Multiranks For Partitions Into Multi-Colors: 1 Introduction and Motivation
Multiranks For Partitions Into Multi-Colors: 1 Introduction and Motivation
Multiranks For Partitions Into Multi-Colors: 1 Introduction and Motivation
Roberta R. Zhou
School of Mathematical Sciences
Dalian University of Technology
Dalian 116024, P. R. China
[email protected]
Submitted: Apr 1, 2012; Accepted: Apr 8, 2012; Published: Apr 16, 2012
Mathematics Subject Classifications: Primary 11P83 and
Secondary 05A17, 05A19, 11F03, 11P81
Abstract
We generalize Hammond-Lewis birank to multiranks for partitions into colors
and give combinatorial interpretations for multipartitions such as b(n) defined by
H. Zhao and Z. Zhong and Qp1 ,p2 (n) defined by Toh congruences modulo 3, 5, 7.
Another proof has been given by B. Kim [24]. He defined a crank analog M 0 (m, N, n) for
a(n) and proved that
for all nonnegative integers n, where M 0 (m, N, n) is the number of partition of n with
crank congruent to m modulo N .
Hammond and Lewis [21] investigated some elementary results for 2-colored partitions
mod 5. They defined
birank(1 , 2 ) = #(1 ) #(2 ),
where #() is the number of parts in the partition . They explained that the residue of
the birank mod 5 divided the bipattitions of n into 5 equal classes provided n 2, 3 or 4
(mod 5). F. G. Garvan [19] found two other analogs the Dyson-birank and the 5-core-
birank.
H. Zhao and Z. ZhongP [31] have also investigated the arithmetic properties of a certain
function b(n) given by n=0 b(n)q n = (q; q)2 2 2 2
(q ; q ) . They have found b(5n + 4) 0
(mod 5) and b(7n + 2) b(7n + 3) b(7n + 4) b(7n + 6) 0 (mod 7) for any n > 0.
Toh [28] has also given lots of partition congruences such as Q(p1 ,p2 ) (n) 0 (mod `)
for prime number `, which is defined in section 3.
The main purpose of this paper is to define multirank for partition into colors and
prove multipartitions congruences applying the method of [18], which uses roots of unity.
It also has used the modified Jacobi triple product identity and quintuple product identity
as follows:
Let T (z, q) and T 2 (z, q) denote the two variable generating functions for NC1222(m, n)
and NC21222(m, n),
XX 1
T (z, q) := NC1222(m, n)z m q n = (5a)
mZ n=0
[zq, q/z; q] [zq 2 , q 2 /z; q 2 ]
XX 1
T 2 (z, q) := NC21222(m, n)z m q n = . (5b)
mZ n=0
[zq, q/z; q] [z 3q 2 , q 2 /z 3 ; q 2 ]
NC12 22 (m, n) = NC12 22 (m, n) and NC12 22 (m, t, n) = NC12 22 (tm, t, n);
NC2 12 22 (m, n) = NC2 12 22 (m, n) and NC2 12 22 (m, t, n) = NC2 12 22 (tm, t, n).
NC12 22 (, 5, n)q n .
P
Denote NC12 22 [] by NC12 22 [] := n>0
b(5n + 4)
NC12 22 (0, 5, 5n + 4) = NC12 22 (1, 5, 5n + 4) = NC12 22 (2, 5, 5n + 4) = .
5
Proof. Putting z = in (5a), where = exp 2i
5
, gives
1
T (, q) =
(q; q) ( 1 q; q) 2 2
(q ; q ) (
1 q 2 ; q 2 )
[q, 2 q, 2 q; q] [q 2 , 2 q 2 , 2 q 2 ; q 2 ]
= .
(q 5 ; q 5 ) (q 10 ; q 10 )
q2 q3
1 q
+( + ) + .
hq 10 ; q 25 i hq 10 ; q 50 i hq 5 ; q 25 i hq 20 ; q 50 i hq 10 ; q 25 i hq 20 ; q 50 i
While the 5-dissection of T (, q) is
T (, q) = NC12 22 [0] NC12 22 [2] + ( + 1 )(NC12 22 [1] NC12 22 [2]).
It follows that
X 1 q3
{NC12 22 (0, 5, n) NC12 22 (2, 5, n)}q n = +
n>0
hq 5 ; q 25 i hq 10 ; q 50 i hq 10 ; q 25 i hq 20 ; q 50 i
X q q2
{NC12 22 (1, 5, n) NC12 22 (2, 5, n)}q n = +
n>0
hq 10 ; q 25 i hq 10 ; q 50 i hq 5 ; q 25 i hq 20 ; q 50 i
q3
.
hq 10 ; q 25 i hq 20 ; q 50 i
No term involving q 5n+4 appears on the right side of the last identity, we finish the
proof of Theorem 1.
Theorem 2. For n > 0,
NC2 12 22 (0, 7, 7n + k) = NC2 12 22 (1, 7, 7n + k) = NC2 12 22 (2, 7, 7n + k)
b(7n + k)
= NC2 12 22 (3, 7, 7n + k) = for k = 2, 3, 4, 6.
