Journal of Number Theory 182 (2018) 8394

Contents lists available at ScienceDirect

Journal of Number Theory

www.elsevier.com/locate/jnt

New congruences for t-core partitions and

Andrews singular overpartitions
T. Kathiravan 1 , S.N. Fathima
Department of Mathematics, Ramanujan School of Mathematics, Pondicherry
University, Puducherry, 605 014, India

a r t i c l e i n f o a b s t r a c t

Article history: A partition of n is called a t-core partition of n if none of

Received 4 April 2017 its hook numbers are multiples of t. Let the number of t-core
Received in revised form 9 May 2017 partitions of n be denoted by at (n). Recently, G. E. Andrews
Accepted 15 May 2017
dened combinatorial objects which he called (k, i) singular
Available online 27 July 2017
Communicated by the Principal overpartitions, overpartitions of n in which no part is divisible
Editors by k and only parts i (mod k) may be overlined. Let the
number of (k, i) singular overpartitions of n be denoted by
MSC: C k,i (n). The object of this paper is to obtain new congruences
05A17 modulo 2 for a15 (n) and a23 (n). We also obtain congruences
11P83 modulo 2 for C 92,23 and C 60,15 .
2017 Elsevier Inc. All rights reserved.
Keywords:
t-Core
Singular overpartition
Congruence
Dissection

* Corresponding author.
E-mail addresses: kkathiravan98@gmail.com (T. Kathiravan), fathima.mat@pondiuni.edu.in
(S.N. Fathima).
1
The rst author research is supported by UGC-BSR, Research Fellowship, New Delhi, Government of
India.

http://dx.doi.org/10.1016/j.jnt.2017.05.021
0022-314X/ 2017 Elsevier Inc. All rights reserved.
84 T. Kathiravan, S.N. Fathima / Journal of Number Theory 182 (2018) 8394

1. Introduction

A partition of the positive integer n is a nonincreasing sequence of positive integers

whose sum is n. The number of partitions of n is denoted by p(n). A partition is very
ofter represented with the help of its FerrersYoung diagram. If = 1 2 s is
a partition of n, then the FerrersYoung diagram of A is the s-row collection of nodes

... 1 nodes
... 2 nodes
.. .. .. . . ..
. . . . .
... s nodes

Let j denote the number of nodes in column j. Then the hook number H(i, j) of the
(i, j) node is dened by H(i, j) := i + j j i + 1. For t > 0, the partition of n is
called a t-core of n if none of the hook numbers of its associated FerrersYoung diagram
is a multiple of t.

Example. The FerrersYoung diagram of the partition = (4, 2, 1) of 7 is

The nodes (1, 1), (1, 2), (1, 3), (1, 4), (2, 1), (2, 2) and (3, 1) have hook numbers 6, 4, 2,
1, 3, 1 and 1, respectively. Therefore, is a 5-core. Obviously, it is a t-core for t 7.
If at (n) denotes the number of partitions of n that are t-cores, then the generating
function for at (n) is given by

 (q t ; q t )t
at (n)q n = . (1.1)
n=0
(q; q)

The study of t-cores for t prime rst arose in connection with Nakayamas conjecture [14,
17]. Garvan, Kim and Stanton [11,12] have found that t-cores are useful in establishing
cranks, which help to give combinatorial proofs of Ramanujans congruences [18] [19, pp.
210213], satised by p(n), namely,

p(5n + 4) 0 (mod 5),

p(7n + 5) 0 (mod 7),
p(11n + 6) 0 (mod 11).

