, a term we shall now define. Definition: A derangement of a set S is a permutation of the elements of S which does not fix any element of S, i.e., it is a rearrangement of the elements of S in which the position of every element is altered.

So, if we treat a permutation as a 1-to-1 function from S to S, then a derangement is a function f:S S such that f(s) s sS. We have the following theorem regarding the number of derangements. (1) i . i =0 i! Proof: Let Ai be the set of all permutations of the n-element set that fix i i = 1, , n. Then
Theorem 10: The number of derangements of an n-element set is Dn = n!
n

Dn = A i = n! Ai + Ai A j ... + (1) n A1 ... An

i =1 i i< j

= n! C(n, 1) (n 1)! + C(n, 2) (n 2)! + (1)nC(n, n)0! 1 1 1 1 = n! 1 + + ... + (1) n , which is the expression we wanted. n! 1! 2! 3! 1 1 1 1 Remark: The expression 1 + + ... + (1) n is the beginning of the 1 ! 2 ! 3 ! n ! 1 expansion for e . Even for moderately large values of n, Dn is very close to n!e1=0.36788 n!. As an extension of Theorem 10, we have the following results.
Theorem 11: For a set of n objects, the number of permutations in which

(i) only r of these n objects are deranged is n! C(r, 1) (n 1)! + C(r, 2) (n 2)! + (1)r C(r, r) (n r)!; (ii) exactly r elements are fixed is C(n, r) Dnr. We will not prove these formulae here, but shall consider some examples of their applications.
Example 12: Let n books be distributed to n children. The books are returned and distributed to the children again later on. In how many ways can the books be distributed so that no child will get the same book twice? Solution: The required number is (n!)2e1, since corresponding to each first distribution, there are (n!)e1 ways of distributing again.

***
Example 13: Suppose 10 people have exactly the same briefcases, which they leave at a counter. The cases are handed back to the people randomly. What is the probability that no one gets the right case? Solution: The number of possibilities favourable to the event is D10. The total number of possibilities is 10!. Thus, the probability that none will get the right briefcase is D10/10! = 0.36788.

***

56

Some More Counting Principles

Note that, since Dn n!e1, the possibility in all such examples is essentially e1, which is independent of n.

You may now try the following exercises. E16) Each of the n guests at a party puts on a coat when s/he leaves. None of them gets the correct coat. In how many ways can this happen? In how many ways can just one of the guests get the right coat? E17) In how many ways can the integers 1, 2, 3,,9 be permuted such that no odd integer will be in its natural position. E18) Find the number of permutations in which exactly four of the nine integers 1, 2,,9 are fixed. With this we come to the end of this unit. In the next unit we shall continue our discussion on counting from a slightly different perspective. Let us now summarise what we have covered in this unit.

3.6 SUMMARY
In this unit, you have studied the following points. 1. The pigeonhole principle, stated in several forms, its proof, and its applications. The generalized pigeonhole principle, its proof, and applications. The inclusion-exclusion principle, and its proof. Finding the number of surjective functions, the discrete probability and the number of derangements, by using the inclusion-exclusion principle.

2. 3. 4.

3.7 SOLUTIONS /ANSWERS

E1) By drawing lines parallel to the sides and through the points trisecting each side, we can divide the equilateral triangle into 9 equilateral triangles of side 1 cm. Thus, if 10 points are chosen, at least two of them must lie in one of the 9 triangles. 5 persons can be paired in C(5, 2) =10 ways. Hence, if pairs are invited 11 times, at least one pair must have been invited twice or more times, by the pigeonhole principle. Four persons can be arranged in a line in 4! = 24 ways. Hence, if we consider 25 occasions, at least on two occasions the same ordering in the queue must have been found, by the pigeonhole principle. Consider the following grouping of numbers: {1, 2, 4, 8, 16}, {3, 9, 18}, {5, 15}, {6, 12}, {7, 14}, {10, 20}, {11}, {13}, {17}, {19}. There are 10 groupings, exhausting all the 20 integers from 1 to 20. If 11 numbers are chosen it is impossible to select at most one from each group. So two numbers have to be selected from some group. Obviously one of them will divide the other.

