Some new mock theta functions

https://doi.org/10.1016/j.aam.2021.102267Get rights and content

Abstract

Mock theta functions can be represented by Eulerian forms, Hecke-type double sums, Appell–Lerch sums, and Fourier coefficients of meromorphic Jacobi forms. In view of some q-series identities, we establish three two-parameter mock theta functions, and express them in terms of Appell-Lerch sums. In addition, we find three mock theta functions in Eulerian forms. Then in light of their Hecke-type double sums, we provide the Hecke-type double sums for some third order mock theta functions, and establish the relations between these functions and some classical mock theta functions.

Introduction

Mock theta functions were first introduced by Ramanujan [30] in his last letter to Hardy. He found that these functions have certain asymptotic properties as q approaches a root of unity, which are similar to theta functions, but they are not really theta functions. Historically, mock theta functions can be represented by Eulerian forms, Hecke-type double sums, Appell–Lerch sums, and Fourier coefficients of meromorphic Jacobi forms. In these years, it has received a great deal of attention to seek new mock theta functions and find different forms for mock theta functions. In this paper, we find the following two-parameter mock theta functions:U1(x,q):=(1+x)(1+xq)(1+x1q)n=0(1;q)2nq3n(xq,x1q;q2)n+1,U2(x,q):=(1+q)(1+xq)(1+x1q)n=0(1,q3;q2)nqn(xq,x1q;q2)n+1,U3(x,q):=(1+q)(1xq)(1x1q)n=0(q,q3;q2)n(1)nq2n+1(xq2,x1q2;q2)n+1. In addition, by finding the following functions:E1(q):=n=0qn2(q;q2)n+1,E2(q):=n=0(1q)qn2+2n(q2;q2)n,E3(q):=n=0(1q)qn2+n(q;q2)n, we establish some relations between these functions and some third order mock theta functions, and then provide the Hecke-type double sums for these third order mock theta functions.

Here and throughout the paper, we use the standard q-series notation [16]. Let q denote a complex number with |q|<1. Then for positive integers n and m,(a;q)0:=1,(a;q)n:=k=0n1(1aqk),(a;q)n:=q(n2)(q/a)n(q/a;q)n,(a;q):=k=0(1aqk),(a1,a2,,am;q):=(a1;q)(a2;q)(am;q). The (unilateral) basic hypergeometric series ϕsr is defined asϕsr(a1,a2,,arb1,,bs;q,x)=n=0(a1,a2,,ar;q)n(q,b1,,bs;q)n((1)nq(n2))1+srxn. Jacobi's triple product identity is given byj(x;q):=(x;q)(q/x;q)(q;q)=n=(1)nq(n2)xn. Let a and m be integers with m positive. Then defineJa,m:=j(qa;qm),Ja,m:=j(qa;qm),Jm:=Jm,3m=i=1(1qmi).

Ramanujan listed 17 mock theta functions which were assigned orders 3, 5, and 7 in the last letter to Hardy. The following two third order mock theta functions are included in the list.ϕ(q)=n=0qn2(q2;q2)n,ψ(q)=n=1qn2(q;q2)n. In 1936, Waston [32] found another three third order mock theta functions. For example,ν(q)=n=0qn2+n(q;q2)n+1. Then Waston [33] proved some linear relations for the fifth order mock theta functions given by Ramanujan. In 1981, by means of Bailey chains, Andrews [2] established the Hecke-type double sums for the fifth and seventh order mock theta functions. Hecke-type double sums have the following form:(m,n)D(1)H(m,n)qQ(m,n)+L(m,n), where H(m,n) and L(m,n) are linear forms, Q(m,n) is an indefinite quadratic form, and D is some subset of Z×Z such that Q(m,n)0 for all (m,n)D. For example, Andrews gave the following Hecke-type identities:J1f0(q)=n=0(1q4n+2)q5n2+n2j=nn(1)jqj2,J1f1(q)=n=0(1q2n+1)q5n2+3n2j=nn(1)jqj2, where the fifth order mock theta functions f0(q) and f1(q) are stated asf0(q)=n=0qn2(q;q)nandf1(q)=n=0qn2+n(q;q)n. Subsequently, Hecke-type double sums were widely used to prove identities related to mock theta functions. Based on (1.8) and (1.9), Hickerson [21] introduced the universal mock theta function g(x,q),g(x,q)=x1(1+n=0qn2(x;q)n+1(x1q;q)n)=n=0qn(n+1)(x,x1q;q)n+1, and then provedf0(q)=J5,10J2,5J12q2g(q2,q10),f1(q)=J5,10J4,5J12q3g(q4,q10). The above identities are called as “mock theta conjectures” which express the fifth order mock theta functions in terms of g(x,q) and theta functions. It is customary to refer to the analogous identities involving other mock theta functions as mock theta conjectures. Later, based on g(x,q) and the Hecke-type double sums given by Andrews [2], three mock theta conjectures related to the seventh order mock theta functions were proved in [20]. Furthermore, McIntosh [28] proved the mock theta conjectures for the third order mock theta functions. For example,ϕ(q)=2qg3(q,q4)+J27J13J43, where he used g3(x,q) to denote g(x,q). In 2012, in view of q-orthogonal polynomials, Andrews [3] established some Hecke-type double sums related to some known and new third order mock theta functions. For example,1+ψ(q)=1J1n=0(1q6n+6)(1)nq2n2+nj=0nq(j+12). In [22], Hickerson and Mortenson gave the following definition for a special type of Hecke-type double sums.