In [10], Garvan proved some Ramanujan type congruences for ap (n) for certain small
primes p, and Hirschhorn and Sellers [13] proved multiplicative formulas for a4 (n). For
example, for 0,
 T. Kathiravan, S.N. Fathima / Journal of Number Theory 182 (2018) 8394 85

 
2+1 5 32 5
a4 3 n+ = 2 a4 (3n).
8

The topic of t-core partitions has been the subject of a number of papers [6,5,4,7,20].
Recently G. E. Andrews [3] introduced (k, i) singular overpartitions, overpartitions in
which no part is divisible by k and only parts i (mod k) may be overlined. Let
C k,i (n) denote the number of such partitions of n. The generating function for C k,i (n),
where k 3 and 1 i  k2  is

 (q k ; q k ) (q i ; q k ) (q ki ; q k )
C k,i (n)q n = . (1.2)
n=0
(q; q)

In his paper [3], G. E. Andrews also proved that for n 0,

C 3,1 (9n + 3) C 3,1 (9n + 6) 0 (mod 3).

Chan et al. [8] generalized and found innite families of congruences modulo 3 for C 3,1 (n),
C 6,1 (n), C 6,2 and modulo 2 for C 4,1 (n). For recent works on singular overpartitions, see
[1,16,15].

The aim of this paper is to prove new congruences for a15 (n), a23 (n), C 92,23 (n) and
C 60,15 (n). The following are our main results:

Theorem 1.1. If  {5, 7, 10, 11, 14, 15, 17, 19, 20, 21, 22} then for all n 0,

a23 (8(23n + ) + 1) 0 (mod 2) (1.3)

and if m {1, 7, 9, 12, 13, 16, 17, 19, 21, 22} then for all n 0,
 
a23 8 232k+2 (23n + m) + 7 232k+2 22 0 (mod 2). (1.4)

Theorem 1.2. For all n 0,

a15 (8n + 2) 0 (mod 2) (1.5)

a15 (8(5n + s) + 6) 0 (mod 2), s {3, 4} (1.6)
 
70 52k 28
a15 85 2k+1
(5n + s) + 0 (mod 2), s {1, 2}. (1.7)
3

Theorem 1.3. If  {5, 7, 10, 11, 14, 15, 17, 19, 20, 21, 22} then for all n 0,
 
232k+1 17
C 92,23 4 23 (23n + ) +
2k
0 (mod 2). (1.8)
6
86 T. Kathiravan, S.N. Fathima / Journal of Number Theory 182 (2018) 8394

Theorem 1.4. For all n 0,

 
19 52k+1 11
C 60,15 20 52k (5n + s) + q n 0 (mod 2), s {1, 2}. (1.9)
6

In order to prove our main results, we collect a few lemmas in section 2. The proofs
of Theorems 1.11.4 are given in section 3.

2. Preliminaries

By the binomial theorem, for any positive integer k,

k k1
f 2 f22 (mod 2k ). (2.1)

Lemma 2.1. (Cui and Gu [9, Theorem 2.2]) If p 5 is a prime and

p 1
p 1 , if p 1 (mod 6),
:= p6 1
6 , if p 1 (mod 6),
6

then
p1

2  
3k2 +k 3p2 +(6k+1)p 3p2 (6k+1)p
(q; q) = (1)k q 2 f q 2 , q 2

k= p12
k= p1
6

p1 p2 1 2 2
+ (1) 6 q 24 (q p ; q p ) . (2.2)

p1 3k2 +k p2 1
Furthermore, if p1
2 k
p1
2 ,k = 6 , then 2  24 (mod p).

Lemma 2.2. (Ahmed and Baruah [2, Eqn. (3.5)])

1 1 
15 15
= 2 2 2 30 30 2 (q 6 )(q 10 ) + qf (q 90 , q 150 )f (q 2 , q 14 )
(q; q) (q ; q ) (q ; q ) (q ; q )

+q 15 f (q 30 , q 210 )f (q 6 , q 10 ) . (2.3)

3. Proofs of Theorems 1.11.4

Theorem 3.1. For any non-negative integer k, we have

  
a23 8 232k+1 n + 232k+1 22 q n f1 f2 + q 2 f1 f4 f46 (mod 2) (3.1)
n=0
 T. Kathiravan, S.N. Fathima / Journal of Number Theory 182 (2018) 8394 87

