C LA SSICAL H C nl~ IQ U[S

t BASIC COMBINATO RiAL NlJMIR',ERS

ho w many ways can the medab be distributed?
Le t, throughout this section, X stand for then-set {I, 2, .. . , n} and A stand fo r the m-s~t
I0 . The Reserve Bank of India prints currency notes in denominations of {a1, a 2, . . . , a,..}. Consider a map ef, : X - > A. There are m" such maps possible. The
fact th at
One Rup ee, Two Rupees, Five Rupees, Ten Rupees, Twenty Rupees, Fifty every such map allows several interpretations is fundamental to the techniques
of combi-
Rupees, a od O ne Hundred Rupees. In how many ways can it display ten oaiorics. The reader must have already noticed the parallel between Problems 2 and 4 and
cu rrency notes , not necessarily of different denomination s'/ How many of bet~veen Problems 3 and 5 of I-I (MF). First, let us ag ree to write ,f; as
th ese w ill ha ve all drnomination s?
l I. In how many ways can an employer distribute R s 100/- as Holiday ( I 2
Bonus to hu; 5 employees? No fraction of a rupee is allowed . Also, do not \,j,( 1) ef,(2) <f,(3) ,f,;nJ
wc,rry abo ut questions of equity and fairness!
12 . The re sults of ~O chess games (win, lose, or draw) have to be predict- Writing
ed How many different forecasts can contain exactly 15 correct results ?
13. How many distinct results can we obtain from one throw of four dice ?
!
j <f,(j) = a;,,
(1)

~
fivt: dice? Can you generalise this? we have
14 . In how many ways can 8 rooks be placed on a standard chess board
so that no rnok can attack another? How many if the rooks are labelled ? t ef> =Caii
2
II)
How would the amwer be modified if we remove the restriction that 'no
fO.)k can attack another'?
15. Show that there are 7 partitions of the integer 5, and 33 partitions of
'
i
i
a;,

wi th the a;'s belonging to A.

a;1 a;.

the m
Let now X be considered as a set of II objects to be distributed or sorted into
the integer 9. How many of these have 4 parts? How many have the largest
part equal to 4? Experiment with other partitions and other numbers.
I boxes a 1, a2 , • • • , a,.,. Then, ef, preoiscly describes a way of sorting the II objects
boxes. With this interpretation, (I) says that the object j goes into the box a;,·
way of sorting shall be called a c,o_mbi!l_atoria/ distribution or, simply, a distribution.
into the m
Each such

II
Example With n = 4 and A = {a, b, c}, the map

C ~ : :)
stands for the distribution which sorts
objects I and 3 into box a,
object 2 into box b,
object 4 into box c. )

The second interpretation for </, : X -> A needs the following:

l>e}inition 1.1 (__~n 11-sample of A or an n-tuple of A is defined as the

ordered multiset
'ordered 11-
{a;,, a;,, .. . , a;,}, where the ai's belong to A . (Note that we use 'n-tuple' for
tuple' .) Written as a string of symbols

••• , a,,, are

it ·is also called an -~~letter word ,C!r a word of /cngtli n. ) rhe symbols a 1 , a2 ,
they agree
called letters of the alphabet A} Two n-tuples arc consiflered to be the same iff
entry by entry. / '-- ·
 ~