E2)

E3)

E4)

57

Basic Combinatorics

E5)

Suppose x1, x2,, x15 are the number of balls in the 15 boxes, listed in increasing order, assuming that all these numbers are different. Then, clearly, xi i 1 for i = 1, 2,,15. But then, x i 14.15 / 2 = 105.
i =1 15

But the total number of balls is only 100, a contradiction. Thus, the xis cannot all be different. E6) In the sequence a1, a2,,an, there are (n+1)/2 odd numbers and (n1)/2 even numbers because n is odd. Hence, it is impossible to pair all the ais with numbers from 1, 2,, n with opposite parity (evenness and oddness). Hence, in at least one pair (i, ai), both the numbers will be of the same parity. This means that the factor (aii) will be even, and hence the product will be even. Consider the seven colours as containers, and the integers getting the respective colour as their contents. Then we have a distribution of an infinite number of objects among 7 containers. Hence, by Theorem 5, at least one container must have an infinity of objects, that is, the colour of that container must have been used an infinite number of times. Let H be the family of 5-element subsets B of A. For each B in H, let f(B) be the sum of the numbers in B. Obviously, we must have f(B) 1 + 2 + 3 + 4 + 5 = 15, and f(B) 46 + 47 + 48 + 49 + 50 = 240. Hence, f: H T where T = {15, 16,, 240}. Since T = 226 andH= C(10,5) = 252, by Theorem 4, H contains different sets with the same image under f, that is different sets, the sums of whose elements are equal. E9) The 100 collections can be considered as containers. There are an infinity of even numbers. When these even numbers are distributed into 100 containers, at least one container must have an infinity of them, by Theorem 5.

E7)

E8)

E10) The inclusion-exclusion formula can be rewritten as

A1 ... A n = N (S1 S 2 + ... + ( 1) n 1 S n .
_ _

Also, we know that A1 ... A n = N A1 ... A n . Hence the result. E11) Let the objects be the integers 0, 1,, 999. Let A1 be the set of numbers divisible by 5, and A2 the set of numbers divisible by 7. Now, N = 1000, A1 = 200, A 2 =143 and A1 A 2 = 29. So, by Theorem 7, the answer is 1000 200 143 + 29 = 686. E12) The answer to this problem is clearly the number of functions from an 8element set (the set of people) onto a set of 4-elements (the set of floors). This number is
i 0

C(4, i)(4 i) 8 = 4 8 4.38 + 6.2 8 4.18.

E13) We can choose three digits in C(10,3) = 120 ways. The number of 6-digit numbers, using all the three digits, is the same as the number of functions from a 6-set onto a 3-set. This number is
58

Some More Counting Principles

363.26 + 3.16 = 540. Hence, the answer is 120.540 = 64800. This will include numbers starting with 0 also. E14) The total number of ways in which 13 cards can be chosen from a deck of 52 cards is C(52, 13). If Ai is a choice of cards, none of which are from the ith suit, for i = 1,2,3,4, then A i = C(39,13), A i A j = C(26,13), and C(AiAj Ak) = C(13,13). So, A i = C(52,13) 4C(39,13) + C( 4,2)C(26,13) C(4,3) C(13, 13)
_

Ai
Hence, the required probability is

C(52,13)

E15) If A,B,C,D are the integers divisible by 2, 3, 5, 7, respectively, then

_ _ 10000 10000 10000 10000 A ... D = 10,000 2 3 5 7

10000 10000 10000 10000 10000 10000 + + + + + + 6 15 35 14 21 10 10000 10000 10000 10000 10000 + 30 42 105 70 210 = 2285, where [x] denotes the greatest integer x. Hence, the required probability is
2285 = 0.23. 10000

E16) If Ar is the event that the rth person gets the right coat, then by Theorem 7, A i = n! A r + A r A s ...
r r ,s _

= n! n(n1)! + C(n,2)(n2)! C(n,3)(n3)!+ = C(n,2)(n2)! C(n,3)(n3)!+

n 1 = n! (1) r r r! =0 The number of ways in which only one person receives the correct coat is the sum of all possible intersections of (n1) A i s . This is
_

n 1 1 (1) r n! n(n1)! = n! r r! =0

n 1 1 (1) r . r r! =0

E17) 1, 3, 5, 7, 9 are the odd integers. By Theorem 11(i), the required number of ways is 9! C(5, 1)8! + C(5, 2)7! C(5,3)6! + C(5, 4)5! C(5, 5)4! E18) By Theorem 11(ii), the required number of permutations is C(9,4)D94=C(9,4)D5.
59

