Skip to main content

Theory and Modern Applications

Circular summation formulas for theta function ϑ 4 (z|τ)

Abstract

In this paper, we obtain the circular summation formulas of the theta function ϑ 4 (z|τ) and show the corresponding alternating summations and inverse relations. Some applications are also considered.

MSC:11F27, 33E05.

1 Introduction

Throughout this paper we take q= e π i τ , Im(τ)>0, zC. The classical Jacobi’s four theta functions ϑ i (z|τ), i=1,2,3,4, are defined as follows, with the notation of Tannery and Molk (see, e.g., [13]):

ϑ 1 (z|τ)=i q 1 4 n = ( 1 ) n q n ( n + 1 ) e ( 2 n + 1 ) i z ,
(1.1)
ϑ 2 (z|τ)= q 1 4 n = q n ( n + 1 ) e ( 2 n + 1 ) i z ,
(1.2)
ϑ 3 (z|τ)= n = q n 2 e 2 n i z ,
(1.3)
ϑ 4 (z|τ)= n = ( 1 ) n q n 2 e 2 n i z .
(1.4)

From the Jacobi theta functions (1.1)-(1.4), via the direct calculation and applying the induction, we have the following properties, respectively:

ϑ 1 (z+nπ|τ)= ( 1 ) n ϑ 1 (z|τ), ϑ 1 (z+nπτ|τ)= ( 1 ) n q n 2 e 2 n i z ϑ 1 (z|τ),
(1.5)
ϑ 2 (z+nπ|τ)= ( 1 ) n ϑ 2 (z|τ), ϑ 2 (z+nπτ|τ)= q n 2 e 2 n i z ϑ 2 (z|τ),
(1.6)
ϑ 3 (z+nπ|τ)= ϑ 3 (z|τ), ϑ 3 (z+nπτ|τ)= q n 2 e 2 n i z ϑ 3 (z|τ),
(1.7)
ϑ 4 (z+nπ|τ)= ϑ 4 (z|τ), ϑ 4 (z+nπτ|τ)= ( 1 ) n q n 2 e 2 n i z ϑ 4 (z|τ).
(1.8)

From the Jacobi theta functions (1.1)-(1.4), via a direct calculation, we also have the following transformation formulas:

ϑ 1 ( z + π 2 | τ ) = ϑ 2 (z|τ), ϑ 1 ( z + π τ 2 | τ ) =i q 1 4 e i z ϑ 4 (z|τ),
(1.9)
ϑ 2 ( z + π 2 | τ ) = ϑ 1 (z|τ), ϑ 2 ( z + π τ 2 | τ ) = q 1 4 e i z ϑ 3 (z|τ),
(1.10)
ϑ 3 ( z + π 2 | τ ) = ϑ 4 (z|τ), ϑ 3 ( z + π τ 2 | τ ) = q 1 4 e i z ϑ 2 (z|τ),
(1.11)
ϑ 4 ( z + π 2 | τ ) = ϑ 3 (z|τ), ϑ 4 ( z + π τ 2 | τ ) =i q 1 4 e i z ϑ 1 (z|τ).
(1.12)

From the above equations we easily obtain

ϑ 1 ( z + π 2 + π τ 2 | τ ) = q 1 4 e i z ϑ 3 ( z | τ ) , ϑ 2 ( z + π 2 + π τ 2 | τ ) = i q 1 4 e i z ϑ 4 ( z | τ ) ,
(1.13)
ϑ 3 ( z + π 2 + π τ 2 | τ ) = i q 1 4 e i z ϑ 1 ( z | τ ) , ϑ 4 ( z + π 2 + π τ 2 | τ ) = q 1 4 e i z ϑ 2 ( z | τ ) .
(1.14)

From (1.5)-(1.12), we can obtain the following lemmas.

Lemma 1.1 For n any positive integer, we have

ϑ 1 ( z + n π 2 | τ ) = { i n ϑ 1 ( z | τ ) , n  is even , i n 1 ϑ 2 ( z | τ ) , n  is odd ,
(1.15)
ϑ 2 ( z + n π 2 | τ ) = { i n ϑ 2 ( z | τ ) , n  is even , i n 1 ϑ 1 ( z | τ ) , n  is odd ,
(1.16)
ϑ 3 ( z + n π 2 | τ ) = { ϑ 3 ( z | τ ) , n  is even , ϑ 4 ( z | τ ) , n  is odd ,
(1.17)
ϑ 4 ( z + n π 2 | τ ) = { ϑ 4 ( z | τ ) , n  is even , ϑ 3 ( z | τ ) , n  is odd .
(1.18)

Lemma 1.2 For n any positive integer, we have

ϑ 1 ( z + n π τ 2 | τ ) = { i n q n 2 4 e n i z ϑ 1 ( z | τ ) , n  is even , i n q n 2 4 e n i z ϑ 4 ( z | τ ) , n  is odd ,
(1.19)
ϑ 2 ( z + n π τ 2 | τ ) = { q n 2 4 e n i z ϑ 2 ( z | τ ) , n  is even , q n 2 4 e n i z ϑ 3 ( z | τ ) , n  is odd ,
(1.20)
ϑ 3 ( z + n π τ 2 | τ ) = { q n 2 4 e n i z ϑ 3 ( z | τ ) , n  is even , q n 2 4 e n i z ϑ 2 ( z | τ ) , n  is odd ,
(1.21)
ϑ 4 ( z + n π τ 2 | τ ) = { i n q n 2 4 e n i z ϑ 4 ( z | τ ) , n  is even , i n q n 2 4 e n i z ϑ 1 ( z | τ ) , n  is odd .
(1.22)

On p.54 in Ramanujan’s lost notebook (see [[4], p.54, Entry 9.1.1], or [[5], p.337]), Ramanujan recorded the following claim (without proof), which is now well known as Ramanujan’s circular summation. The appellation of ‘circular summation’ is due to Son (see [[5], p.338]).

Theorem 1.3 (Ramanujan’s circular summation)

For each positive integer n and |ab|<1,

n / 2 < r n / 2 ( k = k r ( mod n ) a k ( k + 1 ) / ( 2 n ) b k ( k 1 ) / ( 2 n ) ) n =f(a,b) F n (ab),
(1.23)

where

F n (q):=1+2n q ( n 1 ) / 2 +,n3.
(1.24)

Ramanujan’s theta function f(a,b) is defined by

f(a,b)= n = a n ( n + 1 ) / 2 b n ( n 1 ) / 2 ,|ab|<1.
(1.25)

By the definition of Ramanujan’s theta function above and routine calculations, we can rewrite Ramanujan’s circular summation (1.23) as follows (see, for details [[5], p.338]).

Theorem 1.4 (Ramanujan’s circular summation)

Let U k = a k ( k + 1 ) / ( 2 n ) and V k = b k ( k 1 ) / ( 2 n ) . For each positive integer n and |ab|<1,

k = 0 n 1 U k n f n ( U n + k U k , V n k V k ) =f(a,b) F n (ab),
(1.26)

where

F n (q):=1+2n q ( n 1 ) / 2 +,n3.
(1.27)

If we are going to apply the transformation τ 1 τ to Ramanujan’s identity, it will be convenient to convert Ramanujan’s theorem into one involving the classical theta function ϑ 3 (z|τ) defined by (1.3). Surprisingly Chan et al. [6] prove that Theorem 1.5 below is equivalent to Theorem 1.3.

Theorem 1.5 (Ramanujan’s circular summation)

For any positive integer n2,

k = 0 n 1 q k 2 e 2 k i z ϑ 3 n (z+kπτ|nτ)= ϑ 3 (z|τ) F n (τ).
(1.28)

When n3,

F n (q)=1+2n q n 1 +.
(1.29)

Chan et al. [6] also showed that Theorem 1.6 below is equivalent; we have Theorem 1.5 by applying the Jacobi imaginary transformation formula [[3], p.475].

Theorem 1.6 (Ramanujan’s circular summation)

For any positive integer n, there exists a quantity G n (τ) such that

k = 0 n 1 ϑ 3 n ( z + k π n | τ ) = G n (τ) ϑ 3 (nz|nτ),
(1.30)

where

G n (τ)= n ( i τ ) ( 1 n ) / 2 F n ( 1 n τ ) .
(1.31)

Ramanujan’s circular summation is an interesting subject in his notebook. On the subject of Ramanujan’s circular summation and related identities of theta functions and their various extensions, a remarkably large number of investigations have appeared in the literature (see, for example, Andrews, Berndt, Rangachari, Ono, Ahlgren, Chua, Murayama, Son, Chan, Liu, Ng, Chan, Shen, Cai, Zhu, and Xu et al. [525]).

Recently, Zeng [23] extended Ramanujan’s circular summation in the following form.

Theorem 1.7 For any nonnegative integers n, k, a, and b with a+b=k, there exists a quantity R 33 (a,b; y a b , τ k n 2 ) such that

s = 0 k n 1 ϑ 3 a ( z k n + y a + π s k n | τ k n 2 ) ϑ 3 b ( z k n y b + π s k n | τ k n 2 ) = R 33 ( a , b ; y a b , τ k n 2 ) ϑ 3 ( z | τ ) .
(1.32)

Chan and Liu [12] further extended Theorem 1.7 in the following general form.

