1 Introduction

Throughout this paper we take q= e i π τ , where (τ)>0. The classical Jacobi theta functions θ i (z|τ), i=1,2,3,4, are defined as follows:

(1.1)
(1.2)
(1.3)
(1.4)

Recently, Chan, Liu and Ng [1] proved Ramanujan’s circular summation formulas and derived identities similar to Ramanujan’s summation formula and connected these identities to Jacobi’s elliptic functions.

Subsequently, Zeng [2] gave a generalized circular summation of the theta function θ 3 (z|τ) as follows:

(1.5)

where

C 33 (a,b;y,τ)=kn m 1 + + m a + n 1 + + n b = 0 m 1 , , m a , n 1 , , n b = q m 1 2 + + m a 2 + n 1 2 + + n b 2 e 2 k ( m 1 + + m a ) i y .

A special case of formula (1.5) yields the following result (see [[1], Theorem 3.1]):

s = 0 k n 1 θ 3 k ( z k n + π s k n | τ k n 2 ) = C 33 ( k , 0 ; 0 , τ k n 2 ) θ 3 (z|τ),
(1.6)

where

C 33 (k,0;0,τ)=kn m 1 + + m k = 0 m 1 , , m k = q m 1 2 + + m k 2 .
(1.7)

Upon a, b, n and k are any positive integer with k=a+b.

More recently, Liu further obtained the general formulas for theta functions (see [3]), but from one main result, Theorem 1 of Liu, we do not deduce our results in the present paper. Many people research the circular summation formulas of theta functions and find more interesting formulas (see, for details, [415]).

In the present paper, we obtain analogues and uniform formulas for theta functions θ 1 (z|τ), a θ 2 (z|τ), θ 3 (z|τ) and θ 4 (z|τ). We now state our result as follows.

Theorem 1 For any positive integer k, n, a and b with k=a+b, α=1,2, β=3,4.

  • For a, b even, we have

    (1.8)
  • For a even, n and b odd, we have

    (1.9)
    (1.10)

where

C α β ( a , b ; y , τ ) = k n i a α q a 4 e a b i y m 1 , , m a , n 1 , , n b = 2 ( m 1 + + m a + n 1 + + n b ) + a = 0 ( 1 ) α ( m 1 + + m a ) + ( β + 1 ) ( n 1 + + n b ) × q m 1 2 + + m a 2 + n 1 2 + + n b 2 + m 1 + + m a e ( 2 k ( m 1 + + m a ) + a 2 ) i y .
(1.11)

2 Proof of Theorem 1

From Jacobi’s theta functions (1.1)-(1.4), we have the following properties respectively:

(2.1)
(2.2)
(2.3)
(2.4)

From (2.1)-(2.4), by using the induction, we easily obtain

(2.5)
(2.6)
(2.7)
(2.8)

Let

f(z)= s = 0 k n 1 θ α a ( z k n + y a + π s k n | τ k n 2 ) θ β b ( z k n y b + π s k n | τ k n 2 ) .
(2.9)

Case 1. When α=1, β=3.

The function f(z) becomes of the following form:

f(z)= s = 0 k n 1 θ 1 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 ) .
(2.10)

From (2.10) we easily obtain

(2.11)
(2.12)

Comparing (2.11) and (2.12), when a is even, we get

f(z+π)=f(z).
(2.13)

By (2.5) and (2.7), and noting that a+b=k, we obtain

f ( z + π τ ) = s = 0 k n 1 θ 1 a ( z k n + y a + π s k n + n π τ k n 2 | τ k n 2 ) θ 3 b ( z k n y b + π s k n + n π τ k n 2 | τ k n 2 ) = ( 1 ) n a q 1 e 2 i z s = 0 k n 1 θ 1 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 ) .
(2.14)

Obviously, when a is even, we have

f(z+πτ)= q 1 e 2 i z f(z).
(2.15)