Basic Combinatorics

60

UNIT 4 PARTITIONS AND DISTRIBUTIONS

Structure
4.0 4.1 4.2 4.3 Introduction Objectives Integer Partitions Distributions
4.3.1 4.3.2 4.3.3 4.3.4 Distinguishable Objects into Distinguishable Containers Distinguishable Objects into Indistinguishable Containers Indistinguishable Objects into Distinguishable Containers Indistinguishable Objects into Indistinguishable Containers

Partitions and Distributions

Page No.
61 61 61 64

4.4 4.5

Summary Solutions /Answers

69 70

4.0 INTRODUCTION
In the last two units we have exposed you to a variety of combinatorial techniques. In this unit we look at a few more ways of counting arrangements of objects when order matters, and when it doesnt. In Sec. 4.2, we focus on the ways in which a natural number can be written as a sum of natural numbers. In the process you will be introduced to a useful recurrence relation. We link this, in Sec. 4.3, with the different ways in which n objects can be distributed among m containers. As you will see, there are four broad possible kinds of distributions. In each case, we consider ways of counting all the distributions. In the process you will also be introduced to Stirling numbers. With this unit we come to the end of our discussion on counting techniques. Some of the problems you have studied here will be looked at from different approaches in our later course MCS-033. You should attempt the assignment based on the course after studying this unit, and this block.

4.1 OBJECTIVES
After going through this unit, you should be able to: define an integer partition, and count the number of partitions of an integer; count the number of ways of distributing distinguishable and indistinguishable objects, respectively, into distinguishable containers; count the number of ways of distributing distinguishable and indistinguishable objects, respectively, into indistinguishable containers.

4.2 INTEGER PARTITIONS

Suppose a detergent manufacturer wants to promote her product by giving a gift token with 100 bars out of the whole stock. The lucky persons among her customers will get the gift. Some of them may buy more than one bar at a time with the hope of getting gifts. In how many ways can the 100 gift tokens get distributed? One possible way is that all the 100 bars with gifts are bought by 100 different customers. We can indicate +4 12 + ... + 1. Another possibility is that somebody buys 2 bars, this situation by 100 = 1 1 4 3
100 times

61

Basic Combinatorics

somebody else buys 3 bars, and the remaining 95 bars are distributed amongst 95 different people. We are not interested in the order in which the bars are bought. For example, here we are not interested in whether the person who bought 2 bars bought them before the person who bought the 3 bars. So, we can indicate this situation by +4 + ... + 100 = 1 12 1 + 2 + 3 . More generally, we can indicate each way of distributing 1 4 3
95 times

the 100 bars with gifts by 100 = p1 + p2 + p2 + + pk, where the pi are natural numbers, and p1 p2 pk. Each way of writing 100 in this form is called an integer partition of 100. More generally, we have the following definition.
Definition: Any representation of nN as a sum of positive integers in nonincreasing order is called a partition (or integer partition) of n. Each such partition can be written in the form n = p1 + p2 + + pk, where p1 p2 pk.

Here, p1, p2,, pk are called the parts of the partition, and the number of parts of the partition is k. While we chose 100 in the example above, it is really a huge number in the context of integer partitions. Let us consider a smaller number, say 5. How many partitions of 5 can you think of? There are 7 altogether, namely, 5, 1+4, 2+3, 1+1+3, 1+2+2, 1+1+1+2 and 1+1+1+1+1.
In books, you will often come across the notation p(n) for the number of partitions of n.

If we represent the number of partitions of the integer n by Pn, we have shown that P5 = 7. These partitions have 1, 2, 2, 3, 3, 4 and 5 parts, respectively.
k , then we If we represent the number of partitions of n with exactly k parts by Pn 1 2 3 4 5 have P5 = 1, P5 = 2, P5 = 2, P5 = 1, P5 = 1.

To check your understanding of the material so far, try the following exercises. E1) E2)
4 5 Write down all the partitions of 7. Also find P7 and P7 .

Let us consider the situation of the detergent manufacturer again. Suppose she only wants to distribute 10 gift tokens in 5 specific sales districts, where the sales are low. What is the number of ways of doing this?

