Internat. J. Math. & Math. Sci.

625
Vol. i0 No. 4 (1987) 625-640

SRINIVASA RAMANUJAN (1887-1920)AND THE THEORY OF

PARTITIONS OF NUMBERS AND STATISTICAL MECHANICS
A CENTENNIAL TRIBUTE

LOKENATH DEBNATH

Department of Mathematics
University of Central Florida
Orlando, Florida 32816, U.S.A.

(Received June 5, 1987)

ABSTRACT. This centennial tribute commemorates Ramanujan the Mathematician and
Ramanujan the Man. A brief account of his llfe, career, and remarkable mathematical
contributions is given to describe the gifted talent of Srinivasa RamanuJan. As an
example of his creativity in mathematics, some of his work on the theory of partition
of numbers has been presented with its application to statistical mechanics.

KEYWORDS AND PHRASES. Partition of Numbers, Congruence Identities, Asymptotic

formula, Restricted and unrestricted Partitions, Bose-Einstein statistics, Fermi-Dirac
statisitics, Maxwell-Boltzmann statisitlcs, Gentile statistics.
1980 AMS SUBJECT CLASSIFICATION CODE. 01A70, IIP57, IIP72, lIPS0, 82A05, 82A15.

I. INTRODUCTION
Srinivasa Ramanan is universally considered as one of the mathematical geniuses
of all time. He was born in India a hundred years ago on December 22 of that year.
His remarkable contributions to pure mathematics placed him in the rank of Gauss,
Galois, Abel, Euler, Fermat, Jacobi, Riemann and other similar stature. His
contributions to the theory of numbers are generally considered unique. During his
life-time, Ramanujan became a living legend and a versatile creative intellect. His
name will be encountered in the history of mathematics as long as humanity will study
mathematics.

Ramanujan was born on December 22, 1887 in Brahmin Hindu family at Erode near
Kumbakonam, a small town in South India. His father was a clerk in a cloth-merchant’s
office in Kumbakonam, and used to maintain his family with a small income. His mother
was a devoted housewife and had a strong religious belief. However, there was no
family history of mathematical or scientific genius.

At the age of seven, young Ramanujan was sent to the high school of Knbakonam
and remained there until he was sixteen. He was soon found to be a brilliant student
and his outstanding ability had begun to reveal itself before he was ten. By the time
Ramanujan was twelve or thirteen, he was truely recognized as one of the most
outstanding young students. He remained brilliant throughout his life and his talent
and interest were singularly directed toward mathematics. Like Albert Einstein,
Ramanujan became entranced by an elementary text book entitled A Synopsis of
Elementary Results in Pure and Applied Mathematics by George Shoobridge Cart. No
doubt that this book has had a profound influence on him and his familiarity with it
626 L. DEBNATH

marked the true startlng-polnt of his mathematical discovery. In 1903, RamanuJan

passed the Matriculation Examination of the
Madras and Joined the
University of
Goverenment College at Kumbakonam in 1904 with the Subrahmanyan Scholarship which is
usually awarded to students for proficiency in Mathematics and English. At the
College he used to spend most of his time studying mathematics. His consequent
neglect of his other subjects resulted in his failure to get promotion to the senior
class. Consequently, he lost his scholarship. He was so disappointed that he dropped
out from the college. In 1906, he entered Pachalyappa’s College in Madras and
appeared as a private student for the F.A. Examination in December 1907 and
unfortunately again failed. He was very disappointed but continued his independent
study and research in pure mathematics.

During the summer of 1909, RamanuJan married Janakl and it became necessary for
him to find some permanent job. Being unemployed for about six years, he accepted a
small job in 1912 at the Madras Port Trust as clerk. He has now a steady job, and he
found he had enough time to do his own research in mathematics. He had already
published his first paper in the Journal of the Indian Mathematical Society in the
December issue of volume 3, 1911. During the next year, RamanuJan published two more
paers in volume 4 (1912) of the same Journal.
At the advice of his teacher and friend, Seshu Aiyar, RamanuJan wrote a letter on
January 13, 1913 to famous British mathematician G.H. Hardy, then Fellow of Trinity
College, Cambridge. Enclosed also in this letter was a set of mathematical results
incuding one hundred and twenty theorems. After receiving this material, Hardy
discussed it with J.E. Littlewood with regard to RamanuJan’s mathematical talent. At
the beginning Hardy was reluctant, but impressed by RamanuJan’s results on continued
fractions. Finally, Hardy decided to bring RamanuJan to Cambridge in order to pursue
some Joint research on mathematics. RamanuJan was pleased to receive an invitation
from Hardy to work with him at Cambridge. But the lack of his mother’s permission
combined with his strong Hindu religion prejudices forced him to decline Hardy’s
offer. As a result of his further correspondence with Hardy, RamanuJan’s talent was
brought to the attention of the University of Madras. The University made a prompt
decision to grant a special scholarship to RamanuJan for a period of two years. On
May I, 1913, the 25 year old RamanuJan formerly resigned from the Madras Port Trust
Office and Jolned the University of Madras as a research scholar with a small
scholarship. He remained in that position until his departure for Cambridge on March
17, 1914.
During the years 1903-1914, Ramanuj an devoted himself almost entirely to
mathematical research and recorded his results in his own notebooks. Before his
arrival in Cambridge, RamanuJan had five research papers to his name, all of which
appeared in the Journal of the Indian Mathematical Society. He discovered and/or
rediscovered a large number of most elegant and beautiful fornmlas. Thes results were
concerned with Bernoulli’s and Euler’s numbers, hypergeometric series, functional
equation for the Riemann zeta function, definite integrals, continued fractions and
distribution of primes. During his stay in Cambridge from 1914 to 1919, RamanuJan
worked ontinually together with Hardy and Littlewood on many problems and results
 SRINIVASA RAMANUJAN AND PARTITIONS OF NUMBERS 627