7
Proof. Replacing z by $ in (5b), where $ = exp 2i
7
, gives
1
T 2 ($, q) =
($q; q) ($1 q; q) 3 2 2 3 2 2
($ q ; q ) ($ q ; q )
[q, $2 q, $2 q; q] [$4 q, $4 q; q 2 ]
= .
(q 7 ; q 7 )
Using the modified quintuple product identity (4a), we have
2
6n 3n 2+n
P 6n 6n2
($ + $ + + $ )q
T 2 ($, q) = .
(q 7 ; q 7 )
2
Since 3n 2+n 0, 1, 2, 5 (mod 7), and $6n + $6n2 + + $6n = 0 only when n 7 1,
hence the coefficient of q n on the right side of the last identity is zero when n 2 (mod 7),
n 3 (mod 7), n 4 (mod 7), and n 6 (mod 7). The Theorem 2 is completed.
We have
NC12 22 (0, 5, 4) = NC12 22 (1, 5, 4) = NC12 22 (2, 5, 4) = 7
NC2 12 22 (0, 7, 4) = NC2 12 22 (1, 7, 4) = NC2 12 22 (2, 7, 4) = NC2 12 22 (3, 7, 4) = 5.
pod(n) denote the number of partitions of n where the odd parts are distinct;
ped(n) denote the number of partitions of n where the even parts are distinct.
And let Q(p1 ,p2) (n) denote the number of partitions of n into two colors (say, red and blue),
where the parts colored red satisfy restrictions of partitions counted by p1(n), while the
parts colored blue satisfy restrictions of partitions counted by p2(n).
If we denote
then we can say them as partitions into 4-colors. For C134122, we define the sum of
parts s, a weight w13 41 22 and a multirank R13 41 22 (), by
s() = (1 ) + 2(2 ) + (3 ) + (4 )
w13 41 22 () = (1)#(2 )
R13 41 22 () = #(3 ) #(4 ).
Since
R13 41 22 (1 , 22 , 3 , 4 ) = R13 41 22 (1 , 22 , 4 , 3 ),
hence
NC13 41 22(m, n) = NC13 41 22(m, n) and NC13 41 22(m, t, n) = NC13 41 22(tm, t, n).
Suppose is primitive 3th root of unity. By letting z = in (6) and using Jacobi
triple product identity (1), we have
XX (q; q) (q 2 ; q 4 )
NC13 41 22 (m, n) m q n =
mZ n=0
(q; q) ( 1 q; q)
P n 2n2
(q 2 ; q 2 ) (q 2 ; q 4 ) n= (1) q
= = .
(q 3 ; q 3 ) (q 3 ; q 3 )
Since the coefficient of q n on the right side of the last identity is zero when n 1 (mod 3),
and 1 + + 2 is a minimal polynomial in Z[], we must have the result following as
s() = (1 ) + (2 ) + (3 )
R13 23 () = #(2 ) #(3 ).
Let NC13 23 (m, n) denote the number of 3-colored partitions of n if s() = n with multi-
rank m, and NC13 23 (m, t, n) denote the number of 3-colored partitions of n with multirank
congruent to m (mod t), so that
X
NC13 23 (m, n) = 1.
C13 23 ,s()=n,
R13 23 ()=m
Obviously
1
Since the coefficient of q n in the q-expansion of (q 3 ;q 6 )
is zero when n 1 (mod 3), we
have the following Theorem.
Q(po,ped) (3n+1)
NC13 23(0, 3, 3n + 1) = NC13 23(1, 3, 3n + 1) = NC13 23(2, 3, 3n + 1) = .
3
We define
For = (1 ,2 ,23 ,24 ), we denote the sum of parts s, and a multirank R124222(), by
Let NC124222(m, n) denote the number of 4-colored partitions of n if s() = n with multirank
m, and NC124222(m, t, n) denote the number of 4-colored partitions of n with multirank
congruent to m (mod t), so that
X
NC124222(m, n) = 1.
C12 42 22 ,s()=n,
R12 42 22 ()=m
Similarly
NC12 42 22(m, n) = NC12 42 22(m, n) and NC12 42 22(m, t, n) = NC12 42 22(tm, t, n).
Q(pod,pod) (3n+2)
NC12 42 22(0, 3, 3n+2) = NC12 42 22(1, 3, 3n+2) = NC12 42 22(2, 3, 3n+2) = .
3
Chen and Lin [10] has proved Q(pod,pod) (3n+2) 0 (mod 3).
Proof. Putting z = in (8) and utilizing Jacobi triple product identity (1), gives
P n2
XX
m n (q; q 2 )2 n= q
NC12 42 22 (m, n) q = = .
mZ n=0
(q 2 ; q 2 ) ( 1 q 2 ; q 2 ) (q 6 ; q 6 )
Since the coefficient of q n on the right side of the last identity is zero when n 2 (mod 3),
we complete the theorem.