Proof. From (1.1), we have

 3
f184
a23 (n)q n (mod 2). (3.2)
n=0
f1 f23

Now from [21, Lemma 2.1.], we have


1
p11 231 (2n)q 2n + q + q 3 f2 f46 (mod 2) (3.3)
f1 f23 n=0

where p11 231 (n) is dened by

 1
p11 231 (n)q n = .
n=0
f1 f23

From (3.2) and (3.3), we obtain


 
a23 (n)q
n 3
f184 p11 231 (2n)q 2n 3
+ q + q f2 f46 (mod 2). (3.4)
n=0 n=0

Extracting the terms involving q 2n+1 from both sides of (3.4), dividing both sides by q
and then replacing q 2 by q, we obtain

 7
f2 f46
a23 (2n + 1)q n f92
3
+q (mod 2). (3.5)
n=0
f1 f23

Now, substituting (3.3) in (3.5) and extracting the terms involving q 2n from both sides
of the resulting congruence and then replacing q 2 by q, we obtain

 f2 f184
a23 (4n + 1)q n f46
3
+q + q 2 f2 f184 (mod 2). (3.6)
n=0
f1 f23

Again substituting (3.3) in (3.6) and extracting the terms involving q 2n from both sides
of the resulting congruence and then replacing q 2 by q, we obtain


a23 (8n + 1)q n f23
3
+ q 2 f2 f23
5
(mod 2). (3.7)
n=0

Taking p = 23 in (2.2), and q replacing by q 2 , we get


11  
f2 = f q 46(35+3k) , q 46(353k) + q 44 f1058
2

(1)k q 3k +k
(3.8)
k=11
k=4
88 T. Kathiravan, S.N. Fathima / Journal of Number Theory 182 (2018) 8394

Note that for 11 k 11 and k = 4,

3k 2 + k  44 (mod 23).

Employing (3.8) in (3.7) and extracting the terms involving q 23n from both sides of the
resulting congruence and then replacing q 23 by q, we obtain


a23 (184n + 1)q n f1 f2 + q 2 f1 f4 f46 (mod 2), (3.9)
n=0

which is the k = 0 case of (3.1). Now suppose (3.1) holds for some k 0. Again taking
p = 23 in (2.2), we obtain


11 2  
f1 =
(1)k 3k 2+k
q f q 23(35+3k)
, q 23(353k)
+ q 22 f529
(3.10)
k=11
k=4

Note that for 11 k 11 and k = 4,

3k 2 + k
 22 (mod 23).
2

If we replace q by q 4 in (3.10), we get


11  
f4 = (1)k q 2(3k +k) f q 92(35+3k) , q 92(353k) + q 88 f2116
2

(3.11)
k=11
k=4

Note that for 11 k 11 and k = 4,

2 (3k 2 + k)  88 (mod 23).

Employing (3.8), (3.10) and (3.11) in (3.9) and extracting the terms involving q 23n+20
from both sides of the resulting congruence, dividing both sides by q 20 and then replacing
q 23 by q, we obtain

  
a23 8 232k+2 n + 7 232k+2 22 q n q 2 f23
3
+ q 4 f2 f23
5
(mod 2). (3.12)
n=0

Employing (3.8) in (3.12) and extracting the terms involving q 23n+2 from both sides
of the resulting congruence, dividing both sides by q 2 and then replacing q 23 by q, we
obtain
 T. Kathiravan, S.N. Fathima / Journal of Number Theory 182 (2018) 8394 89

  
a23 8 232k+3 n + 232k+3 22 q n f1 f2 + q 2 f1 f4 f46 (mod 2). (3.13)
n=0

This completes the proof by induction of (1.6). 2

We now prove Theorem 1.1.