conjectured by himself. His close association with two great mathematicians enabled
him not only to learn mathematics with rigorous proofs but also to create new
mathematics. Ramanujan was never disappointed or intimidated even when some of his
results, proofs or conjectures were erroneous or even false. Absolutely no doubt, he
simply enjoyed mathematics and deeply loved mathematical formulas and theorems. It
was in Cambridge where his genius burst into full flower and he attained great
eminence as a gifted mathematician of the world. Of his thirty-two papers, seven were
written in collaboration with Hardy. Most of these papers on various subjects took
shape during the super-productive period of 1914-1919. These subjects include the
theory of partitions of numbers, the Rogers-Ramanujan identities, hyper-geometrlc
functions, continued fractions, theory of representation of numbers as sums of
squares, Ramanujan’s Y-function, elliptic functions and q-serles.

In May 1917, Hardy wrote a letter to the University of Madras informing that
Ramanujan was infected with an incurable disease, possibly tuberculosis. In order to
get a better medical treatment, it was necessary for him to stay in England for some
time more. In spite of his illness, Ramanujan continued his mathematical research
even when he was in bed. It was not until fall of 1918 that Ramanujan showed any
definite sign of improvement. On February 28, 1918, he was elected a Fellow of the
Royal Society at the early age of thirty. He was the first Indian on whom the highest
honor was conferred at the first proposal. Niel Bohr was the only other eminent
scientist so elected as the Fellow of the Royal Society. On October 13, 1918, he was
also elected a Fellow of the Trinity College, Cambridge University with a fellowhslp
of 250 a year for the next six years. In his announcement of his election with the
award, Hardy forwarded a letter to the Registrar of Madras University by saying, "He
will return to India with a scientific standing and reputation such as no Indian has
enjoyed before, and I am confident that India will regard him as the treasure he
is." He also asked the University to make a permanent arrangement for him in a way
which could leave him free for research. The University of Madras promptly responded
to Hardy’s request by granting an award of 250 a year for five years from April I,
1919 without any duties or assignments. In addition, the University also agreed to
pay all of his travel expenses from England to India. In the meantime, RamanuJan’s
health showed some signs of improvement. So it was decided to send him back home as
it deemed safe for him to travel. Accordingly, he left England on February 27, 1919
and then arrived at Bombay on March 17, 1919. His return home was a very pleasant
news for his family, but everybody was very concerned to see his mental and physical
conditions as his body had become thin and emaciated. Everyone hoped that his return
to his homeland, to his wife and parents and to his friends may have some positive
impact on his recovery from illness. Despite his loss of weight and energy, RamanuJan
continued his mathematical research even when he was in bed.

In spite of his health gradually deteriorating, RamanuJan spent about nine months
in different places including his home town of Kumbakonam, Madras and a village of
Kodumudi on the bank of the river Kaveri. The best medical care and treatment
availalbe at that time were arranged for him. Unfortunately, everything was
nsuccesful. He died on April 29, 1920 at the age of 32 at Chetput, a suburb of Madras,
surrounded by his wife, parents, brothers, friends and admirers.
628 L. DEBNATH

In his last letter to Hardy on January 12, 1920, three months before his death,
RamanuJan wrote: "I discovered very interesting functions recently I call ’Mock’
O-functlons. Unlike ’False’0-functlons (studied by Professor Rogers in his interesting
paper) they enter into mathematics as beautifully as the ordlnaryO-functions. I am
sending you with this letter some examples." Like his first letter of January 1913,
Ramanujan’s last letter was also loaded with many interesting ideas and results
concerning q-serles, elliptic and modular functions. In order to pay tribute to
Srlnlvasa Ramanujan, G.N. Watson selected the contents of RamanuJan’s last letter to
Hardy along with Ramanujan’s five pages of notes on the Mock Theta functions for his
1935 presidential address to the London Mathematical Society. In his presidential
address entitled "The Final Problem: An Account of the Mock Theta Functions", Watson
(1936) discussed Ramanujan’s results and his own subseqent work with some detail. His
concluding remarks included: "Ramanujan’s discovery of the Mock Theta functions makes
it obvious that his skill and ingenuity did not desert him at the oncoming of his
untimely end. As much as his earlier work, the mock theta functions are an
achievement sufficient to cause his name to be held in lasting remembrance. To his
students such discoveries will be a source of delight Clearly, RamanuJan’s
contributions to elliptic and modular functions had also served as the basis of the
subsequent developments of these areas in the twentieth century.