Theorem 1.8 Suppose y 1 , y 2 ,, y n are n complex numbers such that y 1 + y 2 ++ y n =0; there exists a quantity G m , n ( y 1 , y 2 ,, y n |τ) such that

k = 0 m n 1 j = 1 n ϑ 3 ( z + y j + k π m n | τ ) = G m , n ( y 1 , y 2 ,, y n |τ) ϑ 3 ( m n z | m 2 n τ ) ,
(1.33)

In the present paper, motivated by [10, 12], and [6], by applying the theory and method of elliptic functions, we obtain the circular summation formulas of theta functions ϑ 4 (z|τ) and show the corresponding alternating summations and inverse relations. We also give some applications and derive some interesting identities of theta functions.

2 Ramanujan’s circular summation formula for theta functions ϑ 4 (z|τ)

In the present section, we obtain Ramanujan’s circular summation formula for the theta functions ϑ 4 (z|τ). We now state our main result as follows.

Theorem 2.1 Suppose that m, n are any positive integers, p is any integer; y 1 , y 2 ,, y n are any complex numbers.

  • When y 1 + y 2 ++ y n = p π m + n π 2 , we have

k = 0 m n 1 j = 1 n ϑ 4 ( z + y j + k π m n | τ ) = R 4 , 4 ( y 1 , y 2 ,, y n ;m,n,p;τ) ϑ 3 ( m n z | m 2 n τ ) .
(2.1)
  • When y 1 + y 2 ++ y n = ( 2 p + 1 ) π 2 m + n π 2 , we have

k = 0 m n 1 j = 1 n ϑ 4 ( z + y j + k π m n | τ ) = R 4 , 4 ( y 1 , y 2 ,, y n ;m,n,p;τ) ϑ 4 ( m n z | m 2 n τ ) .
(2.2)

Here

R 4 , 4 ( y 1 , y 2 ,, y n ;m,n,p;τ)=mn r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) .
(2.3)

Proof Let f(z) be the left-hand side of (2.1) with z z m n , τ τ m 2 n . We have

f(z)= k = 0 m n 1 j = 1 n ϑ 4 ( z m n + y j + k π m n | τ m 2 n ) .
(2.4)

By (1.8) for n=1, we easily obtain

f ( z + π ) = k = 0 m n 1 j = 1 n ϑ 4 ( z + π m n + y j + k π m n | τ m 2 n ) = k = 1 m n 1 j = 1 n ϑ 4 ( z m n + y j + k π m n | τ m 2 n ) + j = 1 n ϑ 4 ( z m n + y j | τ m 2 n ) .
(2.5)

Comparing (2.4) and (2.5), we have

f(z)=f(z+π).
(2.6)

By (1.8), we obtain

f ( z + π τ ) = k = 0 m n 1 j = 1 n ϑ 4 ( z + π τ m n + y j + k π m n | τ m 2 n ) = k = 0 m n 1 j = 1 n ϑ 4 ( z m n + y j + k π m n + m π τ m 2 n | τ m 2 n ) = k = 0 m n 1 j = 1 n ( 1 ) m q 1 n e 2 i m ( z m n + y j + k π m n ) ϑ 4 ( z m n + y j + k π m n | τ m 2 n ) = ( 1 ) m n q 1 e 2 i z e 2 i m ( y 1 + y 2 + + y n ) f ( z ) .
(2.7)
  • When y 1 + y 2 ++ y n = p π m + n π 2 in (2.7), we have

    f(z+πτ)= q 1 e 2 i z f(z).
    (2.8)

We construct the function f ( z ) ϑ 3 ( z | τ ) , by (1.7) for n=1, (2.6) and (2.8), we find that the function f ( z ) ϑ 3 ( z | τ ) is an elliptic function with double periods π and πτ and has only a simple pole at z= π 2 + π τ 2 in the period parallelogram. Hence the function f ( z ) ϑ 3 ( z | τ ) is a constant, say C 4 , 4 ( 1 ) ( y 1 , y 2 ,, y n ;τ), and we have

f(z)= C 4 , 4 ( 1 ) ( y 1 , y 2 ,, y n ;τ) ϑ 3 (z|τ),

or, equivalently,

k = 0 m n 1 j = 1 n ϑ 4 ( z m n + y j + k π m n | τ m 2 n ) = C 4 , 4 ( 1 ) ( y 1 , y 2 ,, y n ;τ) ϑ 3 (z|τ).
(2.9)

Letting

zmnzandτ m 2 nτ

in (2.9), and then setting

R 4 , 4 ( 1 ) ( y 1 , y 2 ,, y n ;m,n,p;τ)= C 4 , 4 ( 1 ) ( y 1 , y 2 , , y n ; m 2 n τ ) ,

we arrive at (2.1).

Setting

zz+ y j + k π m n

in (1.4), via the direct calculation, we obtain

j = 1 n ϑ 4 ( z + y j + k π m n | τ ) = r 1 , , r n = ( 1 ) r 1 + + r n q r 1 2 + + r n 2 × e 2 ( r 1 + + r n ) i z e 2 i ( r 1 y 1 + + r n y n ) e 2 k π i m n ( r 1 + + r n ) .
(2.10)

Setting

zmnzandτ m 2 nτ

in (1.3), we get

ϑ 3 ( m n z | m 2 n τ ) = r = q m 2 n r 2 e 2 m n r i z .
(2.11)

Substituting (2.10) and (2.11) into (2.1), we find that

r 1 , , r n = ( 1 ) r 1 + + r n q r 1 2 + + r n 2 e 2 ( r 1 + + r n ) i z e 2 i ( r 1 y 1 + + r n y n ) k = 0 m n 1 e 2 k π i m n ( r 1 + + r n ) = R 4 , 4 ( 1 ) ( y 1 , y 2 , , y n ; m , n , p ; τ ) r = q m 2 n r 2 e 2 m n r i z .
(2.12)

Equating the constants of both sides of (2.12), we get (2.3).

  • When y 1 + y 2 ++ y n = ( 2 p + 1 ) π 2 m + n π 2 in (2.7), we have

    f(z+πτ)= q 1 e 2 i z f(z).
    (2.13)

We construct the function f ( z ) ϑ 4 ( z | τ ) , by (1.8) for n=1, (2.6) and (2.13), we find that the function f ( z ) ϑ 4 ( z | τ ) is an elliptic function with double periods π and πτ and has only a simple pole at z= π τ 2 in the period parallelogram. Hence the function f ( z ) ϑ 4 ( z | τ ) is a constant, say C 4 , 4 ( 2 ) ( y 1 , y 2 ,, y n ;τ), and we have

f(z)= C 4 , 4 ( 2 ) ( y 1 , y 2 ,, y n ;τ) ϑ 4 (z|τ),

or, equivalently,

k = 0 m n 1 j = 1 n ϑ 4 ( z m n + y j + k π m n | τ m 2 n ) = C 4 , 4 ( 2 ) ( y 1 , y 2 ,, y n ;τ) ϑ 4 (z|τ).
(2.14)

Letting

zmnzandτ m 2 nτ

in (2.14), and then setting

R 4 , 4 ( 2 ) ( y 1 , y 2 ,, y n ;m,n,p;τ)= C 4 , 4 ( 2 ) ( y 1 , y 2 , , y n ; m 2 n τ ) ,

we arrive at (2.2).

Similarly, we may obtain

R 4 , 4 ( 2 ) ( y 1 , y 2 ,, y n ;m,n,p;τ)=mn r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) .
(2.15)

Clearly,

R 4 , 4 ( y 1 , y 2 ,, y n ;m,n,p;τ)= R 4 , 4 ( 1 ) ( y 1 , y 2 ,, y n ;m,n,p;τ)= R 4 , 4 ( 2 ) ( y 1 , y 2 ,, y n ;m,n,p;τ).

The proof is complete. □

In the following we will prove that the following main result of Chan and Liu is a special case of Theorem 2.1.

Corollary 2.2 (Chan and Liu [[12], p.1191, Theorem 4])

Suppose that m, n are any positive integers; y 1 , y 2 ,, y n are any complex numbers and y 1 + y 2 ++ y n =0. We have

k = 0 m n 1 j = 1 n ϑ 3 ( z + y j + k π m n | τ ) = R 3 , 3 ( y 1 , y 2 ,, y n ;m,n;τ) ϑ 3 ( m n z | m 2 n τ ) ,
(2.16)

where

R 3 , 3 ( y 1 , y 2 ,, y n ;m,n;τ)=mn r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) .
(2.17)

Proof First, setting y 1 + y 2 ++ y n =0 in (2.1), we have p= m n 2 , noting that p is any integer, hence we say that mn is even. Setting

zz+ π 2

in (2.1) of Theorem 2.1 and applying the properties (1.12) and (1.17), we have

k = 0 m n 1 j = 1 n ϑ 3 ( z + y j + k π m n | τ ) = R 3 , 3 ( y 1 , y 2 ,, y n ;m,n;τ) ϑ 3 ( m n z | m 2 n τ ) .
(2.18)

Next taking y 1 + y 2 ++ y n =0 in (2.2), we have p= m n + 1 2 , noting that p is any integer, hence we say that mn is odd. Setting

zz+ π 2

in (2.2) of Theorem 2.1 and applying the properties (1.12) and (1.18), this leads to the following:

k = 0 m n 1 j = 1 n ϑ 3 ( z + y j + k π m n | τ ) = R 3 , 3 ( y 1 , y 2 ,, y n ;m,n;τ) ϑ 3 ( m n z | m 2 n τ ) .
(2.19)

Obviously, by (2.18) and (2.19), for any positive integers mn, we obtain (2.16).