It can be said as partitions into 3-colors. For = (41 , 2 , 3 ) C12 41 , we define the
sum of parts s, a weight w12 41 and a multirank R12 41 (), by
s() = 4(1 ) + (2 ) + (3 )
w12 41 () = (1)#(1 )
R12 41 () = #(2 ) #(3 ).
Then we have
XX (q 4 ; q 4 )
NC12 41 (m, n)z m q n = . (9)
mZ n=0
(zq; q) (z 1 q; q)
The last line depends only on classical identities of Jacobi (1) and (2).
If and only if n 5 2, we have 2n + 2n2 + + 2n = 0. Obviously
0, m 5 0, 2; 0, n 5 0, 4;
m n+1
12 + 4m 5 4, m 5 1; and 5 1, n 5 1, 3; (10)
2 2
3, m 5 3, 4; 3, n 5 2.
Q(p,ped) (5n+2)
NC1241 (0, 5, 5n+2) = NC1241 (1, 5, 5n+2) = NC1241 (2, 5, 5n+2) = .
5
We call the elements of C134123 6-colored partitions. For C13 41 23 , we define the sum
of parts s, a weight w13 41 23 and a multirank R134123(), by
Q(p,po) (5n+4)
NC134123(0, 5, 5n+4) = NC134123(1, 5, 5n+4) = NC134123(2, 5, 5n+4) = .
5
Proof. By replacing z by in (11), we write
XX (q 4 ; q 4 )
NC13 41 23 (m, n) m q n =
mZ n=0
[q, q, 1 q; q 2 ] [ 2 q 4 , 2 q 4 ; q 4 ]
1 2 2 2 2 4 4 4 4 4 (q 4 ; q 4 )2
= q , q, q; q
q , q ; q
.
(q 5 ; q 10 ) (q 20 ; q 20 ) (q 2 ; q 2 )
We see that in the q-expansion on the right side of the last equation the coefficient of q n
is zero when n 4 (mod 5). The proof of Theorem 7 has been finished.
We define
C13 42 24 = {(1 , 2 , 3 , 24 , 45 , 46 )|1 , 2 , 3 O, 4 D, 5 , 6 P}.
We can say them as partitions into 6-colors. For C13 42 24 , we denote the sum of parts
s, a weight w13 42 24 and a multirank R13 42 24 (), by
s() = (1 ) + (2 ) + (3 ) + 2(4 ) + 4(5 ) + 4(6 )
w13 42 24 () = (1)#(4 )
R13 42 24 () = #(2 ) #(3 ) + #(5 ) #(6 ).
The number of 6-colored partitions of n if s() = n (counted according to the weight
w134224) with multirank m is denoted by NC134224(m, n), so that
X
NC13 42 24 (m, n) = w13 42 24 ().
C13 42 24 ,s()=n,
R13 42 24 ()=m
Q(po,pod)(5n+2)
NC134224(0, 5, 5n+2) = NC134224(1, 5, 5n+2) = NC134224(2, 5, 5n+2) = .
5
Proof. The proof of Theorem 8 is similar to Theorem 7. By letting z = in (13), we get
XX (q 2 ; q 2 )
NC13 42 24 (m, n) m q n =
mZ n=0
[q, q, 1 q; q 2 ] [q 4 , 1 q 4 ; q 4 ]
[q , q, 2 q; q 2 ] [q 4 , 2 q 4 , 2 q 4 ; q 4 ]
2 2
= .
(q 5 ; q 10 ) (q 20 ; q 20 )
Using Jacobi triple product identity (1) and (2), the numerator infinite products have
the following series expression
n+1
(1)n q 4( 2 ) { 2n + 2n2 + + 2n }.
2
X X
(1)m q m 2m
m= n>0
We call them as partitions into 7-colors. For C112242, we define the sum of parts s, a
weight w112242 and a multirank R11 22 42 (), by
If we simply put z = 1 in the identity (14), and read off the coefficients of like powers of
q, we find
X
NC11 22 42 (m, n) = Q(pe,ped) (n).
m=
Using Jacobi triple product identity (1) and Entry 8(x) of [5] P.114, the numerator
infinite products have the following series expression
n
(1 + 3n)q 6( 2 )+5n .
2
X X
(1)m q m 2m
m= n=
References
[1] G. E. Andrews. q-Series: Their Development and Application in Analysis, Number
Theory, Combinatorics, Physics, and Computer Algebra. CBMS Regional Conference
Series in Mathematics, No. 66; Amer. Math. Society, 1986.
[2] G. E. Andrews and F. G. Garvan. Dysons crank of a partition. Bull. Amer. Math.
Soc. (N.S.), 18:167171, 1988.
[3] A. O. L. Atkin and H. P. F. Swinnerton-Dyer. Some properties of partitions. Proc.
London Math. Soc., 4(3): 84106, 1954.
[4] W. N. Bailey. Generalized Hypergeometric Series. Cambridge University Press,
Cambridge, 1935.