2. RAMANUJAN-HARDY’S THEORY OF PARTITIONS

As an example of RamanuJ an’ s creativity and outstanding contribution to

mathematics, we briefly describe some of his work on the theory of partitions of
numbers and its subsequent applications to statistical mechanics. Indeed, the theory
of partitions is one of the monumental examples of success of the Hardy-RamanuJan
partnership. Ramanujan shared his interest with Hardy in the unrestricted partition
function or simply the partition function p(n). This is a function of a positive
integer n which is a representation of n as a sum of strictly positive integers.
Thus p(1) I, p(2)
2, p(3) 3, p(4) 5, p(5) 7 and p(6) II. We define p(0)
I. Thus the map n p(n) defines the partition function. More explicitly, the
unrestricted partitions of a number 6 are given as 6=I+I+I+I+I+I= 2+2+2= 2+2+I+I
2+I+I+I+I=3+3-- 3+2+I= 3+I+I+I= 4+2= 4+I+I= 5+I. Hence p(6) II. There are three
partitions of 6 into distinct integers: 6 5+1 4+2. There are four partitions of
6 into odd parts: 5+1 3+3 3+I+I+I I+I+I+I+I+I.
The number 6 has only one
partition into distinct odd parts: 5+I. We also note that there are 4 partitions of 6
into utmost 2 integers, and there are four partitions of slx into integers which do
not exceed 2. And there are 3 partitions of six into 2 integers and there are equally
3 partitions of 6 into integers with 2 as the largest.

It follows from the above examples that the value of the partition function p(n)
depends on both the size and nature of parts of n. These examples also lead to the
concept of restricted and unrestricted partitions of an integer. The restrictions may
sometimes be so stringent that some numbers have no partitions at all. For example,
I0 cannot be partitioned into three distinct odd parts.

There is a simple geometric representation of partitions which is usually shown

 SRINIVASA RAMANUJAN AND PARTITIONS OF NUMBERS 629

by using a display of lattice points (dots) called a Ferret graph. For example, the
partition of 20 given by 7+4+4+3+I+I can be represented by 20 dots arranged in five
rows as follows:

Reading this graph vertically, we get another partition of 20 which is

6+4+4+3+I+I+I. Two such partitions are called conjugate. Observe that the
part in either of these partitions is equal to the number of parts in the other. This
leads to a simple but interesting theorem which states that the number of partitions
of n into m parts is equal to the number of partitions of n into parts with m as the
largest part. Several theorems can be proved by simple combinatorial arguments
involving graphs.

Above examples with the geometrical representation indicate that partitions have
inherent symmetry. In quantum mechanics, such geometric representations of partitions
are known as Youn Tableaux which was introduced by Young for his study of symmetric
groups. They were also found to have an important role in the analysis of the
symmetries of many-electron systems.

The above discussions also illustrate some important and useful role of the
partition function from mathematical, geometrical and physical points of view. In
additive questions of the above kind it is appropriate to consider a power series

F(x)
n=O
.
generating function of p(n) defined by

p(n)x Ixl <

From this elementary idea of generating function, Euler formulated the analytical
(2.1)

where p(o)
F(x)--
m
(l-x m) .
theory of partitions by proving a simple but a remarkable result:

n=o
p(n)x n, Ix{ < (2.2)

-
If 0 < x < and an integer m > and

m -I
F (x) (I -x k) + Z p(n) x n, Ixl < (2.3)
k=l n=0

then it can be proved that

Pm(n) <_ p(n), Pm(n) p(n), 0 <_ n <_ m, (2.4ab)

and
630 L. DEBNATH

lim Pm(n) p(n) (2.5)

m/

Furthermore

lim F (x) F(x) (2.6)

m
m
Euler’s result (2.2) gives a generating function for the unrestricted partition
of an integer n without any restriction on the number of parts or their properties
such as size, parity, etc. Hence the generating function for the partition of n into
parts with various restriction on the nature of the parts can be found without any
difficulty.

For example, the generating function for the partition of n into distinct
(unequal) integral parts is

F(x) (l+x) (l+x 2) (l+x 3) H (l+x m) (2.7)

This result can be rewritten as
m--I

2 4 6
-x -x -x
F(x)
2 3
-x -x -x

II (1-x 2m-l) (2.8)

(l-x) (l-x 2) (l-x3)...m--I
Obviously, the right hand side is the generating function for the partition of n
into odd integral parts. Thus it follows from (2.7) and (2.8) that the number of
partitions of n into unequal parts is equal to the number of its partitions into odd
parts. This is indeed a remarkable result.