We easily see that the R 4 , 4 ( 1 ) ( y 1 , y 2 ,, y n ;m,n,p;τ) and R 4 , 4 ( 2 ) ( y 1 , y 2 ,, y n ;m,n,p;τ) are independent of z, therefore, when y 1 + y 2 ++ y n =0, we find that

R 3 , 3 ( y 1 , y 2 ,, y n ;m,n;τ)= R 4 , 4 ( 1 ) ( y 1 , y 2 ,, y n ;m,n;τ)= R 4 , 4 ( 2 ) ( y 1 , y 2 ,, y n ;m,n;τ).

The proof is complete. □

Remark 2.3 Obviously, Theorem 2.1 implies the well-known result (2.16), but we see that from (2.16) we do not obtain Theorem 2.1. However, we can reformulate the results of Chan and Liu as Theorem 2.1.

Suppose that m, n are any positive integers, p is any integer; y 1 , y 2 ,, y n are any complex numbers.

  • When y 1 + y 2 ++ y n = p π m , we have

    k = 0 m n 1 j = 1 n ϑ 3 ( z + y j + k π m n | τ ) = R 3 , 3 ( y 1 , y 2 ,, y n ;m,n,p;τ) ϑ 3 ( m n z | m 2 n τ ) .
    (2.20)
  • When y 1 + y 2 ++ y n = ( 2 p + 1 ) π 2 m , we have

    k = 0 m n 1 j = 1 n ϑ 3 ( z + y j + k π m n | τ ) = R 3 , 3 ( y 1 , y 2 ,, y n ;m,n,p;τ) ϑ 4 ( m n z | m 2 n τ ) .
    (2.21)

Here

R 3 , 3 ( y 1 , y 2 ,, y n ;m,n,p;τ)=mn r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) .
(2.22)

3 The imaginary transformations formulas for Theorem 2.1

In the present section, we derive the corresponding imaginary transformations formulas for Theorem 2.1.

Theorem 3.1 Suppose that m, n are any positive integers, p is any integer; y 1 , y 2 ,, y n are any complex numbers.

  • When y 1 + y 2 ++ y n =mnp+ m 2 n 2 2 , we have

    k = 0 m n 1 q k 2 + 2 k p + m n k e ( 2 k + 2 p + m n ) i z j = 1 n ϑ 2 ( m z + y j π τ + m k π τ | m 2 n τ ) = F 4 , 4 ( 1 ) ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 3 ( z | τ ) .
    (3.1)
  • When y 1 + y 2 ++ y n = ( 2 p + 1 ) m n 2 + m 2 n 2 2 , we have

    k = 0 m n 1 q k 2 + k ( 2 p + 1 ) + m n k e ( 2 k + 2 p + m n + 1 ) i z j = 1 n ϑ 2 ( m z + y j π τ + m k π τ | m 2 n τ ) = F 4 , 4 ( 2 ) ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 2 ( z | τ ) .
    (3.2)

Here

F 4 , 4 ( 1 ) ( y 1 , y 2 , , y n ; m , n , p ; τ ) = ( i τ ) 1 n 2 ( m 2 n ) n 2 q y 1 2 + + y n 2 m 2 n R 4 , 4 ( 1 ) ( y 1 π m 2 n , y 2 π m 2 n , , y n π m 2 n ; m , n , p ; 1 m 2 n τ ) ,
(3.3)
F 4 , 4 ( 1 ) ( y 1 , y 2 , , y n ; m , n , p ; τ ) = k = 0 m n 1 q ( k + m n 2 ) 2 + m 2 n 2 2 r 1 , , r n = m ( r 1 + + r n ) = m n + k + p q m 2 n ( r 1 2 + + r n 2 ) 2 ( r 1 y 1 + + r n y n ) ,
(3.4)
F 4 , 4 ( 2 ) ( y 1 , y 2 , , y n ; m , n , p ; τ ) = ( i τ ) 1 n 2 ( m 2 n ) n 2 q y 1 2 + + y n 2 m 2 n R 4 , 4 ( 2 ) ( y 1 π m 2 n , y 2 π m 2 n , , y n π m 2 n ; m , n , p ; 1 m 2 n τ ) ,
(3.5)
F 4 , 4 ( 2 ) ( y 1 , y 2 , , y n ; m , n , p ; τ ) = k = 0 m n 1 q ( k + m n 2 1 2 ) 2 m n k + m 2 n 2 2 r 1 , , r n = m ( r 1 + + r n ) = m n + k + p q m 2 n ( r 1 2 + + r n 2 ) 2 ( r 1 y 1 + + r n y n ) .
(3.6)

Proof In (2.1) making the transformations τ 1 m 2 n τ , and then z z m n τ and y j y j π m 2 n for j=1,2,,n, (2.1) becomes

k = 0 m n 1 j = 1 n ϑ 4 ( m z + y j π τ + m k π τ m 2 n τ | 1 m 2 n τ ) = R 4 , 4 ( 1 ) ( y 1 π m 2 n , , y n π m 2 n ; m , n , p ; 1 m 2 n τ ) ϑ 3 ( z τ | 1 τ ) .
(3.7)

Applying the imaginary transformation formulas (see, e.g., [13])

ϑ 3 ( z τ | 1 τ ) = i τ e i z 2 π τ ϑ 3 (z|τ)and ϑ 4 ( z τ | 1 τ ) = i τ e i z 2 π τ ϑ 2 (z|τ)

to (3.7), via the suitable substitutions of the variable z and τ and noting that y 1 + y 2 ++ y n =mnp+ m 2 n 2 2 , and simplifying, we thus obtain (3.1) and (3.3). Applying the series expressions (1.3) and (1.4) in (3.1), via the direct calculation, we obtain (3.4).

In a similar manner, we use the imaginary transformation formula:

ϑ 4 ( z τ | 1 τ ) = i τ e i z 2 π τ ϑ 2 (z|τ),

and (2.2), and noting that y 1 + y 2 ++ y n = ( 2 p + 1 ) m n 2 + m 2 n 2 2 , we can prove (3.2), (3.5), and (3.6). Therefore we complete the proof of Theorem 3.1. □

4 The alternating Ramanujan’s circular summation formula

In this section, we will obtain the corresponding alternating Ramanujan’s circular summation formula from Theorem 2.1.

Theorem 4.1 Suppose that m, n are any positive integers, p is any integer; y 1 , y 2 ,, y n are any complex numbers.

  • When y 1 + y 2 ++ y n = p π m + n π 2 , we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 4 ( z + y j + k π m n | τ ) = ( 1 ) p R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 2 ( m n z | m 2 n τ ) , m  is even ,
    (4.1)
    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 1 ( z + y j + k π m n | τ ) = ( 1 ) p R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 2 ( m n z | m 2 n τ ) , m  is odd .
    (4.2)
  • When y 1 + y 2 ++ y n = ( 2 p + 1 ) π 2 m + n π 2 , we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 4 ( z + y j + k π m n | τ ) = ( 1 ) p + 1 R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 1 ( m n z | m 2 n τ ) , m  is even ,
    (4.3)
    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 1 ( z + y j + k π m n | τ ) = ( 1 ) p + 1 R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 1 ( m n z | m 2 n τ ) , m  is odd .
    (4.4)

Here

R 4 , 4 ( y 1 , y 2 ,, y n ;m,n,p;τ)=mn r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) .
(4.5)

Proof Let

zz+ m π τ 2

in Theorem 2.1.

By using (1.11), (1.12), and (1.22) we compute

j = 1 n ϑ 4 ( z + y j + k π m n + m π τ 2 | τ ) = { ( 1 ) k i m n q m 2 n 4 e m n i z e m i ( y 1 + + y n ) j = 1 n ϑ 4 ( z + y j + k π m n | τ ) , m  is even , ( 1 ) k i m n q m 2 n 4 e m n i z e m i ( y 1 + + y n ) j = 1 n ϑ 1 ( z + y j + k π m n | τ ) , m  is odd ,
(4.6)
ϑ 3 ( m n z + m 2 n π τ 2 | m 2 n τ ) = q m 2 n 4 e m n i z ϑ 2 ( m n z | m 2 n τ ) ,
(4.7)
ϑ 4 ( m n z + m 2 n π τ 2 | m 2 n τ ) =i q m 2 n 4 e m n i z ϑ 1 ( m n z | m 2 n τ ) .
(4.8)

Substituting (4.6) and (4.7) into (2.1) of Theorem 2.1 and simplifying, we get (4.1) of Theorem 4.1.

Substituting (4.6) and (4.8) into (2.2) of Theorem 2.1 and simplifying, we arrive at (4.3) of Theorem 4.1. This proof is complete. □

Corollary 4.2 Suppose that m, n are any positive integers, p is any integer; y 1 , y 2 ,, y n are any complex numbers.

