Number Theory (Short Work)
Number Theory (Short Work)
Number Theory (Short Work)
Suprokash Hazra
Department of Mathematics,
Ramanujan school of Mathematical Sciences,
Pondicherry University, Puducherry, India
e-mail: [email protected]
Abstract: In this paper we go on to discuss about two different versions for the proof of the
generalization of Stanley’s theorem in integer partition. We illustrate two different techniques
that may be applied to profitably understand the underlying structure behind the theorem.
Keywords: Stanley’s theorem, Tilings, Partition identities.
AMS Classification: 2010, 05A17.
1 Introduction :
Stanley’s theorem is an important result in the theory of partitions. Various generalizations have
been made to it over the course of time inc and Dastidar & Gupta’s subsequent generalization. In
their paper they go on to show how the sum of the number of distinct members of the partition is
not just equal to the number of 1’s present in the partition of the same number but is also equal
to the sum of the number of i’s in the partitions of all the numbers from n to (n + i − 1), where i
can be any positive integer. In this paper we go on to provide different combinatorial arguments
to prove such generalizations. A proof of the generalization involves the use of tilings of a 1x∞
board. This concept can be further generalised to cover similar properties for overpartitions as
well.To prove the generalization we first make use of some lemmas which follow subsequently in
the article.
2 Main Results :
A. Combinatorial proof of the extension of Stanley’s theorem :
We need the following lemma in order to prove our main theorem.
Lemma 2.A.1 : Let n and k be two positive integers with k ≤ n, then for each positive integer
i, (1 ≤ i ≤ n), we have n + i = qi k + ri , where 0 ≤ ri < k, then,
qi = q0 , for all 1 ≤ i ≤ s − 1
= q0 + 1, for all i ≥ s
1
And ri = r0 + i , for all 1 ≤i ≤ s - 1,
= i − s , for all i ≥ s where s = k − r0
Proof : Let us consider the two integers n + i and k,then by division algorithm there exists two
integers qi and ri such that, n + i = qi k + ri with 0 ≤ ri < k. Now, n = q0 k + r0 . We assume,
k − r0 = s, then we have the following,
n = q0 k + r0 ; n + 1 = q0 k + (r0 + 1);
...
n + s − 1 = q0 k + (r0 + s − 1), n + s = (q0 + 1)k ;
n + s + 1 = (q0 + 1)k + 1, ...n + k − 1 = (q0 + 1)k + (r0 − 1)
By the above formula it is clear that
qi = q0 for all 1 ≤ i ≤ s − 1
=q0 + 1 for all i ≥ s
Hence the lemma 2.A.1
of (n + i) for all i.
Pqi
Lemma 2.A.3: Qk (n + i) = j=1 P (n + i − jk), for some fixed i.
Proof : For fixed i, considering the two integers n + i and k, then by division algorithm there
exists two integers qi and ri such that,
For a fixed i, the number of partitions of n + i,where k occurs exactly t-times is,
2
=P (n + i − k) + P (n + i − 2k) + P (n + i − 3k) + ...... + P (n + i − (qi − 1)k)
−(qi − 1)P (n + i − qi k) + qi P (ri )
=P (n + i − k) + P (n + i − 2k) + P (n + i − 3k) + ....... + P (n + i − (qi − 1)k)
− qi P (n + i − qi k) + P (n + i − qi k) + qi P (ri )
=P (n + i − k) + P (n + i − 2k) + P (n + i − 3k) + ....... + P (n + i − (qi − 1)k))+
P (n + i − qi k)[since, n + i − qi k = ri ]
= qj=1
Pi
P (n + i − jk),
Hence the lemma 2.A.3.
Remark 2.A.4 : When t ∈ 1, 2, ..., n is fixed, then for all 1 ≤ i ≤n ,we have,
Pk−1
i=0 [P (n + i − tk)]
3
appears and sum those result for all 1 ≤ i ≤ n. Now the number of partitions of n in which i
appears is P (n − i) for all 1 ≤ i ≤ n − 1 and with the special case i = nis 1.
Hence B(n) = 1 + P (1) + P (2) + ... + P (n − 3) + P (n − 2) + P (n − 1).
Hence A(n) = B(n)
This completes our proof.
Proof : Consider A = Set of all tilings with atleast (k − i) black tiles in position r.
And, B = Set of all tilings with atleast r black tiles in position k.
Now we define T : A→ B, by the following way that, we take a tiling from A, then we remove
(k − i) black tiles from position r and add r black tiles in position k.
Then T is well defined map. In order to prove T is bijective, we define a mapping S:B→A,
by the following way,we take a tiling from B,then we remove r black tiles from position k and
add (k − i) black tiles inposition r. Then S is well defined map. Also ST =IdA and T S = IdB ,
the identity mappings on A and B respectively.
Now clearly the change of measure under the map T is q −r(k−i)+kr = q ir .
Now take A∗= Set of all tilings of A with measure q n . Then for each tiling of A∗ we obtain a
bijective Correspondence with a tiling of B of measure q n+ir .Also every tiling gives a partitions
and vice versa.So by the Partition representation of a tiling, the lemma follows.
4
Hence the lemma 2.B.2.
Lemma 2.B.3: Number of partitions of n with atleast i times r = The number of partitions of
n with atleast r times i.
Proof : Consider M = Set of all tilings with atleast i black tiles in position r. And N = Set of all
tilings with atleast r black tiles in position i. Now we define Q: M→N, by the following way, We
take a tiling from M, then we remove i black tiles from position r and add r black tiles in position
i.