Another beautiful result follows from Euler’s theorem and has the form

-
(1-x 2)
(l-x 4) (l-x6) l+xl+ x3+ x6+ xl0+ (2.9)
(l-x) (l-x 3) (l-x 5)
n
The powers of x are the familiar triangular numbers,
An n(n+l) that can be
represented geometrically as the number of equidistant points in triangles of
different sizes. These points form a triangular lattice. As a generalization of this
idea, the square numbers are defined by the number of points in square lattices of
increasing size, that is, I, 4, 9, 16, 25

We next consider the partition function generated by the product

(l-x m) which is the reciprocal of the generating function of the unrestricted
m=l
partition function p(n) given by (2.2). This product has the representation

m--I
where
(l-x m)

Pe(n)
I+

is
[
n=l

the
Pe(n)
partition
Po(n)}x
of n
n
n=

into
. an
(-I) nxm(n)

even number of distinct

(2.10)

parts, and

Po(n) is the partition of an odd number of distinct parts, and the integers
m(n) (3n2- n) are called the Euler pentagonal numbers which can be represented
geometrcally by the number of equidistant points in a pentagon of increasing size.
 SRINIVASA RAMANUJAN AND PARTITIONS OF NUMBERS 631

These points form a pentagonal lattice. Also, it follows from (2.2) and (2.10) by
actual computation that

F(x) (2.11)
(I x m)
2
l-x-x + x5+ x x
12
x 15+
m

In view of the fact that the generating functions for the partition of numbers
involve infinite products which play an important role in the theory of elliptic and
associated functions (Dutta and Debnath, 1965).

Ramanujan made some significant contributions to the theory of partitions. He

was not only the first but the only mathematicians who successfully proved several
remarkable congruence properties of p(n). Some of his congruences are

p(5m + 4) 0 (mod 5) (2.12)

p(7m + 5) 0 (mod 7) (2.13)

p(llm + 5) --0 (mod II) (2.14)

All these results are included in his famous conjecture: If p--5,7 or II and
24n- --0 (mod pa), > I, then

p(n) 0 (rood pa) (2.15)

This was a very astonishing conjecture and has led to a good deal of theoretical
research and numerical computation on congruence of p(n) using H. Gupta’s table
(1980) of values of p(n) for n < 300. However, S. Chowla found that this conjecture
is not true for n-- 243. For this n, 24n-I 5831 (mod but =- 73
p(243) 133978259344888 0 (mod 72
0 (mod 73 (2.16ab)

Subsequently, D.H. Lehmer (1936) became deeply involved in the proof of the conjecture
and also in the computation of p(n) for large n. G.N. Watson (1938) proved
RamanuJan’s conjecture for powers of 7. Finally, A.O.L. Atkin (1967) settled the
problem by proving the conjecture for powers of II. RamanuJan’s conjecture can now be
a b
stated as an important theorem: If 24n-I 0 (mod dr5 7 llC), then

pCn) 0 (rood d) (2.17)

In connection with his famous discovery of several congruence properties,

RamanuJan also studied two remarkable partition identities:
5
[ pC5m + 4) x m 5
(x5)" (2.18)
6
m=0 (R)(x)
632

where (x)
.
=0
p(7m + 5) x

(l-x n)
m
7
’(x 7
(x
L. DEBNATH

’4
3
+ 49x
((x)
(x)
7
8
7
(2.19)

(2.20)
n

The functions on the right side of (2.18) (2.19) have power series expansions with
integer coefficients, Ramanujan’s congruences (2.12) (2.13) follow from these
identities. Subsequently, these identities have created a tremendous interest among
many researchers including H.B.C. Darling, L.J. Mordell, H. Rademacher and H.S.
Zuckermann. They proved RamanuJan’ s identities by using the theory of modular
functions. Proofs without modular functions were given by D. KruyswlJk (1950) and
later by O. Kolberg. The method of Kolberg gave not only the RamanuJan identities but
many new ones.

Actual computation reveals that the partition function p(n) grows very rapidly
with n. D.H. Lehmer (1936) computed p(n) for n 14,031 to verify a conjecture of

_
Ramanujan which asserts that p(14,031) 0 (mod 114). This assertion was found to be
correct. This leads to the question of asymptotic representation of p(n) for large
n. ,During the early part of the 20th century, Hardy and RamanuJan made significant
progress in the determination of an asymptotic formula for p(n). Using elementary
arguments, they first showed

log p(n) ,--, + 0 (n) as n (2.21)

Then, with the aid of a Tauberlan Theorem, Hardy and RamanuJan (1918) proved that

2
p(n)
4n/3
exp [II [--) as n (2.22)