You may wonder if youve found all the partitions in E2. One way to check is by finding out the required number in terms of partitions of smaller numbers, which may be easier to find. One such relation between partitions of n and n+1, n+2, etc. is given in the following theorem.
1 2 k k 1 n Theorem 1: Pn + Pn + ... + Pn = Pn + k , Pn = Pn = 1, for 1 k n , that is, the number of partitions of n with at most k parts is the same as the number of partitions of n+k with exactly k parts.

Before we begin the proof of this theorem, let us consider an example. Let us take n = 1 2 3 3 + P4 + P4 = P7 4, k = 3. According to Theorem 1, we must have P4 . Note that
1 2 3 P4 + P4 + P4 is the total number of partitions of 4 with 1, 2 or 3 parts, i.e., the number of partitions with at most 3 parts. There is one partition of 4 with one part, 4 = 4. Let us write this as a 3-tuple, (4, 0, 0), adding two more zeroes since we are considering partitions with at most 3 parts. If we add 1 to all the entries of this 3-tuple, we get (4+1,0+1,0+1) = (5,1,1) and (1+1+5 is a partition of 7 with three parts. Similarly, consider the partition 4 = 1+3 of 4 into two parts. Again, we can write this as (1, 3, 0).

62

Now, if we add 1 to each of the entries, we get (2, 4, 1) and 1+2+4 is a partition of 7 into three parts. Conversely, if we take the partition 7 = 1 + 3 +3 of 7 into three parts, write it as (1, 3, 3) and subtract 1 from all the entries, we get (0, 2, 2) which corresponds to the partition 4 = 2+2 of 4 into 2 parts. In this way, we can match every partition of 4 with at most 3 parts with a partition of 7 with exactly 3 parts, and vice versa. This is the basic idea behind our proof of Theorem 1, which we now give.
1 n Proof of Theorem 1: The cases Pn = 1 = Pn follow from the definition.

Partitions and Distributions

We will prove the general formula now. Let M be the set of partitions of n having k or less parts. We can consider each partition belonging to M as a k-tuple after adding as many zeroes as necessary. Define the mapping
,0 ,..., 0) a (p1+1,p2+1,,pm + 1,1 ,1 ,..., 1), m k (p1, p2,, pm, 0 1 2 3 1 2 3
(k m) times (k m) times

from M into the set M of partitions of n+k into exactly k parts. This mapping is bijective, since i) two distinct k-tuples in M are mapped onto two distinct k-tuples in M; ii) every k-tuple in M is the image of a k-tuple of M. This is because, if (p1, p2,,pk) is a partition of n+k with k parts, then it is the image of (p1 1, p2 1,, pk 1) under the mapping above.
1 k k + ... + Pn = M = Pn Therefore, M = Pn + k , and the theorem is proved.

k Note that Pn = 0 if n < k, since there is no partition of n with k parts if n < k. Also,

n 1 Pn = Pn = 1.
r from values of The formula in Theorem 1 is an identity which allows us to find Pn k k Pm , where m < n, k r. This is why it is also called a recurrence relation for Pn .

Theorem 1 is every useful. For instance, to verify your count in E2, you can use it
5 1 2 5 = P5 + P5 + ... + P5 = 7. because P10 k From Theorem 1, the Pn s may be calculated recursively as shown in Table 1.

k Table 1 : Pn for 1 n, k 6

k n
1 2 3 4 5 6

1 1 1 1 1 1 1

2 0 1 1 2 2 3

3 0 0 1 1 2 3

4 0 0 0 1 1 2

5 0 0 0 0 1 1

6 0 0 0 0 0 1

2 In this table, the second entry in the row corresponding to n = 4 is P4 . By Theorem 1, 2 1 2 P4 , which is the sum of the entries in the row corresponding to n = 2. = P2 + P2 3 Similarly, P6 is the sum of the entries in the row corresponding to n = 3.

63

Basic Combinatorics

Now, here is an exercise for you. E3)

k Use Table 1 to find the values of P7 ,1 k 6.

The partition of a number n into k parts also tells us how n objects can be distributed among k boxes. We will now consider all possibilities of such distributions.