  • When y 1 + y 2 ++ y n =0, we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 4 ( z + y j + k π m n | τ ) = i m n R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 2 ( m n z | m 2 n τ ) , m  is even ,
    (4.9)
    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 1 ( z + y j + k π m n | τ ) = i m n R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 2 ( m n z | m 2 n τ ) , m  is odd and  n  is even .
    (4.10)
  • When y 1 + y 2 ++ y n = π 2 m , we have

    k = 0 m n 1 ( 1 ) k + 1 j = 1 n ϑ 4 ( z + y j + k π m n | τ ) = i m n R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 1 ( m n z | m 2 n τ ) , m  is even ,
    (4.11)
    k = 0 m n 1 ( 1 ) k + 1 j = 1 n ϑ 1 ( z + y j + k π m n | τ ) = i m n R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 1 ( m n z | m 2 n τ ) , m  is odd and  n  is even .
    (4.12)
  • When y 1 + y 2 ++ y n =0and m, n are odd, we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 1 ( z + y j + k π m n | τ ) = i m n 1 R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 1 ( m n z | m 2 n τ ) .
    (4.13)

Here

R 4 , 4 ( y 1 , y 2 ,, y n ;m,n,p;τ)=mn r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) .
(4.14)

Proof If mn is even, setting p= m n 2 in y 1 + y 2 ++ y n = p π m + n π 2 , then we have y 1 + y 2 ++ y n =0 and ( 1 ) p = i m n . We directly deduce (4.9) of Corollary 4.2.

If mn is an even, setting p= m n 2 in y 1 + y 2 ++ y n = p π m + n π 2 + π 2 , we have y 1 + y 2 ++ y n = π 2 m . We get (4.11) of Corollary 4.2.

If mn is an odd, setting p= m n + 1 2 in y 1 + y 2 ++ y n = p π m + n π 2 + π 2 and ( 1 ) p = i m n 1 , we have y 1 + y 2 ++ y n =0. We get (4.13) of Corollary 4.2. The proof is complete. □

Corollary 4.3 Suppose that m, n are any positive integers. We have

k = 0 m n 1 ( 1 ) k ϑ 4 n ( z + k π m n | τ ) = i m n R 4 , 4 (m,n;τ) ϑ 2 ( m n z | m 2 n τ ) ,m is even,
(4.15)
k = 0 m n 1 ( 1 ) k ϑ 1 n ( z + k π m n | τ ) = i m n R 4 , 4 ( m , n ; τ ) ϑ 2 ( m n z | m 2 n τ ) , m  is odd and  n is even ,
(4.16)
k = 0 m n 1 ( 1 ) k ϑ 1 n ( z + k π m n | τ ) = i m n 1 R 4 , 4 (m,n;τ) ϑ 1 ( m n z | m 2 n τ ) ,m,n are odd,
(4.17)
k = 0 m n 1 ( 1 ) k + 1 ϑ 4 n ( z + π 2 m n + k π m n | τ ) = i m n R 4 , 4 ( m , n ; τ ) ϑ 1 ( m n z | m 2 n τ ) , m  is even ,
(4.18)
k = 0 m n 1 ( 1 ) k + 1 ϑ 1 n ( z + π 2 m n + k π m n | τ ) = i m n R 4 , 4 ( m , n ; τ ) ϑ 1 ( m n z | m 2 n τ ) , m  is odd and  n  is even ,
(4.19)

where

R 4 , 4 (m,n;τ)=mn r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) .
(4.20)

Proof We set y 1 = y 2 == y n =0 in (4.9), (4.10), and (4.13) of Corollary 4.2. We take y 1 = y 2 == y n = π 2 m n in (4.11) and (4.12) of Corollary 4.2. □

Remark 4.4 Taking m=1 in (4.15) and (4.17), and noting that

R 4 , 4 (n;τ)= R 4 , 4 (n;τ)=n r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 ,

we have

k = 0 n 1 ( 1 ) k ϑ 3 n ( z + k π n | τ ) = { i n R 4 , 4 ( n ; τ ) ϑ 2 ( n z | n τ ) , n  is even , i 1 n R 4 , 4 ( n ; τ ) ϑ 1 ( n z | n τ ) , n  is odd ,
(4.21)

which is just Chan’s result (see [[6], p.634, Theorem 4.2]). Hence (4.15) and (4.17) of Corollary 4.3 are the corresponding extension of Chan’s result.

Remark 4.5 We note that Chan’s result (see [[6], p.632, Theorem 4.2]) has a misprint; we have here corrected this point in (4.21).

Remark 4.6 Setting m=1 in (4.18), we have

k = 0 n 1 ( 1 ) k ϑ 1 n ( z + π 2 n + k π n | τ ) = i n R 4 , 4 (n;τ) ϑ 1 (nz|nτ),n is even.
(4.22)

Setting n=1 in (4.18), we have

k = 0 m 1 ( 1 ) k ϑ 4 ( z + π 2 m + k π m | τ ) = i m m ϑ 1 ( m z | m 2 τ ) ,m is even.
(4.23)

The above two formulas are the analogs of Boon’s results (see [[8], p.3440]).

Theorem 4.7 Suppose that m, n are any positive integers, p is any integer; y 1 , y 2 ,, y n are any complex numbers.

  • When y 1 + y 2 ++ y n = p π m + n π 2 and m is even, we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 3 ( z + y j + k π m n | τ ) = ( 1 ) p i m n R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 2 ( m n z | m 2 n τ ) .
    (4.24)
  • When y 1 + y 2 ++ y n = p π m + n π 2 and m is odd and n is even, we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 2 ( z + y j + k π m n | τ ) = ( 1 ) p i m n R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 2 ( m n z | m 2 n τ ) .
    (4.25)
  • When y 1 + y 2 ++ y n = p π m + n π 2 and m is odd and n is odd, we have

    k = 0 m n 1 ( 1 ) k + 1 j = 1 n ϑ 2 ( z + y j + k π m n | τ ) = ( 1 ) p i m n + 1 R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 1 ( m n z | m 2 n τ ) .
    (4.26)
  • When y 1 + y 2 ++ y n = ( 2 p + 1 ) π 2 m + n π 2 and m is even, we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 3 ( z + y j + k π m n | τ ) = ( 1 ) p i m n R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 1 ( m n z | m 2 n τ ) .
    (4.27)
  • When y 1 + y 2 ++ y n = ( 2 p + 1 ) π 2 m + n π 2 and m is odd and n is even, we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 2 ( z + y j + k π m n | τ ) = ( 1 ) p i m n R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 1 ( m n z | m 2 n τ ) .
    (4.28)
  • When y 1 + y 2 ++ y n = ( 2 p + 1 ) π 2 m + n π 2 and m is odd and n is odd, we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 2 ( z + y j + k π m n | τ ) = ( 1 ) p i m n + 1 R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 2 ( m n z | m 2 n τ ) .
    (4.29)

Here

R 4 , 4 ( y 1 , y 2 ,, y n ;m,n,p;τ)=mn r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) .
(4.30)

Proof Let

zz+ π 2 + m π τ 2

in Theorem 2.1.

By using (1.9), (1.12), and (1.22) we have

j = 1 n ϑ 4 ( z + y j + k π m n + π 2 + m π τ 2 | τ ) = { ( 1 ) k q m 2 n 4 e m n i z e m i ( y 1 + + y n ) j = 1 n ϑ 3 ( z + y j + k π m n | τ ) , m  is even , ( 1 ) k + n q m 2 n 4 e m n i z e m i ( y 1 + + y n ) j = 1 n ϑ 2 ( z + y j + k π m n | τ ) , m  is odd ,
(4.31)
ϑ 3 ( m n z + m 2 n π τ 2 + m n π 2 | m 2 n τ ) = { q m 2 n 4 e m n i z ϑ 2 ( m n z | m 2 n τ ) , m n  is even , i q m 2 n 4 e m n i z ϑ 1 ( m n z | m 2 n τ ) , m n  is odd ,
(4.32)
ϑ 4 ( m n z + m 2 n π τ 2 + m n π 2 | m 2 n τ ) = { i q m 2 n 4 e m n i z ϑ 1 ( m n z | m 2 n τ ) , m n  is even , q m 2 n 4 e m n i z ϑ 2 ( m n z | m 2 n τ ) , m n  is odd .
(4.33)

Substituting (4.31) and (4.32) into (2.1) of Theorem 2.1 and simplifying, we get (4.24), (4.25), and (4.26) of Theorem 4.7.

Substituting (4.31) and (4.33) into (2.2) of Theorem 2.1 and simplifying, we arrive at (4.27), (4.28), and (4.29) of Theorem 4.7. This proof is complete. □

Corollary 4.8 Suppose that m, n are any positive integers; y 1 , y 2 ,, y n are any complex numbers.

  • When y 1 + y 2 ++ y n =0and m is even, we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 3 ( z + y j + k π m n | τ ) = R 4 , 4 ( y 1 , y 2 , , y n ; m , n ; τ ) ϑ 2 ( m n z | m 2 n τ ) .
    (4.34)
  • When y 1 + y 2 ++ y n =0and m is odd and n is even, we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 2 ( z + y j + k π m n | τ ) = R 4 , 4 ( y 1 , y 2 , , y n ; m , n ; τ ) ϑ 2 ( m n z | m 2 n τ ) .
    (4.35)
  • When y 1 + y 2 ++ y n =0and m is odd and n is odd, we have

    k = 0 m n 1 ( 1 ) k + 1 j = 1 n ϑ 2 ( z + y j + k π m n | τ ) = R 4 , 4 ( y 1 , y 2 , , y n ; m , n ; τ ) ϑ 1 ( m n z | m 2 n τ ) .
    (4.36)
  • When y 1 + y 2 ++ y n = π 2 m and m is even, we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 3 ( z + y j + k π m n | τ ) = R 4 , 4 ( y 1 , y 2 , , y n ; m , n ; τ ) ϑ 1 ( m n z | m 2 n τ ) .
    (4.37)
  • When y 1 + y 2 ++ y n = π 2 m and m is odd and n is even, we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 2 ( z + y j + k π m n | τ ) = R 4 , 4 ( y 1 , y 2 , , y n ; m , n ; τ ) ϑ 1 ( m n z | m 2 n τ ) .
    (4.38)
  • When y 1 + y 2 ++ y n =0and m, n are odd, we have

    k = 0 m n 1 ( 1 ) k j = 1 n ϑ 2 ( z + y j + k π m n | τ ) = R 4 , 4 ( y 1 , y 2 , , y n ; m , n ; τ ) ϑ 2 ( m n z | m 2 n τ ) .
    (4.39)

Here

R 4 , 4 ( y 1 , y 2 , , y n ; m , n ; τ ) = m n r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) .
(4.40)

Proof If mn is an even, setting p= m n 2 in y 1 + y 2 ++ y n = p π m + n π 2 , then we have y 1 + y 2 ++ y n =0 and ( 1 ) p = i m n . We directly deduce (4.34) and (4.35) of Corollary 4.8 from Theorem 4.7.