This is one of the most remarkable results in the theory of numbers. Equally
remarkable was Hardy and RamanuJan’s proofs of (2.22). One proof is based on the
elementary recurrence relation

o(k) o(n-k), p(o) (2.23)

where o(k) is the sum of the divisors of k. The asymptotic approximation of

o(n) led to this result (2.22).
RamanuJan and Hardy’s second proof was based upon the Cauchy integral formula in
complex analysis. In order to outline the proof, we replace real x by a complex z in
(2.2) to obtain the Taylor series representation of F(z) and hence the coefficient
p(n) of the resulting series can be expressed by the Cauchy integral formula

p(n)
F(n) (o) _/__I [ F_)dz (2.24)
n! 2i
C z

where F(z) is analytic inside the unit disk Izl in the complex z-plane, C is a
suitable closed contour enclosing the origin and lying entirely within the unit
disk. It turns out that the unit clrcle is a natural boundary for F(z).
We choose C as a circle with the origin as center and radius r (0 < r < I) in the
 SRINIVASA RAMANUJAN AND PARTITIONS OF NUMBERS 633

complex z-plane. We make a change of variable z= exp (2iT) in (2.2) so that

27 iT
F(e (I _e2iTm)
m

(iT] [n(T)] -1
exP,12 (2.25)

where q(T) is the Dedekind n- function which is an analytic function of T for

Im (T) > 0.

We now rewrite (2.24) in terms of n(T) in the form

-2i T
e n dT

-
p(n) n(T) Im (T) > 0 (2.26)
r
where F is the of length unity, parallel to the real axis, from
line segment
2
+ ie to + ie, e
n
n
24 >
We also assume C to be such that
0 and
F is its image under the transformation z exp(2wiT).
The function n(T) has a simple pole at T=0. RamanuJan and Hardy proved
rigorously that the main contribution to the integral for p(n) given by (2.26) comes
from the polar singularity at
given by
T--0 as n . Thus the asymptotic value of p(n) is

p(n)
4
2
n
exp (ln K) + 0
(elK:n ), n+
n
(R), (2.27)

where K

When is replaced by n, then (2.27) becomes identical with (2.22).

n

singular at every point

unity.
T (p,q) ,
Finally, H. Rademacher (1937) further improved and fully completed the evaluation
of the integral for p(n) by proving an exact formula.
on a segment of the real axis of length
He then evaluated contributions to integral (2.26) at its all singular points
of the form T
p
He noted that q(T) has also

and obtained the exact formula

q

p(n) ---/--I
2
Z q Aq (n)
q=l
d
Kn]
[n slnh--- n >_ I, (2.28)
where
A (n) exp(-2ip/q), exp[is(p,q)], (p,q) I, (2.29abc)
q P, q P, q
p--q
and s(p,q) is the Dedekind sum.

The work of RamanuJan-Hardy’s partition function combined with that of Rademacher

can be regarded as truely remarkable and have stimulated tremendous interests in
subsequent developments in the theory of modular functions. The RamanuJan-Hardy
collaboration on the asymptotic analysis for p(n) is one of the monumental results in
the history of mathematics and is perhaps best described by J.E. Littlewood (1929) in
his review of the collected papers of Srlnlvasa RamanuJan in the Nature.
634 L. DEBNATH

3. APPLICATIONS TO STATISTICAL MECHANICS

One of the most remarkable applications of the RamanuJan-Hardy asymptotic formula
for p(n) deals with the problems of statistical mechanics. Several authors including
Auluck and Kotharl (1946), Temperley (1949) and Dutta (1956) discussed the significant
role of partition functions in statistical mechanics. The theory of partitions of
numbers have been found to be very useful for the study of the Bose-Einsteln
condensation of a perfect gas. The central problem is the determination of number of
ways a given amount of energy can be shared out among different possible states of a
thermodynamic assembly. This problem is essentially the same as that of finding the
number of partitions of a number into integers under certain restrictions.

We consider a thermodynamic assembly of N non-lnteractlng identical linear simple

harmonic oscillators. The energy levels associated with an oscillator are
e (m+ where m is a non-negatlve integer, h=2 is the Planck constant and
m
is the angular frequency. If E denotes the energy of the assembly, a number n is
defined by

n(0) E N (3.1)

where n denotes (in units of 00) the energy of the assembly, excluding the residual
energy given by the second term of the right side of (3.1).

We denote ?(E) for the number of distinct wave functions assigned to the
assembly for the energy state E. It is well known that for a Bose-Einsteln assembly
the number of assigned wave functions is the number of ways of distributing n energy
quanta among N identical oscillators without any restriction as to the number of
quanta assigned to the oscillator. For a Fermi-Dirac assembly, the energy quanta
assigned to all oscillators are all different. For the case of a classical Maxwell-
Boltzmann assembly, oscillators are considered as distinguishable from each other, and
the number of wave functions is simply the number of ways of distributing n energy
quanta among N distinguishable oscillators. This is equal to the number of ways of
assigning N elements to n positions, repetitions of any element are permissible.