C LASS!C-&L TECHNIQUES I BASIC COMBINATORlft L ;,· u;4BERS 1

With this definition, the map The statement that 4> is a surjection is equivalent to the requirement th;;;t the word 1.,
exhau~ts _the letters of the ~phabet A_. In oth~r ~or~s, the word ef, contains all the letten :r
,t,=f1,a b2 3 4) .4. T~1s 1s the same as _say1~g- that, m the d1stnbutton ef, of objects into boxe,, every bo~ a,
a c contam~ at_least one_ ot~ect~, 1.e., no box _could be empty. Among the rn" maps, how many
are sur,1ections? This question needs a httle more work before we can ariswer it. We shall
can be inteq,reted as standing for a word abac of length 4 out of an alphabet of 3 letters come back to it later.
a, b, c. Then, the set X = {l, 2, 3, 4} is the set of orders::d places in the word :.) On the other hand, ·t he number of injections</, : X ➔ A is easier to calculate. Clearly
prove that two sets '
The prnof of the following proposition is now easy. Note that t:i </, is an injection iff, in (I},
h.3ve the same cardinality is equivalent to proving that there exists a bijecticn ( =one-one
a,-1 = a;, => j = k,
correspondence onto) between them.
that is, the word</, contains distinct letters from the alphabet. In the interprelation of ef, as a
;' Propa~ition 1,2 There exists a bijection between any two of the following sets; distribution, the equivalent requirement is that every box contains at the most one object.
(a) the set of words of length n on an alphabet of m letters, In other words,(f~ an object is put in a box, no other object can be put into the same box.
Said this way, it is cailed the Exclusion Principle (EP):) We say that the distribution sati~fies
(b) the set of maps of an n-set into an m-set,
the EP. We are now ready for the following propo'1tioo which counts the injections from
(c) the set of distributions of n distinct objects into m distir_ct boxes, XtoA:
(d) the set of n-tuples on m letters,
· Proposition J.3 There: exists a bijection between any two of the following sets:
(e) the set of n-samples of an m-se~.
,, (a) the set of11-letter words with distinct letters out of an alphabet c-:' m letters,
The cardinality of each of these sets is m". ) I
(b) the set of injections of an n-set into an m-set,
(c) the set of distributions of n distinct objects into m distinct boxes, subject to
Remark In the above statement, we have used the word distinct without the EP,
furfaer explanation. It is to be understood in the context of sorting n objects
(d) the set of 11-tuples on m letters, without repetition,
. dra..,.n from an n-set. Recall that the objects of a set are. always taken to be
(e) the set of permutations of m symbols taken n at a tim_e (we caJI these m-per-
disrinct or distinguishable. An equivalent word for the purpose of combi-
natorics is labelled-mean ing, each object has a unique label and as such mutations of an n-set).
is distin~ishable from any other. The corresponding words which give the The cardinality of each d these sets is
opposite meaning are non-distinct, indistinguishabl e, u11label!ed. Sometimes, m(m - l)(m -- 2) ..• (m - n-;- 1).
we also use the word identical for these. The Problem Set 1-1 (MF) has
already used some of this terminology, hopefully without creating a con- If m < n, this number is taken to be zero.
fusion.
Proof The bijection is easily established between the above pairs of sets, in view of the
f Illastratioo Problem 8 of 1-1 (MF) can now be used to illustrate the bijec- discussion preceding the proposition. To calculate the number of elements in each set, Jet
\. tiCJ:l mentioned in Prop. 1.2. We have X = {l, 2, 3, 4, 5J and A= {O, l}. Ln be the list of all n--tuples on m letters without repetitions. To form Ln+i, we have to .add
A word of length 5 on the alphabet A is, for example, OllOl. The corres- to each 11-tuple the (11! + 1)-th letter from the remaining m - n letters. Any (n + 1)-tuple
ponding map from X to A is arises this way -and ~:o this process exhausts all (n + 1)-tuples. Further, there are no repeti-
tions of (n + 1)-tuplc:s obtained by this process, for, two distinct (n + !)-tuples either differ
1 2 3 4 s). in their first n letters, in which case they arise (rom distinct entries in L..,, or they dilfer in
(0 1 1 0 1 the last Jetter. Thus, if two (n + 1)-tuples arc the same, then-tuples from which they arise
are also the same. Hence, · ·
1:b.e 6orresponding distribution of objects in.to boxes is
IL,.+11 = (m ·- n)IL..1
1, 4 go into box 0, = (m -- n)(m - n + l)IL,...11
2, 3, 5 go into box 1.
/
The corresponding 5-tuple is 01101. Tlie 5-sample from A is {O, 1, 1, 0, l}. = (m -- n)(m - n + I) ••• (m - l)IL1I

•ma ,f, : X - A could be ai s11rjecti~~ (~o°:to map) or a~ injection

= (m -- n)(m - n + l) ... (m - l)m.