We construct the function f ( z ) θ 3 ( z | τ ) . By (2.13) and (2.15), we find that the function f ( z ) θ 3 ( z | τ ) is an elliptic function with double periods π and πτ and only has 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, this constant is denoted by C 13 (a,b;y,τ), i.e.,

f ( z ) θ 3 ( z | τ ) = C 13 (a,b;y,τ),

we have

f(z)= C 13 (a,b;y,τ) θ 3 (z|τ).
(2.16)

By (1.3), (2.10) and (2.16), we have

(2.17)

By (1.1) and (1.3), we obtain

(2.18)

By equating the constant term of both sides of (2.18), we obtain

(2.19)

Clearly,

C 13 (a,b;y,τ)= C 13 ( a , b ; y a b , τ k n 2 ) ,
(2.20)

where

(2.21)

In the same manner as in Case 1, we can obtain Case 2 below.

Case 2. When α=2, β=3.

The function f(z) becomes of the following form:

f(z)= s = 0 k n 1 θ 2 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 ) .
(2.22)

From (2.22) we easily obtain

(2.23)
(2.24)

Comparing (2.23) and (2.24), when a is even, we get

f(z+π)=f(z).
(2.25)

By (2.6) and (2.7), and noting that a+b=k, we obtain

f ( z + π τ ) = s = 0 k n 1 θ 2 a ( z k n + y a + π s k n + n π τ k n 2 | τ k n 2 ) θ 3 b ( z k n y b + π s k n + n π τ k n 2 | τ k n 2 ) = q 1 e 2 i z s = 0 k n 1 θ 2 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 ) .
(2.26)

Obviously, we have

f(z+πτ)= q 1 e 2 i z f(z).
(2.27)

We construct the function f ( z ) θ 3 ( z | τ ) . By (2.25) and (2.27), we find that the function f ( z ) θ 3 ( z | τ ) is an elliptic function with double periods π and πτ and only has 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, this constant is denoted by C 23 (a,b;y,τ), i.e.,

f ( z ) θ 3 ( z | τ ) = C 23 (a,b;y,τ),

we have

f(z)= C 23 (a,b;y,τ) θ 3 (z|τ).
(2.28)

By (1.3), (2.22) and (2.28), we have

(2.29)

By (1.2) and (1.3), we obtain

(2.30)

By equating the constant term of both sides of (2.30), we obtain

C 23 ( a , b ; y , τ ) = k n q a 4 k n 2 e i y m 1 , , m a , n 1 , , n b = 2 ( m 1 + + m a + n 1 + + n b ) + a = 0 q m 1 2 + + m a 2 + n 1 2 + + n b 2 + m 1 + + m a k n 2 × e 2 i y a b [ k ( m 1 + + m a ) + a 2 2 ] .
(2.31)

Clearly,

C 23 (a,b;y,τ)= C 23 ( a , b ; y a b , τ k n 2 ) ,
(2.32)

where

C 23 ( a , b ; y , τ ) = k n q a 4 e a b i y m 1 , , m a , n 1 , , n b = 2 ( m 1 + + m a + n 1 + + n b ) + a = 0 q m 1 2 + + m a 2 + n 1 2 + + n b 2 + m 1 + + m a × e ( 2 k ( m 1 + + m a ) + a 2 ) i y .
(2.33)

Case 3. When α=1, β=4.

The function f(z) becomes of the following form:

f(z)= s = 0 k n 1 θ 1 a ( z k n + y a + π s k n | τ k n 2 ) θ 4 b ( z k n y b + π s k n | τ k n 2 ) .
(2.34)

From (2.34) we easily obtain

(2.35)
(2.36)

Comparing (2.35) and (2.36), when a is even, we have

f(z+π)=f(z).
(2.37)