If mn is an even, setting p= m n 2 in y 1 + y 2 ++ y n = p π m + n π 2 + π 2 m , we have y 1 + y 2 ++ y n = π 2 m . We get (4.37), (4.38), and (4.36) of Corollary 4.8 from Theorem 4.7.

If mn is an odd, setting p= m n + 1 2 in y 1 + y 2 ++ y n = p π m + n π 2 + π 2 m and ( 1 ) p = i m n 1 , we have y 1 + y 2 ++ y n =0. We get (4.39) of Corollary 4.8 from Theorem 4.7. The proof is complete. □

Remark 4.9 Equation (4.34) of Corollary 4.8 is just the main result of Zhu (see [[24], p.120, Theorem 1.7]). Of course, we need to make some transformations r j r j m 2 , j=1,2,,n, in (4.40); we readily deduce that

R 4 , 4 ( y 1 , y 2 ,, y n ;m,n;τ)=mn q m 2 n 4 r 1 + + r n = m n 2 r 1 , , r n = q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) ,
(4.41)

which just is the corresponding result of Zhu (see [[24], p.120, (16) of Theorem 1.7]).

Corollary 4.10 Suppose that m, n are any positive integers. We have

k = 0 m n 1 ( 1 ) k ϑ 3 n ( z + k π m n | τ ) = R 4 , 4 ( m , n ; τ ) ϑ 2 ( m n z | m 2 n τ ) , m  is even ,
(4.42)
k = 0 m n 1 ( 1 ) k ϑ 2 n ( z + k π m n | τ ) = R 4 , 4 ( m , n ; τ ) ϑ 2 ( m n z | m 2 n τ ) , m  is odd and  n  is even ,
(4.43)
k = 0 m n 1 ( 1 ) k ϑ 2 n ( z + k π m n | τ ) = R 4 , 4 ( m , n ; τ ) ϑ 1 ( m n z | m 2 n τ ) , m , n  are odd ,
(4.44)
k = 0 m n 1 ( 1 ) k ϑ 3 n ( z + π 2 m n + k π m n | τ ) = R 4 , 4 ( m , n ; τ ) ϑ 1 ( m n z | m 2 n τ ) , m  is even ,
(4.45)
k = 0 m n 1 ( 1 ) k ϑ 2 n ( z + π 2 m n + k π m n | τ ) = R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 1 ( m n z | m 2 n τ ) , m  is odd and  n  is even ,
(4.46)
k = 0 m n 1 ( 1 ) k ϑ 2 n ( z + k π m n | τ ) = R 4 , 4 ( y 1 , y 2 , , y n ; m , n , p ; τ ) ϑ 2 ( m n z | m 2 n τ ) , m , n  are odd ,
(4.47)

where

R 4 , 4 (m,n;τ)=mn r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 .
(4.48)

Proof We set y 1 = y 2 == y n =0 in (4.34), (4.35), and (4.39) of Corollary 4.8. We take y 1 = y 2 == y n = π 2 m n in (4.37) and (4.38) of Corollary 4.8. We deduce Corollary 4.10. □

Remark 4.11 Equation (4.42) of Corollary 4.10 is an extension of Boon’s result. On taking n=1 in (4.42), and noting that R 4 , 4 (m,1;τ)=m, we have

k = 0 m 1 ( 1 ) k ϑ 3 ( z + k π m | τ ) =m ϑ 2 ( m z | m 2 τ ) ,m is even,
(4.49)

which is just Boon’s result (see [[8], p.3440, the second identity of (8)]).

Remark 4.12 Equation (4.45) of Corollary 4.10 is also an extension of Boon’s result. On taking n=1 in (4.45), and noting that R 4 , 4 (m,1;τ)=m, we have

k = 0 m 1 ( 1 ) k ϑ 3 ( z + π 2 m + k π m | τ ) = R 4 , 4 (m,1;τ) ϑ 1 ( m z | m 2 τ ) ,m is even,
(4.50)

which is just Boon’s result (see [[8], p.3440, the first identity of (8)]).

Remark 4.13 Equation (4.42) of Corollary 4.10 is an alternating summation of Ramanujan’s circular summation. Taking m=2 in (4.42), we have

k = 0 2 n 1 ( 1 ) k ϑ 3 n ( z + k π 2 n | τ ) = R 4 , 4 (n;τ) ϑ 2 (2nz|4nτ).
(4.51)

Remark 4.14 We note that Boon’s result (see [[8], p.3440, the first identity of (8)]) has a misprint, we have here corrected this point in our formula (4.50).

Remark 4.15 If we make the transformations zz+ π 2 , zz+ m n π τ 2 , zz+ π 2 + m n π τ 2 , and using the transformation formulas (1.9)-(1.14) for Theorem 3.1, we can also obtain the corresponding alternating circular summation formulas.

5 The inverse formulas for Ramanujan’s circular summation formula

In [15], Liu obtained the following two results.

Lemma 5.1 (see [[15], p.1978, Theorem 1.1])

Suppose that n is a positive integer and f(z|τ) is an entire function of z satisfying the functional equations

f(z|τ)=f(z+π|τ)= q n e 2 n i z f(z+πτ|τ).
(5.1)

Then for any positive integer m, there exists a constant a m ( 1 ) independent of z such that

k = 0 n 1 e 2 m k i π n f ( z + k π n | τ ) =n a m ( 1 ) e 2 m i z ϑ 3 (nz+mπτ|nτ),
(5.2)
f(z|τ)= m = 0 n 1 a m ( 1 ) e 2 m i z ϑ 3 (nz+mπτ|nτ).
(5.3)

Lemma 5.2 (see [[15], p.1978, Theorem 1.2])

Suppose that n is a positive integer and f(z|τ) is an entire function of z satisfying the functional equations

f(z|τ)= ( 1 ) n f(z+π|τ)= q n e 2 n i z f(z+πτ|τ).
(5.4)

Then for any positive integer m, there exists a constant a m ( 2 ) independent of z such that

k = 0 n 1 ( 1 ) k e 2 m k i π n f ( z + k π n | τ ) =n a m ( 2 ) e 2 m i z ϑ 1 (nz+mπτ|nτ),
(5.5)
f(z|τ)= m = 0 n 1 a m ( 2 ) e 2 m i z ϑ 1 (nz+mπτ|nτ).
(5.6)

Liu also found many new identities of the theta functions from Lemma 5.1 and Lemma 5.2.

Below we give some new inverse relations for theta function ϑ 4 (z|τ) by using the results of Liu.

Theorem 5.3 Suppose that m, n are any positive integers, p is any integer; y 1 , y 2 ,, y n are any complex numbers.

  • When y 1 + y 2 ++ y n =pπ+ n π 2 , we have

    k = 0 n 1 e 2 i m k π n j = 1 n ϑ 4 ( z + y j + k π n | τ ) =n a m ( 1 ) e 2 m i z ϑ 3 (nz+mπτ|nτ),
    (5.7)
    j = 1 n ϑ 4 (z+ y j |τ)= m = 0 n 1 a m ( 1 ) e 2 m i z ϑ 3 (nz+mπτ|nτ).
    (5.8)
  • When y 1 + y 2 ++ y n =pπ+ n π 2 + π 2 and n is even, we have

    k = 0 n 1 ( 1 ) k e 2 i m k π n j = 1 n ϑ 4 ( z + y j + k π n | τ ) =n a m ( 2 ) e 2 m i z ϑ 1 (nz+mπτ|nτ),
    (5.9)
    j = 1 n ϑ 4 (z+ y j |τ)= m = 0 n 1 a m ( 2 ) e 2 m i z ϑ 1 (nz+mπτ|nτ).
    (5.10)

Here

a m ( 1 ) = ( 1 ) m r 1 , , r n = r 1 + + r n = m q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) ,
(5.11)
a m ( 2 ) = ( 1 ) m i n + 1 q m n 4 r 1 , , r n = 2 r 1 + + 2 r n = 2 m + n q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) .
(5.12)

Proof Let

f(z|τ)= j = 1 n ϑ 4 (z+ y j |τ).

We easily obtain by using the first and second identities of (1.8) for n=1, respectively:

f(z+π|τ)= j = 1 n ϑ 4 (z+π+ y j |τ)= j = 1 n ϑ 4 (z+ y j |τ)=f(z|τ),
(5.13)
f(z+πτ|τ)= j = 1 n ϑ 4 (z+πτ+ y j |τ)= ( 1 ) n q n e 2 n i z e 2 i ( y 1 + y 2 + + y n ) f(z|τ).
(5.14)
  • When y 1 + y 2 ++ y n =pπ+ n π 2 in (5.14). From (5.13) and (5.14), we have

    f(z|τ)=f(z+π|τ)= q n e 2 n i z f(z+πτ),
    (5.15)

which satisfies condition (5.1). By Lemma 5.1 we obtain (5.7) and (5.8) of Theorem 5.3 at once.