In g::neraJ, the PWh ·t is both, it is called a b1Ject1on, s WC already know.

r·

(=one-one map) or both. en i

S CL,1,:; s 1cA L TE<;H N IQUES I BASIC CO!>lBl N/\TORL'-L rlL::l::::.:.- :.

·T hu ~, Table 1 Values of s:
l)(m - 2) . .. (m - n + I). I 4 6
~
llnl = m(m - 0 2

factorial. Thus , ---- - - --

Notation We write the above number as [rn]. and call it a falling
0 0 0 0 0 0
[m]., = m(m - l)(m - 2) . .. (m - n + I). (2)
0 0 0
2 0 -I 0
see that the number
lllustra1ioos (a) Recalling Problem 3 of I-1 (MF), we 0 2 -3 0 0 0
is [9] 4 = 9 -8-7 -6
or 4-pcrmut ations of the 9-letter word UNIVER SAL -6 11 -6 0 0
0
= _3024. One such permutat ion is IERS. Here, X = {I, 2, 3, 4}, the set to
of 4
correspon ding 0
obJects . and A= {U, N, I, V, E, R, S, A, L}. The map 0 24 -50 35 - 10
the wo rd IERS is -225 85 -15
6 0 -120 274

2 3 S4).
I.]· l E R

The distributi on of objects into boxes is clear.

1-1 (MF) . The
(b) As another illustratio n, let us take Problem 9 of
[!Oh= 720 ways. Each such way of awarding the
medal s can be awarded in
medals into the Notation The number m(m + l)(m + 2) ... (m + n - I) is called a 'rising facto'rial'
medab correspo nds to a distributi on ,f, of the set of three
example, Team 5, and is denoted by [m]n. Thus,
set of learns {Team 1, Team 2, ... , Team 10}. If, for
and bronze-, the
Team 1, and Team 9 get the three medals-g old, silver, [rn]n = m(m + l)(m + 4) ... (m + n - 1). (4)
map cf, ts
into m distinct boxes with
Proposition I.I The number of distributi ons of n distinct objects
/Gold Silver Bi;onze ). the objects in each box arranged in a definite order is [m]n.
\Team 5 Team 1 Team 9 · ·-\
on, we shall take
,/
Discussion Before we take up the proof of this propositi
trade! No combi-
STIRLIN G NUMBE RS OF THE FIRST KIND the student-n ovice of combinat orics into the secret of the
having experi-
it as natorialis t begins to prove a theorem in combinat orics without
The falling factorial (x]. is a polynom ial of n-th degree in x. Let us write mented with a number of special cases. He goes from simple
to complica ted
ds how the argu-
situations in successive steps of abstractio n and understap
[x],. = ~ + s!x + s;x + ... + s:x•.
2
ments in the concrete special cases form building blocks for
the argument in
of all science,
of the first kind. To calculate their the general case. This may be true of all mathemat ics (and
Then, the numbers s! are called the Stirling numbers this process is
perhaps), but in combinat orics, where the universe is 'finite',
values, we first note that not only a must, but is very often the only way to construct
a proof or under-
combinat orics is
[xln+I = (x - n)[x] •. stand one, if it is already there. In fact, the crucial art of
ess! For the
to be able to reason a concrete case without using the concreten
recurrenc e relations for s! (a novice, therefore, a specific advice, which this author
is never tired of
Equating the co~fficients of xk both sides, we get the following
in Section 2):
formal definition of the term 'recurren ce relation' will come repeating , is:
of ten it
If you do not understand something in combinatorics, nine out
s:+i = s:-• - ns! with s~ = 0, s: = l.
(3)
is because you /lave not examined a sufficient number of special
cases.
of each
We postulate : One of the main objectives of the MF problems at the beginning
n of ideas in the
section is for them to serve as a springboa rd for generatio
show how an
s~ = O if k > n. general case. In the process of proving Prop. I .4, we shall
examinati on of the special cases is helpful.
Table 1 gives the first few values of/,:.
 14 C LASS IC AL TECHNJQ UFS I BASIC COMB , N ATO RIAL NU !,eu .:

d as asking Now, using the notation

In the same way, ~ro_ble~ 11 of 1-J (MF) could be considere
objects (one hundred
for the nu~ber of distribut ions of JOO non-distinct Tb
I Rupee coins, for example) among the five distinct
employee s. e answer (:),
C(m, 11) for
is