By (2.5) and (2.8), and noting that a+b=k, we obtain

f ( z + π τ ) = s = 0 k n 1 θ 1 a ( z k n + y a + π s k n + n π τ k n 2 | τ k n 2 ) θ 4 b ( z k n y b + π s k n + n π τ k n 2 | τ k n 2 ) = ( 1 ) k n q 1 e 2 i z s = 0 k n 1 θ 1 a ( z k n + y a + π s k n | τ k n 2 ) θ 4 b ( z k n y b + π s k n | τ k n 2 ) .
(2.38)
  • When a and b are even, then kn is also even, we have

    f(z+πτ)= q 1 e 2 i z f(z).
    (2.39)

We construct the function f ( z ) θ 3 ( z | τ ) , by (2.37) and (2.39), we find that the function f ( z ) θ 3 ( z | τ ) is an elliptic function with double periods π and πτ and only has 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, this constant is denoted by C 14 (a,b;y,τ), i.e.,

f ( z ) θ 3 ( z | τ ) = C 14 (a,b;y,τ),

we have

f(z)= C 14 (a,b;y,τ) θ 3 (z|τ).
(2.40)

By (1.3), (2.34) and (2.40), we have

(2.41)

By (1.1) and (1.4), we obtain

(2.42)

By equating the constant term of both sides of (2.42), we obtain

C 14 ( a , b ; y , τ ) = k n i a q a 4 k n 2 e i y m 1 , , m a , n 1 , , n b = 2 ( m 1 + + m a + n 1 + + n b ) + a = 0 ( 1 ) m 1 + + n b × q m 1 2 + + m a 2 + n 1 2 + + n b 2 + m 1 + + m a k n 2 e 2 i y a b [ k ( m 1 + + m a ) + a 2 2 ] .
(2.43)
  • When a is even, n and b are odd, then kn is also odd, we have

    f(z+πτ)= q 1 e 2 i z f(z).
    (2.44)

We construct the function f ( z ) θ 4 ( z | τ ) . By (2.37) and (2.44), we find that the function f ( z ) θ 4 ( z | τ ) is an elliptic function with double periods π and πτ and only has a simple pole at z= π 2 + π τ 2 in the period parallelogram. Hence the function f ( z ) θ 4 ( z | τ ) is a constant, say, this constant is denoted by C 14 (a,b;y,τ), i.e.,

f ( z ) θ 4 ( z | τ ) = C 14 (a,b;y,τ),

we have

f(z)= C 14 (a,b;y,τ) θ 4 (z|τ).
(2.45)

By (1.4), (2.34) and (2.45), we have

(2.46)

By (1.1) and (1.4), we obtain

(2.47)

By equating the constant term of both sides of (2.47), we obtain

C 14 ( a , b ; y , τ ) = k n i a q a 4 k n 2 e i y m 1 , , m a , n 1 , , n b = 2 ( m 1 + + m a + n 1 + + n b ) + a = 0 ( 1 ) m 1 + + n b × q m 1 2 + + m a 2 + n 1 2 + + n b 2 + m 1 + + m a k n 2 e 2 i y a b [ k ( m 1 + + m a ) + a 2 2 ] .
(2.48)

Clearly, in (2.43) and (2.48), we have

C 14 (a,b;y,τ)= C 14 ( a , b ; y a b , τ k n 2 ) ,

where

C 14 ( a , b ; y , τ ) = k n i a q a 4 e a b i y m 1 , , m a , n 1 , , n b = 2 ( m 1 + + m a + n 1 + + n b ) + a = 0 ( 1 ) m 1 + + n b × q m 1 2 + + m a 2 + n 1 2 + + n b 2 + m 1 + + m a e ( 2 k ( m 1 + + m a ) + a 2 ) i y .
(2.49)

In the same manner as in Case 3, we can obtain Case 4 below.

Case 4. When α=2, β=4.