If Pd(n) denotes the number of partitions of n into exactly d or less than d

parts, then Pd(n)--p(n) for d >_ n where p(n) is the number of partitions of n as a
s of positive integers. On the other hand, qd(n) representSdthe number of
partitions of n into exactly d unequal parts so that Pd(n) qk(n). On the
k-I
other

hand, the number of partitions of n into exactly d or less different parts is denoted
by qd(n) and Qd(n) sands for the number of partitions of n into exactly d unequal
parts so that [ Qk(n)
qd(n) k--1 We also observe the following results:

Pd(n) qd(n + d) qd(n) d(n +- d(d-1)), (3.2ab)

Pd(n) Qd(n + d(d+l)), Qd(n+d) Qd(n) + Qd_t(n), (3.3ab)

It turns out that

?(E) PN(n) for the Bose-Einsteln assembly, (3.4)
 SRINIVASA RAMANUJAN AND PARTITIONS OF NUMBERS 635

9(E) QN(n) + QN_I(n) QN(n+N) for the Fermi-Dirac assembly, (3.5)

N
H
n (N+n-l) (3.6)
9(E) for the Maxwell-Boltzmann assembly,
N-T- N!(N-I)!n!

where N! in (3.6) is inserted to make the entropy expression meaningful. It is

important to point out that if N=0(n ), both PN(n) and QN(n+N) tend to N
H
/N!. This
means that for N < < n both the Bose-Einstein statistics and the ermi-Dirac
statistics tend to the classical Maxwell-Boltzmann statistics.

where
-I
Z

is
.
The state function Z for an assembly is defined by

(E

the
e
-E m

temperature measured in energy units. For the Bose-Einstein

(3.7)

where
Z

kT
-,
n=l
.
assembly of the linear oscillators, the state function Z is given by

PN(n) exp[-(n +

and k is the Boltzmann constant.

N) ]

We can rewrite (3.8) as

(3.8)

N -I
ZeN-- (l_er) (3.9)
r--I
For the Fermi-Dirac assembly, we have

ZeN n:Z [QN (n) + QN-I(n) exp(-n)

N
exp [- N(N-I) ,] (l-er) -I (3.1o)
r:l

-
For the classical Maxwell-Boltzmann case, we have
-N
N= NI_.!
Ze
" I-e (3.11)

- -
It is interesting to point out that as N2/ 0
N -I -N N
II (l-e -r (l-e II {l+e +e-(r-l)}-I
r=l
-N
-!(1 _e- (3.12)

This means that the classical statistics is the limit of both results (3.9) and
(3.10).
The above expressions for the state function Z were used to obtain the result for
the energy E (or n) and the entropy S. Using the expression for S, asymptotic
formulas for the partition functions PN(n) and p(n) as follows:

N-I
PN(n NI_ ! N
H
n
exp(2N) n
27 N
2N
for N < < n (3.13)
636 L. DEBNATH

PN (n) 4n 3--I exp [7 22 I n exp(-N/6/n)] for N > > n (3.14)

Thus, in the limit, n

PN(n) p(n) exp (7 (3.15)

This is the RamanuJan-Hardy asymptotic formula.

A more general result can be derived in the form

D x )} t x
D exp[{( (I- --x
e -I
f _-
et-I
log (l-e-t)} dt +
2(eX-1) (3.16)
PN(n) o

2n [2e(l-e -x) (I
x2
}l I/2
x
e -I
where D is given by

x
D fo (et-l)-lt dt (3.17)

and x=lN is given by

x
D N xe + 0( 2)
(eX-l) 2}
n ++ (3.18)
p
2
2 2(eX-l) 4+ 2 e
x -I

It is important to point out that this general result for PN(n) reduces to (3.13) as
xffiN 0, and to (3.14) as xfN
On the other hand, Temperley (1949) applied the RamanuJan-Hardy theory of
partitions to discuss results of the Bose-Einsteln condensation theory. He considered
a problem different from that of Auluck and Kothari. His model consists of N
particles obeying Bose-Einstein statistics distributed among infinitely many energy
levels 0, e, 2e, 3, of uniform spacing E in such a way that the total energy
is cE. The partition function PN(E) representing the number of ways of dividing an

I/(l-x)(l-xz)(1-xz2)(l-xz 3
Temperley used the RamanuJan-Hardy asymptotic method to compute
.
integer E (energy) into N or less integral parts has the generating function

N E Pn (E) xNzE,
PN(E)
(3.19)

for large N.
His analysis gives

pN(Z) i/ Eexp
[7 exp I- J6E N > > (3.20)

This result is similar to that of Auluck and Kotharl (1946) who considered the problem
of distribution of a fixed amount of energy between N harmonic oscillators of equal
frequency, and solved it from the statistical mechanics of harmonic oscillators.