= we .have
[5] 100/ 100! 4598126.
(•) C(-m , n) = (-l)"C(r n , 11).
sets:
Proposit ion J.7 There exists a bijection between any two of the following rs and the C-numbers. This is
an ordered set of m lette,s, This last result(•) establishes a duality between the C-numbe
(a) the set of str'.ctl~ in~reasing words of length n from an example of combinatoria:/ reciprocity.
into m distinct boxes, with the ations (of an m-set) without
(b) the set of d1stnbut ions of n non-distinct objects Define the C-numbers as those which count n-combin
EP, n-combinations (of an m-set) with
repetitions. Define the C-numbe rs as those which count
(c) the set of 11-subsets of an m-set, values of either the C-oumbers or the
repetitions. Then, even without knowing the explicit
a time, called n-combinations. rs, we can conclude from(•), for example,
(~) t_he set of combina tions of m symbols, taken n at C-numbe
is taken t 0 b O ·f + 1, n) .
The cardmalJty of each of these sets is [m]./n! This number e I m < n and I C(m, n) = (-1)"<:( -m - n
if n = o.
m-set with repetitions is in bijectioL
B • For this, first note that tb.e set of n-combinations of an
without repetitions. (Prove this.)
with the set of n-combinations of an (m + n - 1)-set
bijectio
Proof It is left to the reader to exhibit the necessary
we get pns.. · ylperhmutmg the letters of
each strictly increasin g word in all possible wavs· ' recise Y t e set of all words of Therefor e,
·h · ·
1engt h n wit d1stmct letters. Now apply Prop. 1.3 to get the result.
I
C(m, n) = C(m +n- I, n)
D~jinition J.8 We define the binomial numbe r(;) as = (-l)"C( -m - n + I, n) from(•).

rrmJ./n! if n =I= 0, m ~ n This is the significance of the reciprocity law.

(:)=r ., ifn=O,m~O (5)
Remark If the reader thinks that the above is 'Much Ado about Nothing'
(!),
lO ifm < n This whole paper,
he is well advis,!d to refer to the paper by Stanley (l].
a 'further comment'
The other notation s for the above are "'C., c::,, C(m, n), {m} •. Also, we have of 59 pages, ac:cording to the author of the paper, is

(m)
on(•)!
[m]n m!
n = ,iJ = if m ~ n > 0,

[m]" = (m
(m _ n)! n!

+ n - n!I)! = (m + nn - I) (6) f
STlRLIN G NUMBERS OF THE SECOND KIND
objects into m non-distinct box«:_s, no
n! (m-l)! ifm>l ,n>O. ( Let n ~ ·m. The number of distribut ions of n distinct
It is called a Stirli11g number of the second kind~,The
box being empty, is defined as S::'. ,.-
by the binomial expansions numb.ers S'; are in some way omnTpresent in combinatorics.
We note the parallel between [m],, and [mr as given
objects into m classes, each
( The sorting-here. is a process of partition ing the n distinct
(1 + t)" = 1 + E[nJ,.,, (I - t)_,, = t + .E [n]
r.1
',,.
class being non-emp ty; We shall now make this 'partition
ing' concept pcecise.
r! ' of X (to be more expiicit,
· The subsets Ai, ·A2 , • • • , Ap of X are said to form a partition
when m is a negative set-partit ion of X) if
The binomia l coeffic ients(:) have no combina torial meaning
(6) to negative integers via the falling factorial thus: if
-m A 1 if:. <f, for any i,
integer. Howeve r, we can extend
is a negative integer, then iif:.}=>A 1 n A1 = </,,
-m)- [-m],, _ (-mX- m-1) •.. (-m - n + 1) A1 U A2 U ... U Ap = X.
( - n! - , n!
n . '.fhe number of such partition s
The sets A 1 are called the classes or blocks of the partition
·= (-1:i[m ]" = (-1)"(m +:- 1)- of an n-set X into m noii:cmp ty blocks is the same as~. ;.
'
~;'_/As.an illustrati on, the partition-s
,