The function f(z) becomes of the following form:

f(z)= s = 0 k n 1 θ 2 a ( z k n + y a + π s k n | τ k n 2 ) θ 4 b ( z k n y b + π s k n | τ k n 2 ) .
(2.50)
  • When a and b are even, we have

    (2.51)
    (2.52)
  • When a is even, n and b are odd, we have

    f(z)= C 24 (a,b;y,τ) θ 4 (z|τ),
    (2.53)
C 24 ( a , b ; y , τ ) = k n q a 4 k n 2 e i y m 1 , , m a , n 1 , , n b = 2 ( m 1 + + m a + n 1 + + n b ) + a = 0 ( 1 ) n 1 + + n b × q m 1 2 + + m a 2 + n 1 2 + + n b 2 + m 1 + + m a k n 2 e 2 i y a b [ k ( m 1 + + m a ) + a 2 2 ] .
(2.54)

Clearly, in (2.52) and (2.54), we have

C 24 (a,b;y,τ)= C 24 ( a , b ; y a b , τ k n 2 ) ,

where

C 24 ( a , b ; y , τ ) = k n q a 4 e a b i y m 1 , , m a , n 1 , , n b = 2 ( m 1 + + m a + n 1 + + n b ) + a = 0 ( 1 ) n 1 + + n b q m 1 2 + + m a 2 + n 1 2 + + n b 2 + m 1 + + m a × e ( 2 k ( m 1 + + m a ) + a 2 ) i y .
(2.55)

Therefore we complete the proof of Theorem 1.

3 Some special cases of Theorem 1

In this section we give some special cases of Theorem 1 and obtain some interesting identities of theta functions.

Corollary 1 For any positive integer n, we have

(3.1)
(3.2)
(3.3)
(3.4)

where C 13 (2,2;y,τ) and C 24 (2,2;y,τ) are defined by (2.21) and (2.55), respectively.

Proof Taking a=b=2 and α=1, β=3 in (1.11), we have

(3.5)

Taking a=b=2 and α=2, β=4 in (1.11), we have

C 24 ( 2 , 2 ; y , τ ) = 4 n q 1 2 m 1 , m 2 , n 1 , n 2 = m 1 + m 2 + n 1 + n 2 + 1 = 0 ( 1 ) n 1 + n 2 q m 1 2 + m 2 2 + n 1 2 + n 2 2 + m 1 + m 2 e 8 i y ( m 1 + m 2 ) e 5 i y = 4 n e 3 i y θ 1 2 ( 4 y | τ ) θ 3 2 ( 0 | τ ) .
(3.6)

Obviously, we find that C 13 (2,2;y,τ)= C 24 (2,2;y,τ)=4n e 3 i y θ 1 2 (4y|τ) θ 3 2 (0|τ). □

Taking n=1 and letting z4z, y2y, τ4τ in Corollary 1, we get the following identities for theta functions:

(3.7)
(3.8)
(3.9)

Further, taking y=0 in the above identities, we obtain the following additive formulas of the theta function θ 3 (z|τ):

θ 3 ( 4 z | 4 τ ) = 1 4 θ 1 2 θ 3 2 [ θ 1 2 ( z | τ ) θ 3 2 ( z | τ ) + θ 1 2 ( z + π 4 | τ ) θ 3 2 ( z + π 4 | τ ) + θ 1 2 ( z + π 2 | τ ) θ 3 2 ( z + π 2 | τ ) + θ 1 2 ( z + 3 π 4 | τ ) θ 3 2 ( z + 3 π 4 | τ ) ]
(3.10)
= 1 4 θ 1 2 θ 3 2 [ θ 2 2 ( z | τ ) θ 4 2 ( z | τ ) + θ 2 2 ( z + π 4 | τ ) θ 4 2 ( z + π 4 | τ ) + θ 2 2 ( z + π 2 | τ ) θ 4 2 ( z + π 2 | τ ) + θ 2 2 ( z + 3 π 4 | τ ) θ 4 2 ( z + 3 π 4 | τ ) ] ,
(3.11)

where θ 1 = θ 1 (0|τ), θ 3 = θ 3 (0|τ).