4.3 DISTRIBUTIONS
By a distribution we mean a way of placing several objects into a number of containers. For example, consider the distribution of 6 balls among 3 boxes. We may have all 6 balls of different shapes, sizes and colours, i.e., they are all distinguishable. Or, all the balls could be exactly the same, i.e., they are all indistinguishable. Similarly, all 3 boxes may look different, or all 3 could be exactly the same. So, we see that there are 4 possibilities here. In fact, we have the following possibilities for any set of n objects and m boxes.

Case 1: The objects are distinguishable, and so are the boxes; Case 2: The objects are distinguishable and the boxes are indistinguishable; Case 3: The objects are indistinguishable and the boxes are distinguishable; Case 4: The objects are indistinguishable, and so are the boxes.
You may be surprised to know that in each of the cases the number of such distributions is different. In fact, the distribution problem is to count all possible distributions in any of these situations, or in a combination of these cases. A general guideline for modelling a distribution problem is that a distribution of distinct objects corresponds to an arrangement, and a distribution of identical objects corresponds to a selection. Let us consider examples of each of the four cases given above. (a) There are twenty students and four colleges. In how many ways can the students be accommodated in the four colleges? In this example the students, as well as the colleges, are clearly distinguishable. This comes under Case (1). (b) Suppose we want to group 100 students into 10 groups of 10 each for the purpose of a medical examination. In how many ways can this be done? Here the groups are indistinguishable, though the students in them are distinguishable. Hence, this falls under Case (2). (c) An employer wants to distribute 100 one-rupee notes among 6 employees. What is the number of ways of doing this? Though the one-rupee notes can be distinguished by their distinct numbers, we dont consider them to be distinguishable as far as their use is concerned. The employees, of course, are distinguishable. Hence, this is an example of Case (3). (d) There are 1000 one-rupee notes. In how many ways can they be bundled into 20 bundles?

64

As before, the rupee notes are treated as indistinguishable. Clearly, the bundles are, by themselves, not distinguishable. Only the quantity in each may vary. Hence, this falls under Case (4). Let us consider each case in some detail now.

Partitions and Distributions

4.3.1 Distinguishable Objects into Distinguishable Containers

Let us consider the example (A) above. Since any number of students can be put in a college, there are 20 20 20 20 possibilities, by the multiplication principle. More generally, suppose we are distributing n objects into m containers, both being distinguishable. Then the total number of such distributions is nm. Let us look at an example.

Example 1: Show that the number of words of length n on an alphabet of m letters is mn. Solution: The m letters of the alphabet can be used any number of times in a word of n letters. The word can be considered as n ordered boxes, each holding a letter from the alphabet. The boxes become distinguishable because they are ordered. The letters of the alphabet are clearly distinguishable. So, the number of ways of doing this is mn. ***
Several people are confused while solving the problem above. They tend to take the m letters as the containers instead! Lets consider another example.

Example 2: Suppose we have a set S with n objects. An m-sample from this set S is an ordered arrangement of m letters taken from S, with replacement at every draw, in m draws. Find the number of m-samples from an n-element set. Solution: Every m-sample is a word of length m from the alphabet S containing n letters. Hence, the required number is nm.
*** Now here are some exercises for you to solve. E4) Find the number of three-letter words that can be formed using the letters of the English alphabet. How many of them end in x? How many of them have a vowel in the middle position? E5) How many five-digit numbers are even? How many five-digit numbers are composed of only odd digits? E6) There are 4 women and 5 men. A committee of three, a president, a vicepresident, and a secretary, has to be formed from them. In how many ways can this be done if i) the vice-president should be a woman? ii) exactly one out of the vice-president and the secretary should be a woman? iii) there is at least one woman in the committee? Now suppose, we want to find the number of distributions of n distinguishable objects into m distinguishable containers, with the extra condition that no container should
65

Basic Combinatorics

contain more than one object. It is clear that this requires m n. Then we can get all these arrangements by first choosing n containers to contain exactly one object, and then permuting the n objects among the chosen containers. This can be done in C(m, n). n! = P(m, n) ways. So, we have proved the following result.