lllliii' llrnw lUilsS MIII II~~ ..;;g ;;; a;g,g; ; cc

 I BAS IC COMBINATORIA L NUM BERS 17
j (: CLASS!CAL TECH~IQUES

r of a 4-set {a, b, c, d} into 3 class~ are as follows-with a misuse of notation, but

ing heing obvious:
a I b I cd, a Id I be, c Id I ab,
the mean- f"roposi tiora 1 .9

Proof
T he number of su rjcctions of the n-sct into the m-set A ism\
Let cf, : X -+ A be a surjcctioo. Then,
s~n

b / c / ad, a I c I bd, !JI d I ac. is a set-pa rtition of X into m non- empty classes. Conversely, to each set-partition
of X into
m non-empty classes , there correspond m! surjections. I
Thus, Sl = 6.
{.._The number ~ S; + +... +
S;:' then gives the number of all partitions of a set of n
into m
From this and Prop. 1.3, we get the following :
objects. It gives the number of ways in which n distinct objects can be distributed = n, the number of bijections of X into A is
by Proposition J .JO When m
non-distinct boxes. If m = n, this number is called the n-th Bell number and is denoted
Bn. Thus,
(7;)11! = n! = n! s:. B
Bn = S~ + S + ... + S~. (7)
into
Let us consolidate what we know so far regarding the distribution of objects
Rtcarrence Formula for S;:1
number of ways in boxes. We shall use the following symbolism:
H 1 < m < n, S::'+ 1 can be broken as the sum of two parts: one, the 1
which a certain object i is the sole member of a class, which is s:;i- ; and two, the number DO : Distinct Objects
we have
ofways in which that object i is not the sole member of a class, which is mS:;". Thus, NDO : Non-Distinct Objects
th.c recurrence formula:
DB : Distinct Boxes
SZ'+1 = s:-• + mS~ if 1 < m < n. NDB : Non-Distinct Boxes
We postulate: EP : (With) Exclusion Principle
(8) : (With) No Exclusion Principle
NEP
s1 = 1 = s~, S;:' = 0 ifn > m,
are
sr;: = ~ = O > The number of distributions of n objects into m boxes with the various specifications
S8= I, if m, n O.
are summarised in (9):
The calculations for S;:' c:in now be systematically made and the first few results
a5 in Table 2. DO DB NEP : m" [Prop. 1.2]

Table 2 Values of S;:' and Bn DO DR EP : [m]n (:)n!

= [Prop. 1.3]

NEP : [m]"/n\ = (m+n-1)

4 5 6
~
2 3 B,,
NDO DB [Prop. 1.6]
11
(9)
0 0 0 0 0 1
2 NDO DB EP : [m]n/n! = (:;) [Prop. 1.7)
I 0 0 0 0
2
3 3 I 0 0 0 s~ +s; + .. . +s;· [definition of S::']
0 0 15 DO NDB NEP:
4 7 6 1 Bn ifm = 11 [in view of (8))
0 52
5 15 25 10 I if,1 =m
65 15 213 DO NDB EP
6 31 90 1 0 ifn =F Ill

Now we are
.
ID a
..
pos1tl\ln o
t
c
alculate the number of surjections of X into A.
I
'!
What remains now; in terms of this classification, is the distribution of n non-distinct
objects into 111 non-distinct boxes , with or without EP . With each such distributio!l
no box is empty, we can associate the 111-tuple (,\ 1, >. 2 , ••• , Am) satisfying
for which

(10)

I
I
18 CLASS ICAL TECHN IQU J'S
i H A ~a r· CO iV! 0 l NATO~ f f . L NIJ U St.~ .; r~