Next we compute the constant a m ( 1 ) independent of z. We apply the definitions (1.3) and (1.4).

Setting

znz+mπτandτnτ

in (1.3), we get

ϑ 3 (nz+mπτ|nτ)= r = q n r 2 e 2 r i ( n z + m π τ ) .
(5.16)

Setting

zz+ y j + k π n

in (1.4), we get

j = 1 n ϑ 4 ( z + y j + k π n | τ ) = r 1 , , r n = ( 1 ) r 1 + + r n q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) e k π i n ( r 1 + + r n ) e 2 i ( r 1 + + r n ) z .
(5.17)

Substituting (5.16) and (5.17) into (5.7), we find that

k = 0 n 1 e 2 i m k π n r 1 , , r n = ( 1 ) r 1 + + r n q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) e k π i n ( r 1 + + r n ) e 2 i ( r 1 + + r n ) z = n a m e 2 m i z r = q n r 2 e 2 r i ( n z + m π τ ) .
(5.18)

Comparing the constants of both sides of (5.18) and noting that y 1 + y 2 ++ y n =pπ+ n π 2 , we obtain (5.11) of Theorem 5.3.

  • When y 1 + y 2 ++ y n =pπ+ n π 2 + π 2 in (5.14) and n is only even. From (5.13) and (5.14), we have

    f(z|τ)= ( 1 ) n f(z+π|τ)= q n e 2 n i z f(z+πτ),
    (5.19)

which satisfies condition (5.4) of Lemma 5.2. From Lemma 5.2 we obtain (5.9) and (5.10) of Theorem 5.3 immediately. In a similar way we can prove (5.12). The proof is complete.

 □

Corollary 5.4 Suppose that m, n are any positive integers; y 1 , y 2 ,, y n are any complex numbers.

  • When y 1 + y 2 ++ y n =0and n is even, we have

    k = 0 n 1 j = 1 n ϑ 4 ( z + y j + k π n | τ ) =n a 0 ( 1 ) ϑ 3 (nz|nτ),
    (5.20)
    j = 1 n ϑ 4 (z+ y j |τ)= m = 0 n 1 a m ( 1 ) e 2 m i z ϑ 3 (nz+mπτ|nτ).
    (5.21)
  • When y 1 + y 2 ++ y n = π 2 and n is even, we have

    k = 0 n 1 ( 1 ) k j = 1 n ϑ 4 ( z + y j + k π n | τ ) =n a 0 ( 2 ) ϑ 1 (nz|nτ),
    (5.22)
    j = 1 n ϑ 4 (z+ y j |τ)= m = 0 n 1 a m ( 2 ) e 2 m i z ϑ 1 (nz+mπτ|nτ).
    (5.23)

Here

a 0 ( 1 ) = r 1 , , r n = r 1 + + r n = 0 q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) ,
(5.24)
a 0 ( 2 ) = i n + 1 q n 4 r 1 , , r n = 2 r 1 + + 2 r n = n q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) ,
(5.25)
a m ( 1 ) = ( 1 ) m r 1 , , r n = r 1 + + r n = m q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) ,
(5.26)
a m ( 2 ) = ( 1 ) m i n + 1 q m n 4 r 1 , , r n = 2 r 1 + + 2 r n = 2 m + n q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) .
(5.27)

Proof If n is an even, setting p= n 2 in y 1 + y 2 ++ y n =pπ+ n π 2 , then we have y 1 + y 2 ++ y n =0. We directly deduce (5.20) (setting m=0) and (5.21) of Corollary 5.4 from Theorem 5.3.

If n is an even, setting p= n 2 in y 1 + y 2 ++ y n =pπ+ n π 2 + π 2 , we have y 1 + y 2 ++ y n = π 2 . We get (5.22) (setting m=0) and (5.23) of Corollary 5.4 from Theorem 5.3. The proof is complete. □

Setting n=2 in (5.20) noting that y 1 + y 2 =0 and a 0 ( 1 ) = ϑ 3 (2 y 1 |2τ), we have

ϑ 3 (z+ y 1 |τ) ϑ 3 (z y 1 |τ)+ ϑ 4 (z+ y 1 |τ) ϑ 4 (z y 1 |τ)=2 ϑ 3 (2 y 1 |2τ) ϑ 3 (2z|2τ).
(5.28)

Setting n=2 in (5.21), noting that y 1 + y 2 =0 and a 0 ( 1 ) = ϑ 3 (2 y 1 |2τ), a 1 ( 1 ) = e 2 i y 1 ϑ 3 (2 y 1 |2τ), we have

ϑ 4 ( z + y 1 | τ ) ϑ 4 ( z y 1 | τ ) = ϑ 3 ( 2 y 1 | 2 τ ) ϑ 3 ( 2 z | 2 τ ) e 2 i ( z y 1 ) ϑ 3 ( 2 y 1 | 2 τ ) ϑ 3 ( 2 z + π τ | 2 τ ) .
(5.29)

Setting n=2 in (5.22) noting that y 1 + y 2 = π 2 and a 0 ( 2 ) = ϑ 3 (2 y 1 |2τ), we have

k = 0 n 1 ( 1 ) k j = 1 n ϑ 4 ( z + y j + k π n | τ ) =n a 0 ( 2 ) ϑ 1 (nz|nτ),
(5.30)
j = 1 n ϑ 4 (z+ y j |τ)= m = 0 n 1 a m ( 2 ) e 2 m i z ϑ 1 (nz+mπτ|nτ),
(5.31)
a m ( 2 ) = ( 1 ) m i n + 1 q m n 4 r 1 , , r n = 2 r 1 + + 2 r n = 2 m + n q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) .
(5.32)

Corollary 5.5 Assume m is any positive integer.

  • When n is even, we have

    k = 0 n 1 ϑ 4 n ( z + k π n | τ ) =n a 0 ( 1 ) ϑ 3 (nz|nτ),
    (5.33)
    ϑ 4 n (z|τ)= m = 0 n 1 a m ( 1 ) e 2 m i z ϑ 3 (nz+mπτ|nτ).
    (5.34)
  • When n is even, we have

    k = 0 n 1 ( 1 ) k ϑ 4 n ( z + π 2 n + k π n | τ ) =n a 0 ( 2 ) ϑ 1 (nz|nτ),
    (5.35)
    ϑ 4 n ( z + π 2 n | τ ) = m = 0 n 1 a m ( 2 ) e 2 m i z ϑ 1 (nz+mπτ|nτ).
    (5.36)

Here

a 0 ( 1 ) = r 1 , , r n = r 1 + + r n = 0 q r 1 2 + + r n 2 ,
(5.37)
a 0 ( 2 ) = i n q n 4 r 1 , , r n = 2 r 1 + + 2 r n = n q r 1 2 + + r n 2 ,
(5.38)
a m ( 1 ) = ( 1 ) m r 1 , , r n = r 1 + + r n = m q r 1 2 + + r n 2 ,
(5.39)
a m ( 2 ) = ( 1 ) m i n q m n 4 e m π i n r 1 , , r n = 2 r 1 + + 2 r n = 2 m + n q r 1 2 + + r n 2 .
(5.40)

Proof We set y 1 = y 2 == y n =0 in (5.20) and (5.21) of Corollary 5.4. We take y 1 = y 2 == y n = π 2 n in (5.22) and (5.31) of Corollary 5.4. We obtain Corollary 5.5. □

6 Some applications

The well-known cubic theta function a(τ) is defined (see [9]) by

a(τ)= r 1 , r 2 = q r 1 2 + r 1 r 2 + r 2 2 .
(6.1)

We define the multiple theta series a( y 1 , y 2 |τ) as

a( y 1 , y 2 |τ)= r 1 , r 2 = q r 1 2 + r 1 r 2 + r 2 2 e 2 i ( r 1 y 1 + r 2 y 2 ) .
(6.2)

Obviously, a(τ)=a(0,0|τ).

We give some applications of Theorem 2.1 below.

Corollary 6.1 Suppose that m, n are any positive integers; y 1 , y 2 ,, y n are any complex numbers.

  • When y 1 + y 2 ++ y n =0, we have

    k = 0 m n 1 j = 1 n ϑ 4 ( z + y j + k π m n | τ ) = { R 4 , 4 ( y 1 , y 2 , , y n ; m , n ; τ ) ϑ 3 ( m n z | m 2 n τ ) , m n  is even , R 4 , 4 ( y 1 , y 2 , , y n ; m , n ; τ ) ϑ 4 ( m n z | m 2 n τ ) , m n  is odd .
    (6.3)
  • When y 1 + y 2 ++ y n = π 2 m and mn is even, we have

    k = 0 m n 1 j = 1 n ϑ 4 ( z + y j + k π m n | τ ) = R 4 , 4 ( y 1 , y 2 ,, y n ;m,n;τ) ϑ 4 ( m n z | m 2 n τ ) ,
    (6.4)

where

R 4 , 4 ( y 1 , y 2 ,, y n ;m,n;τ)=mn r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 e 2 i ( r 1 y 1 + + r n y n ) .
(6.5)

Proof When mn is even, taking p= m n 2 in (2.1); when mn is odd, taking p= m n + 1 2 in (2.2), we easily get (6.3).