Theorem 2: The number of ways of distributing n distinguishable objects into m distinguishable containers such that no container contains more than one object is P(m, n).
For example, the cardinality of the set of 5-digit numbers with all digits being distinct odd numbers is P (5,5). This is because the possible digits are 1, 3, 5, 7, 9. Why dont you try an exercise now? E7) Find the number of m-letter words with distinct letters, all taken from an alphabet with n letters, where n m. Is this different from the number of injective mappings from an m-element set into an n-element set, where n m? Give reasons for your answer.

Let us now consider the second type of distribution.

4.3.2 Distinguishable Objects into Indistinguishable Containers

Here we shall find the number of ways of distributing n distinguishable objects into m indistinguishable containers. For this, we first find the number when exactly k of the containers are occupied. This brings us to Stirling numbers of the second kind, named after James Stirling (1692-1770). Suppose n m. The number of distributions of n distinguishable objects into m indistinguishable containers such that no container is empty is represented by S m n . This number is called the Stirling number of the second kind. As you can see, this is also the number of partitions of a set of n objects into m classes.

Definition: For natural numbers n and m, the Stirling number of the second kind, Sm n , is the number of partitions of an n-element set into exactly m parts.
Note that: i)

Sm n = 0 if n < m, for, if the number of containers exceeds the number of objects, then it is impossible to have all the containers non-empty. Sn n = 1, since there is only one way of putting n distinguishable objects in n indistinguishable boxes so that no box is empty.

ii)

iii) S1 n = 1. Now, we shall use the inclusion exclusion principle to find the value of S m n .

Theorem 3: S m n =

1 m (1) k C(m, m k )(m k ) n , n m. m! k =0

Proof: If the m classes are distinguishable, the number of partitions is the same as the number of functions from an n-element set onto an m-element set. As the classes are distinguishable here, we have to divide this number by m!. The result follows from Theorem 8, Unit 3.
66

For example, to obtain the Stirling number, S3 5 , we know that the number of functions from a 5-element set onto a three-element set is 150. So, by Theorem 3, S3 5 = 150/3! = 25.

Partitions and Distributions

Remark: You may be wondering how we have jumped straightaway to the Stirling numbers of the second kind. What about the first kind? We wont be using them in any way here. However, for a the sake of completeness, we define Stirling numbers of the first kind, s(n, k), as follows.
For a positive integer n, and 0 k n, s(n, k) is the coefficient of xk in the expansion of the multinomial x(x 1)(x 2)(x n + 1). Getting back to S m n , you may feel that the formula in Theorem 3 is a little cumbersome. Sometimes, the following recurrence relation for S m n may be more useful.
m 1 Theorem 4: If 1 < m n, then S m + mS m n +1 = S n n .

Proof: Let us take n+1 objects, mark one of them, and consider the distribution of these n+1 objects into m indistinguishable containers. Then we have 2 situations. Case (1) (The marked object is placed in one container without any other objects.): In 1 ways. this case, the remaining n objects can be placed in (m 1) containers in S m n Case (2) (The marked object is placed with at least one more object in a container.): In this case, we can first distribute the n unmarked objects into m containers, and then put the marked objects m to one of these m containers. So, the number of such partitions is m S m n .
m 1 Therefore, by the addition principle, we get S m + mS m n +1 = S n n .

There is a generalisation of Theorem 4 that is of independent interest, which we now state.

m 1 Theorem 5: S m n +1 = C( n , k ).S k k =0 n

Proof: Let us mark one object in a set of (n+1) objects. Suppose the marked object is present in a box with (n k + 1) elements, where m 1 k n. Then we can choose n k more objects to go with the marked object in C(n, n k) ways. The remaining k
1 objects can be distributed into (m 1)n boxes in S m k ways. So the number of ways of 1 distributing the n k objects is C(n, n k) S m k . The result now follows from the addition principle by allowing k to vary from 0 to n.

Let us see some examples of the use of these recurrences.

2 Example 3: Calculate S3 and S 2 4. 2 2 Solution: Using Theorem 4, we get S3 = S1 2 + 2 S 2 = 1 + 2 1 = 3, and 1 2 S2 4 = S 3 + 2 S3 = 1 + 2 3 = 7.

*** Now let us find what we had started with in this sub-section.
67

Basic Combinatorics

Theorem 6: The number of ways of distributing n distinguishable objects into m 2 m indistinguishable containers is S1 n + S n ++ S n , where n m. (Note that here we do not insist that no container is empty.) Proof: When we distribute n distinguishable objects into m indistinguishable containers there are m cases. Case (k) is that exactly k containers are non-empty.