Then Q is well defined map. In order to prove Q is bijective we define a mapping Z:N→M,
bythe following way,we take a tiling from N, then we remove r black tiles from position i and add
i black tiles in position r. Then Z is well defined map. Also QZ= IdN and ZQ = IdM , the identity
mappings on N and M respectively. Now clearly there is no change of measure under the map
Q. Now take M*=Set of all tilings of A with measure q n . Then for each tiling of A* we obtain
a bijective Correspondence with a tiling of N of measure q n . Now every tiling gives a partitions
and vice versa. So by the Partition representation of a tiling, we have the lemma 2.B.3.
Lemma 2.B.4 : The number of partitions of n with atleast 1 times r = The number of partitions
of (n + kj − r) with atleast j times k. we assume n = qk + r,and 1 ≤ j ≤ q.
In particular,Vjk (n + kj − r) = Vsk (n + ks − r),for all 1 ≤ j, s ≤ n, where Vpq (m)= The
number of partitions of m with atleast p times q.
Proof : Let C = Set of all tilings with atleast 1 black tiles in position r. And D= Set of all tilings
with atleast j black tiles in position k. Now we define F: C → D, by the following way, We take
a tiling from C, then we remove 1 black tiles from position r and add j black tiles in position k.
Then F is well defined map. In order to prove F is bijective we define a mapping G:D→ C,by
the following way, we take a tiling from D, then we remove j black tiles from position k and add
1 black tiles in position r. Then G is well defined map. Also GF = IdC and FG = IdD , the identity
mappings on C and D respectively.
Now clearly the change of measure under the map F is q kj−r .
Denote, C # =Set of all tilings of C with measure q n . Then for each tiling of C # we obtain a
bijective Correspondence with a tiling of D of measure q n+kj−r .
Now every tiling gives a partition and vice versa. So by the Partition representation of a
tiling, we have that, the number of partitions of n with atleast 1 times r = number of partitions of
(n + kj − r) with atleast j times k.
So notationally, Vjk (n + kj − r) = V1r (n).
Since the right hand side of the above expression does not depend on j,so clearly we have
5
that, Vjk (n + kj − r) = Vsk (n + ks − r), for all 1 ≤ j, s ≤ n
Hence the lemma 2.B.4.
We now give the tiling proof of the extension of Stanley’s theorem.
Next we put r = 1 in lemma 2.B.3 and obtain, Vi1 (n) = V1i (n)
We sum this over all k ≤ i ≤ n and obtain,
Pn Pn
i=k Vi1 (n) = i=k V1i (n) .......(2)
= k−1
P k
Pn i
Pn 1
Pk−1 k Pn i
i=1 V1 (n + i) + i=k V1 (n) i=1 Vi (n)= i=1 V1 (n + i) + i=k V1 (n)
= k−1
P k
Pn i
i=1 V1 (n + i) + i=k V1 (n)
= k−1
P k k
Pn i
Pk−1 k Pn i
i=1 V1 (n + i) + V1 (n) + i=k+1 V1 (n) = i=0 V1 (n + i) + i=k+1 V1 (n) ......(3)
= 2k
P k
P3k k
Pqk k
i=k+1 Vj (n + kjr) + i=2k+1 Vj (n + kj − r) + .....+ i=(q−1)k+1 (Vj (n + kj − r))
= 2k
P k
P3k k
Pqk k
i=k+1 V2 (n + 2kr) + i=2k+1 V3 (n + 3k − r) + ......... + i=(q−1)k+1 (Vq (n + kq − r))
Hence we obtain,
Pn r
Pk−1 k Pk−1 k Pk−1 k
i=k+1 V1 (n) = i=0 V2 (n + i) + i=0 V3 (n + i) + ......... + i=0 Vq (n + i) .......(4)
= k−1
P k
Pk−1 k Pk−1 k Pk−1 k
i=0 V1 (n + i) + i=0 V2 (n + i) + i=0 V3 (n + i) + ...... + i=0 Vq (n + i)
6
Pq Pq Pq
= j=1 Vjk (n) + j=1 Vjk (n + 1) + .... + j=1 Vjk (n + k − 1)
= Qk (n) + Qk (n + 1) + Qk (n + 2) + .... + Qk (n + k − 1)
= k−1
P
i=0 Qk (n + i)
3 Conclusion :
In this paper we have demonstrated how simple combinatorial proofs can be effectively applied
to problems involving unordered partitions. Similar to the extension of Stanley’s theorem,we can
apply similar techniques to prove further extensions of Elder’s theorem and not just for regular
partitions but for all classes of overpartitions. The concept of tilings could also pe profitably
extended to study similar results in the area of planar partition.
References
[1] Wolfram Mathworld. Elder’s Theorem. 2010. Available at http://mathworld. wol-
fram.com/EldersTheorem.html
[3] Andrews, George, The Theory of Partitions, in “Encyclopedia of Mathematics and Its Appli-
cations,” Vol. 2, Addison-Wesley,Reading, MA, 1976.
[4] Robson da Silva, Jorge F. A. Lima, Jose Plnio O. Santos and Eduardo C. Stabel, Generating
function and combinatorial proofs of Elder’s theorem, Notes on Number Theory and Discrete
Mathematics, ISSN 1310–5132, Vol. 21, 2015, No. 4, 30–35.