Proof of Theorem 1.1. Employing (3.8) in (3.7) and then equating the coecients of
q 23n+ from both sides we obtain (1.3). And also employing (3.8) in (3.12) and then
equating the coecients of q 23n+m from both sides we obtain (1.4). 2

Theorem 3.2. For any non-negative integer k, we have

  
70 52k 28
a15 8 5 2k+1
n+ q n f5 f33 (mod 2) (3.14)
n=0
3

Proof. Again from (1.1), we have

 f240
a15 (n)q n (mod 2). (3.15)
n=0
f1 f15

Substituting (2.3) in (3.15) and extracting the terms involving q 2n from both sides of
the resulting congruence and then replacing q 2 by q, we obtain

 f120 f12 f20
a15 (2n)q n (mod 2). (3.16)
n=0
f2 f30 f3 f5

Now, from [6, Eq. (4.11)], we have

 f22 f30
2
p31 51 (2n + 1)q n = q 2 2 , (3.17)
n=0
f3 f5 f1 f15

where p31 51 (n) is dened by

 1
p31 51 (n)q n = .
n=0
f3 f5

Extracting the terms involving q 2n+1 from both sides of the congruence, dividing both
sides by q, replacing q 2 by q and then employing (3.17), we obtain


a15 (4n + 2)q n qf2 f30
3
(mod 2). (3.18)
n=0

From (3.18) we have

90 T. Kathiravan, S.N. Fathima / Journal of Number Theory 182 (2018) 8394


a15 (8n + 6)q n f1 f15
3
(mod 2). (3.19)
n=0

Ramanujan [19] stated the following identity without proof:

 
f1 = f25 R1 q q 2 R , (3.20)

where

(q 5 , q 20 ; q 25 )
R= .
(q 10 , q 15 ; q 25 )

Substituting (3.20) in (3.19) and extracting the terms involving q 5n+1 from both sides of
the resulting congruence, dividing both sides by q and then replacing q 5 by q, we obtain


a15 (40n + 14)q n f5 f33 (mod 2), (3.21)
n=0

which is the k = 0 case of (3.14). Now suppose (3.14) holds for some k 0. Next, take
power three on both side in (3.20) and replacing q by q 3 , we obtain
 3 
f33 = f75
3
S 3q 3 S 2 + 5q 9 3q 15 S 2 q 18 S 3 (3.22)

where

(q 15 , q 60 ; q 75 )
S= .
(q 30 , q 45 ; q 75 )

Substituting (3.22) in (3.21) and extracting the terms involving q 5n+4 from both sides
of the resulting congruence, dividing both sides by q 4 and then replacing q 5 by q, we
obtain

  
110 52k+1 28
a15 8 52k+2
n+ q n qf1 f15
3
(mod 2). (3.23)
n=0
3

Substituting (3.20) in (3.23) and extracting the terms involving q 5n+2 from both sides
of the resulting congruence, dividing both sides by q 2 and then replacing q 5 by q, we
obtain

  
70 52k+2 28
a15 8 52k+3 n + q n f5 f33 (mod 2). (3.24)
n=0
3

This completes the proof by induction of (3.14). 2

 T. Kathiravan, S.N. Fathima / Journal of Number Theory 182 (2018) 8394 91

We now prove Theorem 1.2.

Proof of Theorem 1.2. The result (1.5) follows from (3.18). Employing (3.20) in (3.19)
and then equating the coecients of q 5n+s from both sides we obtain (1.6). Employing
(3.22) in (3.14), we obtain (1.7). 2

Theorem 3.3. For any non-negative integer k, we have

  
232k+1 17
C 92,23 4 232k n + q n f23 + qf1 f46 + q 2 f2 f23
3
(mod 2). (3.25)
n=0
6

Proof. From (1.2) we have

 2
f46
C 92,23 (n)q n (mod 2). (3.26)
n=0
f1 f23

From (3.3) and (3.26), we have


 
C 92,23 (n)q
n 2
f46 p11 231 (2n)q 2n 3
+ q + q f2 f46 (mod 2). (3.27)
n=0 n=0

Extracting the terms involving q 2n+1 from both sides of (3.27), dividing both sides by q
and then replacing q 2 by q, we obtain