Here k varies from 1 to m. The number of distributions in Case (k) is S k n . The result now follows from the addition principle. Let us consider an example.
Example 4: In how many ways can 20 students be grouped into 3 groups?
2 3 Solution: Theorem 6 says that this can be done in S1 20 + S 20 + S 20 ways.

Now, using Theorem 3, we get this number to be 1+ 1 2 1 3 (1) k C(2,2 k )(2 k ) 20 + (1) k C(3,3 k )(3 k ) 20 2 k =0 6 k =0

= 581,130,734. *** Try some exercises now. E8) E9) Find the number of surjective functions from an n-element set onto an melement set. Find the number of ways of placing n people in n 1 rooms, no room being empty.

Let us now consider the third possibility for distributing objects into containers.

4.3.3 Indistinguishable Objects into Distinguishable Containers

Suppose there are n indistinguishable objects and m distinguishable containers. As the objects are indistinguishable, the distributions depend only on the number of objects in each container. As the containers are distinguishable, they can be assumed to be arranged in a line. Hence, the number of distributions is the number of ways of writing the number n as the sum x1+x2++xm, where the xis are non-negative integers. We have covered this situation in Theorem 5 of Unit 2. Over there we have shown that the number of distributions of n indistinguishable objects into m distinguishable containers is C (m+n1, n). In particular, the number of nonnegative integral solutions of the equation x1+x2++xm = n is C (m+n1, n). Incidentally, we note that the number of distributions of n indistinguishable objects into m distinguishable containers with at most one object per container is C (m, n). Let us consider an example.
Example 5: How many distinct solutions are there of x + y + z + w = 10

i) in non-negative integers? ii) in positive integers?

68

Solution:

i) From the result quoted above, the answer is C(4 + 10 1, 10) = 286. ii) We want x, y, z, w to be positive. Hence, we can write them respectively as X+1, Y+1, Z+1, W+1, where X, Y, Z, W are non-negative. Hence we want the number of non-negative solutions of the equation X+1+Y+1+Z+1+W+1=10, i.e., X+Y+Z+W= 6. The answer, now, is C (4 + 6 1, 6) = 84. Try some exercises now. E10) Show that the number of positive solutions of the equation x1+x2++ xn = m is C(m 1, m n). E11) In how many ways can an employer distribute 100 one-rupee notes among 6 employees so that each gets at least one note? Let us now consider the fourth case.

Partitions and Distributions

4.3.4 Indistinguishable Objects into Indistinguishable Containers

Suppose there are n indistinguishable objects and m indistinguishable containers. Any distribution is determined purely by an unordered m-tuple of non-negative integers with sum n. This is equivalent to the number of increasing sequences of length m of non-negative integers with sum n. But this is precisely the number of partitions of the
1 2 m m + Pn + ... + Pn = Pn integer n with at most m parts, viz., Pn + m , from Theorem 1 of this unit.

Let us consider an example of this case.

Example 6: In how many ways can 20 identical books be placed in 4 identical boxes?
1 2 3 4 4 Solution: The answer is P20 + P20 + P20 + P20 = P24

Why dont you try some exercises now? E12) In how many ways can 1000 one-rupee notes be bundled into a maximum of 20 bundles? E13) A car manufacture has 5 service centres in a city. 10 identical cars were served in these centres for a particular mechanical defect. In how many ways could the cars have been distributed at the various centres? With this we have come to the end of this unit. Let us take a quick look at what we have studied in this unit.

4.4 SUMMARY
1. A partition of nN into k parts is x1+x2++xk = n, where x1 x2 xk. Pn k is the set of all partitions into exactly k is the set of all partitions of n, and Pn parts. The proof and applications of the recurrence relation, 1 2 k k 1 n Pn + Pn + ... + Pn = Pn + k , Pn = Pn = 1, 1 k n. The number of ways of distributing n objects into m containers is :
69

2.

3.