Similarly, we have the following.

Corollary 2 For any positive integer n, we have

(3.12)
(3.13)
(3.14)
(3.15)

where C 14 (2,2;y,τ)= C 23 (2,2;y,τ)=4n e 3 i y θ 2 2 (4y|τ) θ 3 2 (0|τ) are defined by (2.49) and (2.33), respectively.

Taking n=1 and letting z4z, y2y, τ4τ in Corollary 2, we get the following identities of theta functions:

(3.16)
(3.17)
(3.18)

Further, taking y=0 in the above identities, we obtain other additive formulas for the theta function θ 3 (z|τ) as follows:

θ 3 ( 4 z | 4 τ ) = 1 4 θ 2 2 θ 3 2 [ θ 1 2 ( z | τ ) θ 4 2 ( z | τ ) + θ 1 2 ( z + π 4 | τ ) θ 4 2 ( z + π 4 | τ ) + θ 1 2 ( z + π 2 | τ ) θ 4 2 ( z + π 2 | τ ) + θ 1 2 ( z + 3 π 4 | τ ) θ 4 2 ( z + 3 π 4 | τ ) ]
(3.19)
= 1 4 θ 2 2 θ 3 2 [ θ 2 2 ( z | τ ) θ 3 2 ( z | τ ) + θ 2 2 ( z + π 4 | τ ) θ 3 2 ( z + π 4 | τ ) + θ 2 2 ( z + π 2 | τ ) θ 3 2 ( z + π 2 | τ ) + θ 2 2 ( z + 3 π 4 | τ ) θ 3 2 ( z + 3 π 4 | τ ) ] ,
(3.20)

where θ 2 = θ 2 (0|τ), θ 3 = θ 3 (0|τ).

Taking a=2, b=1 in (1.9), (1.10) and (1.11), we have the following.

Corollary 3 For n odd, we have

(3.21)
(3.22)
(3.23)
(3.24)

where C 14 (2,2;y,τ) and C 24 (2,2;y,τ) are defined by (2.49) and (2.55), respectively.

Proof Taking a=2, b=1 and α=(1,2), β=4 in (1.11), we have

(3.25)
(3.26)

 □

Taking n=1 and letting z3z, y2y, τ3τ in Corollary 3, we get the following identities for theta functions:

(3.27)

and

(3.28)

Further taking y=0 in the above identities, we obtain the following additive formulas for the theta function θ 4 (z|τ) as follows:

θ 4 ( 3 z | 3 τ ) = 1 3 θ 3 θ 2 2 [ θ 1 2 ( z | τ ) θ 4 ( z | τ ) + θ 1 2 ( z + π 3 | τ ) θ 4 ( z + π 3 | τ ) + θ 1 2 ( z + 2 π 3 | τ ) θ 4 ( z + 2 π 3 | τ ) ]
(3.29)

and

θ 4 ( 3 z | 3 τ ) = 1 3 θ 3 θ 1 2 [ θ 2 2 ( z | τ ) θ 4 ( z | τ ) + θ 2 2 ( z + π 3 | τ ) θ 4 ( z + π 3 | τ ) + θ 2 2 ( z + 2 π 3 | τ ) θ 4 ( z + 2 π 3 | τ ) ] ,
(3.30)

where θ 1 = θ 1 (0|τ), θ 2 = θ 2 (0|τ), θ 3 = θ 3 (0|τ).

Corollary 4 When n is odd, we have

s = 0 k n 1 ( 1 ) s θ 2 k ( z + π s k n | τ ) = C 33 (k,0;0,τ) θ 2 ( k n z | k n 2 τ ) ,
(3.31)

where C 33 (k,0;0,τ) is defined by (1.7).

Proof Replacing z by knz+knπ+ k n 2 π τ 2 and τ by k n 2 τ in (1.6), we obtain

(3.32)

Substituting n by 2n+1 in the left-hand side of (3.32), we get (3.31). □