 2
f2 f46
C 92,23 (2n + 1)q n f46 + q (mod 2). (3.28)
n=0
f1 f23

Now substituting (3.3) in (3.28) and extracting the terms involving q 2n from both sides
of the resulting congruence and then replacing q 2 by q, we obtain


C 92,23 (4n + 1) q n f23 + qf1 f46 + q 2 f2 f23
3
(mod 2), (3.29)
n=0

which is the k = 0 case of (3.25). Now suppose (3.25) holds for some k 0. Employing
(3.10) and (3.8) in (3.29) and extracting the terms involving q 23n from both sides of the
resulting congruence and then replacing q 23 by q, we obtain
  
232k+1 17
C 92,23 4 23 2k+1
n+ q n f1 + qf2 f23 + q 2 f1 f2 f46 (mod 2).
n=0
6
(3.30)

Employing (3.10) and (3.8) in (3.30) and extracting the terms involving q 23n from both
sides of the resulting congruence, dividing both sides by q 22 and then replacing q 23 by q,
we obtain
92 T. Kathiravan, S.N. Fathima / Journal of Number Theory 182 (2018) 8394

  
232k+3 17
C 92,23 4 23 2k+2
n+ q n f23 + qf1 f46 + q 2 f2 f23
3
(mod 2). (3.31)
n=0
6

This completes the proof by induction of (3.25). 2

We now prove Theorem 1.3.

Proof of Theorem 1.3. Employing (3.10), (3.8) in (3.25) and then equating the coe-
cients of q 23n+ from both sides we obtain Theorem 1.3. 2

Theorem 3.4. For any non-negative integer k, we have

  
19 52k+1 11
C 60,15 20 5 n +2k
q n f33 f10 (mod 2). (3.32)
n=0
6

Proof. From (1.2) we have

 2
f30
C 60,15 (n)q n (mod 2). (3.33)
n=0
f1 f15

Substituting (2.3) in (3.33) and extracting the terms involving q 2n from both sides of
the resulting congruence and then replacing q 2 by q, we obtain

 f62 f10
2
C 60,15 (2n)q n (mod 2). (3.34)
n=0
f2 f3 f5

Extracting the terms involving q 2n+1 from both sides of the congruence, dividing both
sides by q, replacing q 2 by q and then employing (3.17), we obtain


C 60,15 (4n + 2)q n qf2 f15
3
(mod 2). (3.35)
n=0

Now replacing q by q 2 in (3.20), we get

 
f2 = f50 M 1 q 2 q 4 M , (3.36)

where

(q 10 , q 40 ; q 50 )
M= .
(q 20 , q 30 ; q 50 )

Substituting (3.36) in (3.35) and extracting the terms involving q 5n+3 from both sides
of the resulting congruence, dividing both sides by q 3 and then replacing q 5 by q, we
obtain
 T. Kathiravan, S.N. Fathima / Journal of Number Theory 182 (2018) 8394 93


C 60,15 (20n + 14)q n f33 f10 (mod 2), (3.37)
n=0

which is the k = 0 case of (3.32). Now suppose (3.32) holds for some k 0. Substituting
(3.22) in (3.37) and extracting the terms involving q 5n+4 from both sides of the resulting
congruence, dividing both sides by q 4 and then replacing q 5 by q, we obtain

  
23 52k+2 11
C 60,15 20 52k+1 n + q n qf2 f15
3
(mod 2). (3.38)
n=0
6

Substituting (3.36) in (3.38) and extracting the terms involving q 5n+3 from both sides
of the resulting congruence, dividing both sides by q 3 and then replacing q 5 by q, we
obtain

  
19 52k+3 11
C 60,15 20 5 2k+2
n+ q n f33 f10 (mod 2). (3.39)
n=0
6

This completes the proof by induction of (3.32). 2

We now prove Theorem 1.4.