S uch an m-tuple is called a partition of the integer n. The partition above is usually written
When a symbol A .Hands for a partition , it is n,·c<=snry to distingui sh be twe,·,i A, / ;_,,
also as ,\ 1,\ 2 • •• ,\m without commas or parentheses. The number of partitions of the integer
A3, - . . '. and l'•2'•J', . . . . Except where we state oth erwi se , we sbal! , in general, adopt, rle
ri into exactly m classes (ca lled parts in this context) is denoted by P::' . Hence, the numbe r of
convention that
distributions of n non-distinct objects into m identical boxes (boxes could be empty or could
have one or more objects) is
P! + P; + ... + P'; .
This is the total number of partitions of n into m or fewi;_r parts. If rn = n, this number is
Recm-rencc Formula for P';
denoted by p(n), the number of all partitions of the integer n:,So we can continue the tabula-
tion made in (9) :l!nd write: · Let E be the set of partitions of n having m or fewer parts. Each such partitio n may be
considered as an m-tuple. Define on Ethe mapping
Number of distributions
is f : (<X1, at2, <X3, •• • , IXk, 0, 0, . .. • 0)
with spec(fications
P! + P; + .. . + p;,• ➔ (oc, + I, <X2 + 1, .. . , IXk+ I, I, I, . . . , I).
NDO NDB NEP
The latter is a partition of n + m into
exactly m parts. Call the set of such partitions E'.
p(n) ifm = n
( I I) The map f: E ➔ E' i!: a bijection, for, (a) every m-tuple of£' is the image of an m-tuple of
ifn = m · E and (b) distinct m-tuples of E are mapped into distinct m-tuples of£'. Hence,
NDO EP
NDB
0 if n *m 1£1=1£'1-
This gives the recurre1ace relation
Example 1.J I
P!+ P;,+ .. . -I- P::' = P::'+m,
The partitions whence (12)
are
of P! =I= P:.
2 2, 11 Pl= I = Pi With the help of (12) , the first few values of P'; have been calculated in Table 3.
3, 21, I I I P! = p~ =Pl = I Teble 3 Values of P:;' 2nd p(n)
4,31,22,2}1., 1111 Pl = Pl = p~ = I, Pi = 2
'--,"--..m 2 3 4 6 7 8 9
3,1• P~ = P~ = I, P~ = P~ = 2, P! = I p(n)
5, 41, 2111, 1111 I n '---..
0 0 0 0 0 0 0 0 I
Thus, p(2) = 2, p(3) = , 3, = 5, p(S) = 7, and so on. Srinivasa Ramanujan
iathematician, guessed and proved many remarkable 2 0 0 0 0 0 0 0 2
(1887-1920), the self-taught Ind1
curiosities about the function p(n). l I 0 0 0 0 0 0 3
If,\= (,\ 1 , ,\2 , ••• , ,\p) is a pa, ion of n, we write this fact as,\ I- n. Note that, unless
2 1 0 0 0 0 0
the contrary is indicated, the convc. don of (JO) is valid, that is, the parts are in non-
5 2 2 0 () 0 0 7
increasing order. ,\ may also be writtetJ as
6 3 2 0 0 0 11
7 4 2 0 0 IS
where r, is the number of parts equal to i in A, i varying in {I, 2, 3, ... }. For example, all 8 4 5 3 2 1 0 22
the following denote the same partition of 17: 9 4 7 6 2 30

4+3+3 + 2 + 2 + 2 + ),
3 2
(conden.sed as 43 22 31; in the other notation, it is 12 3 4).
43322.21
 20 CLASSICAL TECHNIQUES