When mn is even, taking p= m n 2 in (2.2), we easily obtain (6.4). □

Corollary 6.2 Suppose that m, n are any positive integers, a, b are nonnegative integers and a+b=n; y is any complex numbers.

k = 0 m n 1 ϑ 4 a ( z + b y + k π m n | τ ) ϑ 4 b ( z a y + k π m n | τ ) = { R 4 , 4 ( y ; m , n , a , b ; τ ) ϑ 3 ( m n z | m 2 n τ ) , m n  is even , R 4 , 4 ( y ; m , n , a , b ; τ ) ϑ 4 ( m n z | m 2 n τ ) , m n  is odd .
(6.6)
  • When mn is even, we have

    k = 0 m n 1 ϑ 4 n ( z + π 2 m n + k π m n | τ ) = R 4 , 4 (m,n;τ) ϑ 4 ( m n z | m 2 n τ ) .
    (6.7)

Here

R 4 , 4 (y;m,n,a,b;τ)=mn r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 e 2 n i y ( r 1 + + r a ) ,
(6.8)
R 4 , 4 (m,n;τ)=mn r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 .
(6.9)

Proof When y 1 + y 2 ++ y n =0, taking y 1 = y 2 == y a =by and y a + 1 = y a + 2 == y n =ay, a+b=n in Corollary 6.1, we have (6.6).

When y 1 + y 2 ++ y n = π 2 m , taking y 1 = y 2 == y n = π 2 m n , in Corollary 6.1, we have (6.7). □

Remark 6.3 Let zz+ π 2 in (6.7) and using the transformation formulas (1.8) and (1.12), we have

k = 0 m n 1 ϑ 3 n ( z + π 2 m n + k π m n | τ ) = R 4 , 4 (m,n;τ) ϑ 4 ( m n z | m 2 n τ ) ,
(6.10)

which is a new formula; it is also an extension of Boon’s result.

If we take n=1 in (6.10) and (6.9), we deduce

k = 0 m 1 ϑ 3 ( z + ( 2 k + 1 ) π 2 m | τ ) =m ϑ 4 ( m z | m 2 τ ) ,
(6.11)

which is just Boon’s result (see [[8], p.3440]).

Remark 6.4 We here have corrected a misprint concerning the formulas of ϑ 1 (nz|nτ) and ϑ 4 (nz|nτ) in [[8], p.3440].

Taking y 1 = y 2 == y n =0 in (6.3), we have the following.

Corollary 6.5 Suppose that m, n are any positive integers. We have

k = 0 m n 1 ϑ 4 n ( z + k π m n | τ ) = { R 4 , 4 ( m , n ; τ ) ϑ 3 ( m n z | m 2 n τ ) , m n  is even , R 4 , 4 ( m , n ; τ ) ϑ 4 ( m n z | m 2 n τ ) , m n  is odd .
(6.12)

Here

R 4 , 4 (m,n;τ)=mn r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 .
(6.13)

Taking n=1 in (6.12), we deduce

k = 0 m 1 ϑ 4 ( z + k π m | τ ) = { m ϑ 3 ( m z | m 2 τ ) , m  is even , m ϑ 4 ( m z | m 2 τ ) , m  is odd ,
(6.14)

which is just Boon’s result (see [[8], p.3440, Eq. (7)]), with zz+ π 2 and the transformation equation (1.7) and (1.12).

Taking n=2 in (6.12) and noting that

R 4 , 4 ( 1 ) (m;τ)=2m r 1 + r 2 = 0 r 1 , r 2 = q r 1 2 + r 2 2 =2m ϑ 3 (0|2τ),
(6.15)

we have the following.

Corollary 6.6 Suppose that m is any positive integer; we have

k = 0 2 m 1 ϑ 4 2 ( z + k π 2 m | τ ) =2m ϑ 3 (0|2τ) ϑ 3 ( 2 m z | 2 m 2 τ ) .
(6.16)

The above formula is just Boon’s result [[8], p.3440, Eq. (10)], with zz+ π 2 and the transformation equation (1.7) and (1.12).

Taking n=3 in (6.12) and noting that

R 4 , 4 (m;τ)=3m r 1 + r 2 + r 3 = 0 r 1 , r 2 , r 3 = q r 1 2 + r 2 2 + r 3 2 =3ma(2τ),
(6.17)

we have the following.

Corollary 6.7 Suppose that m is a positive integer.

k = 0 3 m 1 ϑ 4 3 ( z + k π 3 m | τ ) = { 3 m a ( 2 τ ) ϑ 3 ( 3 m z | 3 m 2 τ ) , m  is even , 3 m a ( 2 τ ) ϑ 4 ( 3 m z | 3 m 2 τ ) , m  is odd .
(6.18)

We set m=1,

ϑ 4 3 ( z | τ ) + ϑ 4 3 ( z + π 3 | τ ) + ϑ 4 3 ( z π 3 | τ ) =3a(2τ) ϑ 4 (3z|3τ).
(6.19)

Taking m=1 in (6.12), we have

k = 0 n 1 ϑ 4 n ( z + k π n | τ ) = { R 4 , 4 ( n ; τ ) ϑ 3 ( n z | n τ ) , n  is even , R 4 , 4 ( n ; τ ) ϑ 4 ( n z | n τ ) , n  is odd .
(6.20)

Here

R 4 , 4 (n;τ)=n r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 .
(6.21)

The above formula is just Ramanujan’s circular summation formula (see [[6], p.632, Theorem 3.1]), with zz+ π 2 and the transformation equation (1.7) and (1.12).

We set m=1 in (6.10),

k = 0 n 1 ϑ 3 n ( z + π 2 n + k π n | τ ) = R 4 , 4 (n;τ) ϑ 4 (nz|nτ),
(6.22)

where

R 4 , 4 (n;τ)=n r 1 + + r n = 0 r 1 , , r n = q r 1 2 + + r n 2 .
(6.23)

We set n=2 in (6.10),

ϑ 3 2 ( z + π 4 | τ ) + ϑ 4 2 ( z + π 4 | τ ) =2 ϑ 3 (0|2τ) ϑ 4 (2z|2τ).
(6.24)

We set n=3 in (6.10),

ϑ 3 3 ( z + π 6 | τ ) + ϑ 3 3 ( z π 6 | τ ) + ϑ 4 3 ( z | τ ) =3a(2τ) ϑ 4 (3z|3τ).
(6.25)

Comparing (6.19) and (6.25), we find that

ϑ 3 3 ( z + π 6 | τ ) + ϑ 3 3 ( z π 6 | τ ) = ϑ 4 3 ( z + π 3 | τ ) + ϑ 4 3 ( z π 3 | τ ) .
(6.26)

Taking n=3 in (6.3) and noting that y 1 + y 2 + y 3 =0, we obtain

R 4 , 4 ( y 1 , y 2 ,;τ)=3ma( y 1 y 3 , y 2 y 3 |2τ).

Hence, we have

k = 0 3 m 1 ϑ 4 ( z + y 1 + k π 3 m | τ ) ϑ 4 ( z + y 2 + k π 3 m | τ ) ϑ 4 ( z + y 3 + k π 3 m | τ ) = { 3 m a ( y 1 y 3 , y 2 y 3 | 2 τ ) ϑ 3 ( 3 m z | 3 m 2 τ ) , m  is even , 3 m a ( y 1 y 3 , y 2 y 3 | 2 τ ) ϑ 4 ( 3 m z | 3 m 2 τ ) , m  is odd ,
(6.27)

where a( y 1 , y 2 |2τ) is defined by (6.2).

Taking m=2 in the first equation of (6.27), we get

ϑ 3 ( z + y 1 | τ ) ϑ 3 ( z + y 2 | τ ) ϑ 3 ( z + y 3 | τ ) + ϑ 4 ( z + y 1 | τ ) ϑ 4 ( z + y 2 | τ ) ϑ 4 ( z + y 3 | τ ) + ϑ 4 ( z + y 1 + π 6 | τ ) ϑ 4 ( z + y 2 + π 6 | τ ) ϑ 4 ( z + y 3 + π 6 | τ ) + ϑ 4 ( z + y 1 π 6 | τ ) ϑ 4 ( z + y 2 π 6 | τ ) ϑ 4 ( z + y 3 π 6 | τ ) + ϑ 4 ( z + y 1 + π 3 | τ ) ϑ 4 ( z + y 2 + π 3 | τ ) ϑ 4 ( z + y 3 + π 3 | τ ) + ϑ 4 ( z + y 1 π 3 | τ ) ϑ 4 ( z + y 2 π 3 | τ ) ϑ 4 ( z + y 3 π 3 | τ ) = 6 a ( y 1 y 3 , y 2 y 3 | 2 τ ) ϑ 3 ( 6 z | 12 τ ) .
(6.28)

Setting y 1 = y 2 = y 3 =0 in (6.28), we get

ϑ 3 3 ( z | τ ) + ϑ 4 3 ( z | τ ) + ϑ 4 3 ( z + π 6 | τ ) + ϑ 4 3 ( z π 6 | τ ) + ϑ 4 3 ( z + π 3 | τ ) + ϑ 4 3 ( z π 3 | τ ) = 6 a ( 2 τ ) ϑ 3 ( 6 z | 12 τ ) .
(6.29)

Taking m=1 in the second equation of (6.27), we get

ϑ 4 ( z + y 1 | τ ) ϑ 4 ( z + y 2 | τ ) ϑ 4 ( z + y 3 | τ ) + ϑ 4 ( z + y 1 + π 3 | τ ) ϑ 4 ( z + y 2 + π 3 | τ ) ϑ 4 ( z + y 3 + π 3 | τ ) + ϑ 4 ( z + y 1 π 3 | τ ) ϑ 4 ( z + y 2 π 3 | τ ) ϑ 4 ( z + y 3 π 3 | τ ) = 3 a ( y 1 y 3 , y 2 y 3 | 2 τ ) ϑ 4 ( 3 z | 3 τ ) .
(6.30)

Setting y 1 = y 2 = y 3 =0 in (6.28), we get again

ϑ 3 3 ( z + π 6 | τ ) + ϑ 3 3 ( z π 6 | τ ) + ϑ 4 3 (z|τ)=3a(2τ) ϑ 4 (3z|3τ).
(6.31)

Taking n=3 in (6.4) and noting that y 1 + y 2 + y 3 = π 2 m , we obtain

R 4 , 4 ( y 1 , y 2 ,;τ)=3ma( y 1 y 3 , y 2 y 3 |2τ).