Temperley also investigated the Bose-Einsteln perfect gas model which is also
equivalent to a certain problem in the partition of numbers into sums of squares. The
 SRINIVASA RAMANUJAN AND PARTITIONS OF NUMBERS 637

energy levels available to particles in a cubical box of side-length d are given by

the expresion

where
h
8md
2

K(r,s,t)
2
(p 2+

number of times or not at all.

s2+ t
2

(r2+ s2+ t2).

-- h
2

8md
2

Each
K(r s t)

of these levels may be

The problem is to find the asymptotic form for the
occupied an
(3 21)

integral

number of distinct partition of an integer E, representing the ratio of the total to

the lowest possible energy-separatlon h2/Smd2, into a sum of the numbers where the
order in which the K’s are arranged is neglected, but on the other hand, K’s like
K(1,2,3), K(2,1,3) are distinct from one another, and such K’s have to be treated as
different, even though they are numerically equal. In all cases, the quantitics r,s,t
are positive integers. The upshot of this analysis is the existence of an
intermediate temperature region within which the results of earlier theory are
unreliable. It is also confirmed the existence of the phenomena of condensation into
lowest energy-levels. At the same time, the present investigation gives a transition
departure far below K for a perfect gas of heillum atoms. However, the earlier
theory can provide physically sensible results at very high and at very low
temperatures.

All these above discussions show a clear evidence for the great importance as
well as success of the RamanuJan-Hardy theory of partitions in statistical mechanics.

In an essentially statistical approach to thermodynamic problems, Durra (1953,

1956) obtained some general results from which different statistics viz., those of
Bose, Fermi and Gentile, Maxwell-Boltzmann can be derived by using different
partitions of numbers. It is noted that mathematical problems of statisitics of Bose,
Fermi, and Gentile are those of partitions of numbers (energy) into partitions in
which repetition of parts are restricted differently. In partitions corresponding to
Bose statistics any part can be repeated any number of times, that to Gentile
statistics any part can be repeated upto d times where d is a fixed positive integers,
and that to Fermi statistics no part is allowed to repeat, that is, d=l. All these
led to an investigation of a new and different type of partitions of numbers in which
repetition of any part is restricted suitably. Motivated by the need of such
partition functions and its physical applications to statistical physics, Dutta (1956,
1957) Durra and Debnath (1959) studied a new partition of number n into any number of
parts, in which no part is repeated more than d times. Dutta’s partition function is
denoted by dP(n). Dutta himself and in collaboration with De bnath proved algebraic
and congruence properties of dP(n).
The generating function of this partition
function is

f(x) --I dP(n) xn* (I+ xn+ x 2n + + xdn)

n--I
l-x n"
d+
E (l-x n(d+l) g p(n) xn
[[ (3.22)
l-x
n
E (l-x n) r. p(n) x n(d+l)

where p(n) is the unrestricted partition function due to Hardy and RamanuJan (1918).
638 L. DEBNATH

He proved the following congruence properties for dP(n):

4p (5m+ 4) E 0 (mod 5) (3.23)

6p (7m + 5) --0 (rood 7) (3.24)

(llm + 6) 0 (rood II)

120P (3.25)

Subsequently, Dutta and Debnath (1957) introduced a new partition function

dP(n|m) representing the number of partition of an integer n into m parts with utmost
d repetitions. They proved the generating function, congruence properties, recurrence
relations and other related properties with examples. Several special cases of this
partition functions are discussed with examples. Dutta obtained a simple algebraic
p(n) of
formula to calculate successively the numerical values of d from the values
p(n) and so ultimately from Euler’s table. Using a Tauberlan Theorem, Dutta proved an
pn)
asymptotic forumla correct up to the exponential order above for d for large n:

z d
dP(n) exp [{’" n(d----) (3.26)

In particular, for partition of unequal parts (d=l), (3.26) becomes

(3.27)

For unrestricted partitions (d ) (3.26) reduces to

dP(n) exp [/ for n + (3.28)

These results are in excellent agreement with those of Hardy and RamanuJan (1918) upto
the expotentlal order. The asymptotic result for the unrestricted partition is found
to be very useful for computing the dominant term in the expression for the entropy of
the corresponding thermodynamic system.

Dutta’s partition function is not only more general than that introduced by
earlier authors, but also it is more useful for the study of problems in statistical
physics. Mathematical problems of Gentile statistics deals with the partitions of numbers
(energy) into parts in which repetition of parts are restricted differently. In
partitions corresponding to Bose statistics, any part can be repeated any number of
times (d / (R)). The Fermi statistics deals with the partitions of number into parts in
which no part can repeat (d=l). In partitions corresponding to Gentile statistics,
any part can be repeated up to d times. In other words, Dutta’s partition function
p(n)
d is found to be useful for an investigation of thermodynamic problems.