I BASIC COMBINATORIAL NUMBERS 21

. If,\ = (i\i, Ai, · · · • r..P) f- 11, the Ferrers ra J , .
is defined as an arrangement of d t ( II g Pr or ~he } ourrg diagram associated with ,\ with the_ i boxes of each of the s1 classes being indistinguishable among themselves.
the i-th row. As an illustration ~hs Fca ed nodes), Ill /J left justified rows with ,\; dots in With X and A as above, we now look at the combinatorial distribution X - A. Two
' e errers graph of 54431 I is ·
· given · ·
T1, T2, . •• , Tq be the lengths of the col f . Ill Diagram I . If
~uc~ combinatorial di~tributions are said to be equivalent if they are indistinguishable, that
of n is called tire co.qjugate partition to ~m~s o s~ch a diagram, the partition h, ,·2, . .. , ,·q) is, ,r one can be obtamcd from the other by a permutation of objects of the same kind
,\' is obtained by flipping the diagr ~• o ,\ "· It is usua_lly denoted by .\'. The diagram for
( = class) or of boxes of the same kind ( = class). The equivalence classes of distributions
conjugate to ,\ = (54431 I) given amb or. along the diagonal. For instance, the partition which arise are called pa/ferns.
· , . a ove 1s as shown JO · o·
Cl early, 1f ,\ 1s conj·Joate to ,\ th , . . 1agram 2. Hence, ,\' = (64431)
Let R'(l''2'• . . • ,f•; 1·' 2 • . . . ms") be the number of patterns relative to the distribu-
1 1
., , en " IS CODJUgate to ,\' . .
tions of a collcctioo X of objects of type 1' 12'1 • • • ,f• into a collection A of boxes of type
15 1 211 •• • m1 • .
Let R( 1''2'' . . . n'•; P 1 2'• .. . m'·) be the number of the above patterns in which no
box is empty.
Now, if,\ = (.\,, .\2, .. . , Ap) f- rn, then one cao easily see that the R' -number is obtained
as the sum of the various R-numbers as

R'(I ' 12'• ... n'•; .\1.\2 . . . i\p) = I: R( 1' 12'' .. . rf• ; k 1k 2 • •• kp). (13)
~k,<;),

This relation expresses the R'-numbers in terms of the R-numbers. An inverse relation which
cxpn~sscs the R-numbers in terms of the R' -numbers is desirable and also possible. But, to
Diagram 1 obtain this, we have to 'invert' (13), and for this, a powerful process, called Mobius inver~
Diagram 2
. sion, is required (sec Section VII-2). ln the mean time, the following proposition formalises
. The Ferrers g~ph is not just a visual hel It .
partitions, evidence for which . thp. IS an unportant tool in the investii:a- some special cases of the R-numbers and R'-oumbers which can be easily calculated by
twos of <nic
is e proof of the following: means of tbe elementary combinatorial nambers already introduced:
Propositi<>n I 11 P':' - number o f part1t1ons · ·
of n into parts the largest of wb · b .
p if If,\
• n -

.
IC IS Ill. / ' Proposition I .I 3
roo · = (Ai, A2, · • - , A,,,), its Ferrers graph has m Th ( R(I" ; m) = S;:';
(14)
then m as its largest narL The map.>. ,\' . b .. . rows. e con1ugate partition ,\' IJas
into m parts and the -set of partitions o➔ f I~ ah I 1Ject100 between th,e set of partition s of ,, R'(l"; m) = s!, + S. + .. . + S::'; (15)
n Wit argest part m. I
R(I"; !'") = m! S;:'; (16)
We shall see more about partition~ later.
R'(l"; Im) = m"; (17)

!
1;£ ,\ - l
( ) -
• .
r, r, r, PA ~RNS OF DISTRIBUTIONS
2 3 . . . r.' 1s a part.Jt1on of n and !XI -- n , a set-part1t1on
·· · R(n; m) = P:0' ; (18)
"--- _ of X which contains
r1 classes of cardinality I each (19)
R'(n ; 111) = P~ + P~ + . .. + ,0:;
r2 classes of cardinality 2 each
R(rr I"') =
'
cl - l)
111 -
·
I'
(20)

r,. classes of cardinality n each

1 1 (21)
R' (11 ; I"') = ("+ ::- ).
is said to be a parlr"tion of type V Y• . .. n'•. X itself is said to be a collection of ty r,
1

ir•2r, . .. nr•) Sup?ose the objects of this co!lection have to be sorted into tbe box~s
which themselves form a collect100 A of type 1s,2s, . .. m 1 -, that is, there are
Proof R(I"; 111) (respectively, R(11 ; 111)) expresses the number of ways in which II distinct
01, a2 , . . .__ ,.,.am
(respectively, non-distinct) objects can be sorted into exactly n, indistinguishable boxes, and
s 1 classes of boxes each containing I box so, by the definition, is S: (respectively, P:'). Th1s gi,~ (\4) and (18). The corresponding
numbers, with no rcstTictions on the empty boxes., are available as the results for the cases
s2 classes of bous each containing 2 boxes