Hence, we have

k = 0 3 m 1 ϑ 4 ( z + y 1 + k π 3 m | τ ) ϑ 4 ( z + y 2 + k π 3 m | τ ) ϑ 4 ( z + y 3 + k π 3 m | τ ) = 3 m a ( y 1 y 3 , y 2 y 3 | 2 τ ) ϑ 4 ( 3 m z | 3 m 2 τ ) , m  is even .
(6.32)

Taking m=2 in (6.32) and noting that y 1 + y 2 + y 3 = π 4 , we get

ϑ 3 ( z + y 1 | τ ) ϑ 3 ( z + y 2 | τ ) ϑ 3 ( z + y 3 | τ ) + ϑ 4 ( z + y 1 | τ ) ϑ 4 ( z + y 2 | τ ) ϑ 4 ( z + y 3 | τ ) + ϑ 4 ( z + y 1 + π 6 | τ ) ϑ 4 ( z + y 2 + π 6 | τ ) ϑ 4 ( z + y 3 + π 6 | τ ) + ϑ 4 ( z + y 1 π 6 | τ ) ϑ 4 ( z + y 2 π 6 | τ ) ϑ 4 ( z + y 3 π 6 | τ ) + ϑ 4 ( z + y 1 + π 3 | τ ) ϑ 4 ( z + y 2 + π 3 | τ ) ϑ 4 ( z + y 3 + π 3 | τ ) + ϑ 4 ( z + y 1 π 3 | τ ) ϑ 4 ( z + y 2 π 3 | τ ) ϑ 4 ( z + y 3 π 3 | τ ) = 6 a ( y 1 y 3 , y 2 y 3 | 2 τ ) ϑ 4 ( 6 z | 12 τ ) .
(6.33)

Setting y 1 = y 2 = y 3 = π 12 in (6.33), we get the following new identity for the theta function ϑ 4 (z|τ):

ϑ 3 3 ( z + π 12 | τ ) + ϑ 4 3 ( z + π 12 | τ ) + ϑ 4 3 ( z π 12 | τ ) + ϑ 4 3 ( z + π 4 | τ ) + ϑ 4 3 ( z π 4 | τ ) + ϑ 4 3 ( z + 5 π 12 | τ ) = 6 a ( 2 τ ) ϑ 4 ( 6 z | 12 τ ) .
(6.34)

References

  1. Bellman R: A Brief Introduction to Theta Functions. Holt, Rinehart & Winston, New York; 1961.

    Google Scholar 

  2. Chandrasekharan K: Elliptic Functions. Springer, Berlin; 1985.

    Book  Google Scholar 

  3. Whittaker ET, Watson GN: A Course of Modern Analysis. 4th edition. Cambridge University Press, Cambridge; 1927. Reprinted (1963)

    Google Scholar 

  4. Ramanujan S: The Lost Notebook and Other Unpublished Papers. Narosa, New Delhi; 1988.

    Google Scholar 

  5. Andrews GE, Berndt BC: Ramanujan’s Lost Notebook. Part III. Springer, New York; 2012.

    Book  Google Scholar 

  6. Chan HH, Liu Z-G, Ng ST: Circular summation of theta functions in Ramanujan’s lost notebook. J. Math. Anal. Appl. 2006, 316: 628-641. 10.1016/j.jmaa.2005.05.015

    Article  MathSciNet  Google Scholar 

  7. Ahlgren S: The sixth, eighth, ninth and tenth powers of Ramanujan theta function. Proc. Am. Math. Soc. 2000, 128: 1333-1338. 10.1090/S0002-9939-99-05181-3

    Article  MathSciNet  Google Scholar 

  8. Boon M, Glasser ML, Zak J, Zucker IJ: Additive decompositions of ϑ -functions of multiple arguments. J. Phys. A, Math. Gen. 1982, 15: 3439-3440. 10.1088/0305-4470/15/11/020

    Article  MathSciNet  Google Scholar 

  9. Borwein JM, Borwein PB: A cubic counterpart of Jacobi’s identity and the AGM. Trans. Am. Math. Soc. 1991, 323: 691-701.

    Google Scholar 

  10. Cai Y, Chen S, Luo Q-M: Some circular summation formulas for the theta functions. Bound. Value Probl. 2013., 2013: Article ID 59

    Google Scholar 

  11. Cai Y, Bi Y-Q, Luo Q-M: Some new circular summation formulas of theta functions. Integral Transforms Spec. Funct. 2014, 25: 497-507. 10.1080/10652469.2014.883388

    Article  MathSciNet  Google Scholar 

  12. Chan SH, Liu Z-G: On a new circular summation of theta functions. J. Number Theory 2010, 130: 1190-1196. 10.1016/j.jnt.2009.09.011

    Article  MathSciNet  Google Scholar 

  13. Chua KS:The root lattice A n and Ramanujan’s circular summation of theta functions. Proc. Am. Math. Soc. 2002, 130: 1-8. 10.1090/S0002-9939-01-06080-4

    Article  MathSciNet  Google Scholar 

  14. Chua KS: Circular summation of the 13th powers of Ramanujan’s theta function. Ramanujan J. 2001, 5: 353-354. 10.1023/A:1013935519780

    Article  MathSciNet  Google Scholar 

  15. Liu Z-G: Some inverse relations and theta function identities. Int. J. Number Theory 2012, 8: 1977-2002. 10.1142/S1793042112501126

    Article  MathSciNet  Google Scholar 

  16. Murayama T: On an identity of theta functions obtained from weight enumerators of linear codes. Proc. Jpn. Acad., Ser. A, Math. Sci. 1998, 74: 98-100. 10.3792/pjaa.74.98

    Article  MathSciNet  Google Scholar 

  17. Montaldi E, Zucchelli G:Additive decomposition for the product of two ϑ 3 functions and modular equations. J. Math. Phys. 1989, 30: 2012-2015. 10.1063/1.528238

    Article  MathSciNet  Google Scholar 

  18. Ono K: On the circular summation of the eleventh powers of Ramanujan’s theta function. J. Number Theory 1999, 76: 62-65. 10.1006/jnth.1998.2354

    Article  MathSciNet  Google Scholar 

  19. Rangachari SS: On a result of Ramanujan on theta functions. J. Number Theory 1994, 48: 364-372. 10.1006/jnth.1994.1072

    Article  MathSciNet  Google Scholar 

  20. Shen LC: On the additive formula of the theta functions and a collection of Lambert series pertaining to the modular equations of degree 5. Trans. Am. Math. Soc. 1994, 345: 323-345.

    Google Scholar 

  21. Son SH: Circular summation of theta functions in Ramanujan’s lost notebook. Ramanujan J. 2004, 8: 235-272.

    Article  MathSciNet  Google Scholar 

  22. Xu P: An elementary proof of Ramanujan’s circular summation formula and its generalizations. Ramanujan J. 2012, 27: 409-417. 10.1007/s11139-011-9364-4

    Article  MathSciNet  Google Scholar 

  23. Zeng X-F: A generalized circular summation of theta function and its application. J. Math. Anal. Appl. 2009, 356: 698-703. 10.1016/j.jmaa.2009.03.047

    Article  MathSciNet  Google Scholar 

  24. Zhu J-M: An alternate circular summation formula of theta functions and its applications. Appl. Anal. Discrete Math. 2012, 6: 114-125. 10.2298/AADM120204004Z

    Article  MathSciNet  Google Scholar 

  25. Zhu J-M: A note on a generalized circular summation formula of theta functions. J. Number Theory 2012, 132: 1164-1169. 10.1016/j.jnt.2011.12.016

    Article  MathSciNet  Google Scholar 

Download references

Acknowledgements

The authors express their sincere gratitude to the referee for many valuable comments and suggestions. The present investigation was supported by Natural Science Foundation Project of Chongqing, China, under Grant CSTC2011JJA00024, Research Project of Science and Technology of Chongqing Education Commission, China under Grant KJ120625, Fund of Chongqing Normal University, China, under Grant 10XLR017 and 2011XLZ07.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Qiu-Ming Luo.

Additional information

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

All authors contributed equally in writing this paper, and read and approved the manuscript.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0), which permits use, duplication, adaptation, distribution, and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhou, Y., Luo, QM. Circular summation formulas for theta function ϑ 4 (z|τ). Adv Differ Equ 2014, 243 (2014). https://doi.org/10.1186/1687-1847-2014-243

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/1687-1847-2014-243

Keywords