Basic Combinatorics

i) nm, if the objects and containers are distinguishable. ii) S in , if the objects are distinguishable but the containers are not.
i =1 m

(Here Sij is a Stirling number of the second kind).

iii) C(m+n1, n) if the objects are not distinguishable but the containers are distinguishable. m iv) Pn + m , if neither the objects nor the containers are distinguishable. Further, in (i) above, if there is an extra requirement that each container contain at most one object, then the number of distributions is P(m, n). Again, in (iii) above, with the same extra requirement, the number of distributions is C(m, n).

4.5 SOLUTIONS /ANSWERS

E1) In the table below we give all possible partitions of 7.
Table 2

Number of parts 1 2 3 4 5 6 7

Partitions
7 1+6, 2+5, 3+4 1+1+5, 1+2+4, 1+3+3, 2+2+3 1+1+1+4, 1+1+2+3, 1+2+2+2 1+1+1+1+3, 1+1+1+2+2 1+1+1+1+1+2 1+1+1+1+1+1+1

4 5 From the table, we see that P7 = 3, P7 = 2.

E2) E3)

5 = 7. The required number is P10 1 7 P7 = 1 = P7 . 2 1 2 P7 = P5 + P5 = 1 + 2 = 3 , from Table 1. 3 1 2 3 P7 = P4 + P4 + P4 = 1 + 2 + 1 = 4, from Table 1. 4 1 2 3 4 5 1 2 6 1 = P3 + P3 + P3 + P3 = 3, P7 = P2 + P2 = 2 and P7 = P1 = 1. Similarly, P7

E4)

The 26 letters are distinguishable objects. We have to fill then in three distinguishable containers, viz., the first, second, and third positions of a threelettered word. The solution is 263. If the last letter is to be x, the number is only 262 1. If the middle letter is a vowel, then by the multiplication principle, the answer is 26 5 26.

E5)

The total number of even numbers is 9 10 10 10 5 = 45,000, since the last digit can only by 0, 2, 4, 6 or 8. The number of 5-digit numbers composed of only odd digits (i.e., 1, 3, 5, 7, 9) is clearly 55 = 3125.

70

E6)

i) We can choose a woman for vice-president in 4 ways. To fill the remaining 2 positions we can select 2 from the remaining 8 persons in 8 7 = 56 ways. Hence, the required number is 4 56 = 224. ii) If the vice-president is a woman (chosen in 4 ways), others can be selected in 5 4 = 20 ways. Similarly, if the woman is a secretary, the others can be chosen in 20 ways. Hence, by the addition and multiplication principles, the answer is 20 4 + 20 4 = 160. iii) Without any restriction, three can be selected in 9 8 7 = 504 ways. If no woman is to be selected, then it can be done in 5 4 3 = 60 ways. What we need is the complement of this. Thus, the required answer is 504 60 = 444.

Partitions and Distributions

E7)

If the alphabet has n letters, the m-letter words with distinct letters can be formed in n (n 1)(n 2)(n m + 1) = P(n, m) ways. Now, in an injective mapping, images of distinct elements should be distinct (see Unit 1). There are n possible images for the first element of the m-set, n1 possible images for the second, and so on. Hence, the number of such mappings is also P(n, m).

E8)

Suppose N = {1, 2, ,n} and M = {1, 2, , m}. If f is an onto function from N to M, then the inverse images, f1(1), f1(2),, f1(k) constitute a partition of N into m classes. The number of ways in which this can be done is S m n , where the order of partition is immaterial. But, in functions, the order cannot be ignored. So, the distribution can be done in m!. S m n ways.

E9)

1 This is S n n . This can be done by putting one person each in n 2 rooms and 2 1 persons in 1 room. This can be done in C(n, 2) ways. So S n n = C( n ,2).

E10) If a positive solution is x1, x2, , xn, then it can be written as X1+1, X2+1,, Xn + 1, where the Xis are non-negative. Thus, the required number is the number of non-negative solutions of X1+X2++Xn+n = m, which is C(n + m n 1, m n) = C(m 1, m n). E11) This is the number of positive solutions of x1++x6 = 100. So, the required number is C(100 1, 100 6) = C(99, 94) = 71,523,144.
1 20 20 + P1000 + P1020 E12) P1000 .

Had the requirement been that there be exactly 20 bundles, then the number
20 . would have been P1000 1 2 3 4 5 5 + P10 + P10 + P10 + P10 = P15 E13) P10 .

71