.
sm classes
.
of boxes each containing m boxes
DO NOB NEP NDO NOB NeP
of (9) am\ (11), and hence, we get (15) and (19) . R ' (I"; I"') is just the number of ways of
 CLASSICAL TECHNIQUES

. . P ROBLEM SET 1-1 (DAD)

saning n distinct objects into m dislinct bo
R(I•; Im) is the
correspond ing number with no box empty !bx~s,_ a~d _this is mn. Hence (17). l. ln how many ways can 5 men and 5 women be seated in a roUJl
o tab!~
from then-
set X in_to_ the m-set A. Now, Prop. J .8 i~es alt• it, is _1h~' n_umber of surjections if no two women may be seated side by side?
n non-d1stmc t objects into m distinct i:JO;e s. T(h_is >:is Rh(n, 1 ) is the number of ways of sorting
t e case 2. Six generals propose locking a safe containing top secret with a numbei
NDO DB NEP of different locks. Each general will be given keys to certain of these locks.
How many locks are required and how many keys must each general
of (9). Hence (21). ha\>e so that, unless at least four generals are present, the safe raonot be
It remains to prove (20) R(n· I ·" ) is t::ie b opened?
er of ways in which n non-distinct ob'}·ects
~an be distributed into m di~tincr' bo xes suehnumh t at no bo · . 3. How many integers between l000 and 9999 inclusive have Jislinct
d" . . o .. x JS empty: Smee the objects are
m istingu1shable, our only concern is abouc the digits? Of these, ho\V many are even numbers? How many consist entirely
box. So R(n; Jm) is the number of soJutio~ p s1tive number of obJects that get into each
U1, u2, ... • um of U1 + U2 + .. . + um = n such
of odd digits?
that the u,'s are positive integers Each solut"ion
+ · corresponds to a d 4. In how many ways can 9 distinct objects be placed in 5 distinct boxes i:l
v _
k - ui • a2 + •· • + uk and 1 ~ v 1 < v, < <
~
wor V1V2 • • - Vm-1, where
such a way that 3 of these boxes would be occupied and 2 would be empty?
leads umquely to a solution required He - ~ .R( .. 1 vm_-1 n - I . Conversely, such a word
the (n - 1)-set (I, 2, 3, ... '(n - l)}. lnc:J.I n, m) is the number of
(m - !)-subsets of 5. In how many permutation s of the word AUROBIND do the vowels
:,-1 appear in the alpha·:,etical ~rder?
.
6. There is an unlimited supply of weights of integral numbers of grams
Using n or fewer weights, find the number of ways in which a weight of m
grams can be obtained. Prove that there is a bijection of the set of all such
ways on the set of increasing words of length n - I on (m + 1) ordered
le-tiers. ·
7. How many dirtinct solutions are there of x + y + z + w = 10 (a) in
positive integers. and (b) in non-negative integers?
8. A train with n passengers aboard makes m stops. In bow many ways
can the passengers distribute themselves a~ong these m stops as alighting
passengers? If we are concerned only with the number of alighting
passengers at each stop, how would the answer be modified?
9. There are 16 books on a bookshelf. In how many ways can I: of these
books be select,ed ff a selection must not include two neighbourin i boob?

IO. Show that there are (n t 5) distinct throws of a throw with nnon-
<listinct dice.
11. Given n indistinguishable objects and n additional distioct objects-
also distinct fr,om the earlier n objects-, in how many ways can we choose
n out of the 2n objects?
12. Establish the following relations:

,(a) Bn+I
\/
(kn)sk.
i:
= k-0
{b) .E (1:)( q ) = (P + q).
k le n-k n