As a final example of physical application of RamanuJan-Hardy’s theory of

partitions of numbers, mention may be made of a paper by Bohn and Kalckar (1937)
dealing with calculation of the density of energy levels for a heavy nucleus.
 SRINIVASA RAMANUJAN AND PARTITIONS OF NUMBERS 639

4. CONCLUDING REMARKS

It is hoped that enough has been discussed to give some definite impression of
Ramanujan’s great character as well as of the range and depth of his contributions to
pure mathematics. Throughout his life, Ramanujan was deeply committed to his family
and friends. He also expressed an unlimited interest in education and deep compassion
for poor students and orphans who needed support for their education. He also
profoundly believed in the dignity and work of human being. RamanuJan’s entire life
was totally dedicated to the pursuit of mathematical truth and dissemination of new
mathematical knowledge. His genius was recognized quite early in his llfe and has
never been in question. Indeed, Hardy in his "A Mathematician’s Apology" wrote: "I
have found it easy to work with others, and have collaborated on a large scale with
two exceptional mathematicians (Ramanujan and Littlewood) and this has enabled me to
add to mathematics a good deal more than I could reasonably have expected." Also, he
said: "All of my best work since then (1911 and 1913) has been bound up with theirs
(Ramanujan and Littlewood), ...". There is no doubt at all about RamanuJan’s profound
and ever-lasting impact on mathematics and mathematical community of the world.
Today, one hundred years after his birth, we pay tribute to this great man, and at the
same time, we can assess and marvel at the magnitude of his outstanding
achievements. By any appraisal, RamanuJan was indeed a noble man and a great
mathematician of all time.

5. ACKNOWLEDGEMENTS

The biographical sketch of Srinivasa RamanuJan described in this article is based

upon G.H. Hardy’s two books entitled Collected Papers of Srinivasa Ramanuan (1927)
and Ramanuan (1940). This work was partially supported by the University of Central
Florida.

REFERENCES

I. ANDREWS, G.E., The Theory of Partitions, Vol. 2, Addison-Wesley, London, (1976).

2. ATKINS, A.O.L., Proof a Conjecture of RamanuJan, Glasgow. Math. J. 8 (1967),
14-32.
3. AULUCK, F.C. and KOTHARI, D.S., Statistical Mechanics and the Partitions of
Numbers, Proc. Camb. Phil. Soc. 42 (1946), 272-277.
4. BOHR, N., and KALCKAR, F., Kgl Danske Vid. Selskab Math. Phys. Medd 14 (1937)
and also in Van Lier, C., and Uhlenbeck, G.E., Physica 4 (1937) 531.
5. DEBNATH, L., Some Recent Results of Number Theory Discovered with Electronic
Computers, Internat. J. Math. Edu. Sci.Technol. 13 (1982) 603-617.
6. DUTTA, M., An Essential Statistical Approach to Thermodynamic Problem II, Proc.
Nat. Inst. Scl., India A 19 (1953) 109-126.
7. DUTTA, M., On New Partition of Numbers, Rend Seminario Math della Univ. di
Padova XXV (1956) 138-143.
8. DUTTA, M., On New Partitions of Numbers, Bull.Cal. Math. Soc. 49 (1957) 221-224.
9. DUTTA, M. and DEBNATH, L., On New Partitions of Numbers II, Bull. Cal. Math Soc.
42 (1959) 77-79.
640 L. DEBNATH

I0. DUTTA, M. and DEBNATH, L., Elements of the .Th.eorjof Elliptlc and Associated
Functions with Applications, World Press, Calcutta (1965).
II. GUPTA, H., Selected Topics in Number Theory, Aabacus Press, England, 1980.
12. HARDY, G.H., A Mathematician’s. Apology, Cambridge University Press, Cambridge,
1927.
13. HARDY, G.H., Collected Papers of Srlnlvasa Ramanuan, Cambridge University
Press, Cambridge, 1927.
14. HARDY, G.H., Ramanujan, Chelsea Publishing Company, New York, 1940.
15. HARDY, G.H. and RAMANUJAN, S., Asymptotic Formulae in Combinatory Analysis,
Proc. Lond. Math. Soc. 17 (1918) 75-115.
16. LEHMER, D.H., On a Conjecture of RamanuJan, Jour. Lond. Math. Soc. II (1936),
114-118.
17. LITTLEWOOD, J.E., Srlnlvasa RamanuJan, Nature 123 (1929) 631-633.
18. RADEMACHER, H., On the Partition Function p(n), Proc. Land. Math. Soc. 43
(1937) 241-254.
19. RANGANATHAN, S.R., Ramanuan, .The Man...a.nd the Mathematician, Asia Publishing
Company, Calcutta, 1967.
20. 0TEMPERLEY, H.N.V., Statistical Mechanics and the Partition of Numbers, I. The
Transition of Liquid Helium,
Pro.c.. Lond..R.o. Soc__.__. A199 (1949) 361-375.
21. WATSON, G.N., RamanuJans Vermtung uber Zerfallungsazahlen, J. Reine Angew
Math 179 (1938) 97-128.
22. WATSON, G.N., The Final Problem: An Account of the Mock Theta Functions, J.
Lond. Math. Soc. II (1936) 55-80.