Proof of Theorem 1.4. Employing (3.22) in (3.32), we obtain (1.9). 2

Acknowledgments

The authors are extremely grateful to Professor Michael Hirschhorn who read our
manuscript with great care, uncovered several errors and oered his valuable suggestions
which have substantially improved our paper. The authors also thank the anonymous
referee for his/her helpful suggestions and comments.

References

[1] Z. Ahmed, N.D. Baruah, New congruences for Andrews singular overpartitions, Int. J. Number
Theory 7 (11) (2015) 22472264.
[2] Z. Ahmed, N.D. Baruah, Parity results for broken 5-diamond, 7-diamond and 11-diamond partitions,
Int. J. Number Theory 2 (11) (2015) 527542.
[3] G.E. Andrews, Singular overpartitions, Int. J. Number Theory 11 (2015) 15231533.
[4] N.D. Baruah, K. Nath, Some results on 3-cores, Proc. Amer. Math. Soc. 142 (2) (2014) 441448.
[5] N.D. Baruah, K. Nath, Innite families of arithmetic identities and congruences for bipartitions
with 3-cores, J. Number Theory 149 (2015) 92104.
[6] N.D. Baruah, K.K. Ojah, Analogues of Ramanujans partition identities and congruences arising
from his theta function and modular equation, Ramanujan J. 28 (2012) 385407.
[7] S.C. Chen, Congruences for t-core partition functions, J. Number Theory 133 (12) (2013) 40364046.
[8] S.C. Chen, M.D. Hirschhorn, J.A. Sellers, Arithmetic properties of singular overpartitions, Int. J.
Number Theory 11 (2015) 14631476.
[9] S.P. Cui, N.S.S. Gu, Arithmetic properties of -regular partitions, Adv. Appl. Math. 51 (2013)
507523.
94 T. Kathiravan, S.N. Fathima / Journal of Number Theory 182 (2018) 8394

[10] F. Garvan, Some congruence for partitions that are p-cores, Proc. Lond. Math. Soc. 66 (1993)
449478.
[11] F. Garvan, D. Kim, D. Stanton, Cranks and t-cores, Invent. Math. 101 (1990) 117.
[12] F. Garvan, D. Kim, D. Stanton, More cranks and t-cores, Bull. Aust. Math. Soc. 63 (2001) 379391.
[13] M.D. Hirschhorn, J.A. Sellers, Some amazing facts about 4-cores, J. Number Theory 60 (1996)
5169.
[14] G. James, A. Kerber, The Representation Theory of the Symmetric Group, AddisonWesley Pub-
lishing Co., Reading, Mass., 1981.
[15] T. Kathiravan, S.N. Fathima, Some new congruences for Andrews singular overpartitions, J. Num-
ber Theory 173 (2017) 378393.
[16] M.S. Mahadeva Naika, D.S. Gireesh, Congruences for Andrews singular overpartitions, J. Number
Theory 165 (2016) 109130.
[17] G. de B. Robinson, R. Brauer, On a conjecture by Nakayama, Trans. R. Soc. Can. Sect. III 41
(1947) 2025.
[18] S. Ramanujan, Some properties of p(n), the number of partitions of n, Proc. Cambridge Philos.
Soc. 19 (1919) 207210.
[19] S. Ramanujan, Collected Papers, Cambridge Univ. Press, Cambridge, UK, 1927, reprinted by
Chelsea, New York, 1962, reprinted by the Amer. Math. Soc., Providence, RI, 2000.
[20] E.X.W. Xia, Arithmetic properties of bipartitions with 3-cores, Ramanujan J. 38 (3) (2015) 529548.
[21] O.X.M. Yao, New parity results for broken 11-diamond partitions, J. Number Theory 140 (2014)
267276.