Definition 1.1

Let x, yC:=C{0} and define sg(r) :=1 for r0 and sg(r) :=1 for r<0. Thenfa,b,c(x,y,q):=sg(r)=sg(s)sg(r)(1)r+sxrysqa(r2)+brs+c(s2).

It is clear thatfa,b,a(x,y,q)=fa,b,a(y,x,q). In [29], Mortenson obtained different Hecke-type double sums for some third order mock theta functions. For example,1+2ψ(q)=1J1n=0(1+q2n+1)(1)nq2n2+nj=nnq(j+12), which can be rewritten asψ(q)+1=1J1f3,5,3(q2,q3,q).

For the even order mock theta functions, in view of Bailey's Lemma, Andrews and Hickerson [6] established the Hecke-type double sums for the sixth order mock theta functions, and proved the linear relations for them given by Ramanujan. In [7], Berndt and Chan found two more sixth order mock theta functions. Furthermore, by means of some classical q-series identities, Lovejoy [25] proved some relations for the sixth order mock theta functions. In view of the half-shifted method, Gordon and McIntosh [17] found eight eighth order mock theta functions. Later, McIntosh [26] studied three second order mock theta functions. In [14], the authors provided the Hecke-type double sums for the second and eighth order mock theta functions. For the tenth order mock theta functions appearing in the Lost Notebook [31], Choi [10], [11], [12], [13] established the Hecke-type double sums for these functions, and proved eight linear relations given by Ramanujan.

In [19], it shows that the mock theta functions with odd order can be expressed by the universal mock theta function g3(x,q). Similarly, the mock theta functions with even order are related to another universal mock theta function g2(x,q),g2(x,q)=n=0(q;q)nqn(n+1)/2(x,x1q;q)n+1.

In 2014, Hickerson and Mortenson [22] provided the following definition of Appell–Lerch sums.

Definition 1.2

Let x, zC:=C{0} with neither z nor xz an integral power of q. Thenm(x,q,z):=1j(z;q)r=(1)rq(r2)zr1qr1xz.

Changing r to r+1 in the above series gives another useful form for m(x,q,z) [22, Eq. (3.1)]:m(x,q,z)=zj(z;q)r=(1)rq(r+12)zr1qrxz. Hickerson and Mortenson showed that all of Ramanujan's classical mock theta functions as well as those of Gordon and McIntosh can be expressed in terms of Appell–Lerch sums. In light of Appell-Lerch sums, results for mock theta functions are much more accessible than they were 20 years ago. For more on mock theta functions, one can see [1], [4], [5], [8], [15], [18], [27].

In this paper, we establish the Appell–Lerch sums for U1(x,q), U2(x,q), and U3(x,q).

Theorem 1.3

We haveU1(x,q)=2xq1(1q)m(x,q2,q)2(1+x)m(q,q2,q).

Theorem 1.4

We haveU2(x,q)=2(1+x)m(x,q2,q)+2(1q)m(q,q2,q).

Theorem 1.5

We haveU3(x,q)=2(1x)m(x,q2,1)2(1q)m(q,q2,1).

Notice that by using Corollary 3.2 and Theorem 3.3 in [22], we can evaluate the following terms.m(q,q2,q)=m(q,q2,1)=12. In addition, in view of some q-series identities given by Liu [23], [24] and Chen and Wang [9], we obtain the following Hecke-type double sums related to E1(q), E2(q), and E3(q).

Theorem 1.6

We haveE1(q)1=2q5J1,4f1,2,1(q9,q11,q4).

Theorem 1.7

We haveE2(q)2=1J1,4f1,2,1(iq,iq,q).

Theorem 1.8

We haveE3(q)1=qJ2,4f1,2,1(iq32,iq32,q).

Meanwhile, we provide some relations between these functions and the third order mock theta functions.

Theorem 1.9

We haveE1(q)2ψ(q)=1,E2(q)+ϕ(q)=2,E3(q)+qν(q)=1.

Then according to Theorem 1.6, Theorem 1.9, we derive the following Hecke-type double sums for ψ(q), ϕ(q), and ν(q).

Corollary 1.10

We haveψ(q)=q5J1,4f1,2,1(q9,q11,q4),ϕ(q)=1J1,4f1,2,1(iq,iq,q),ν(q)=1J2,4f1,2,1(iq32,iq32,q).

Notice that the Hecke-type double sum of ν(q) [9, Equation (6.30)] can be converted to (1.14).

This paper is organized as follows. In Section 2, we state some lemmas which are used to prove the main theorems. In Section 3, we prove Theorem 1.3, Theorem 1.5. In Section 4, we prove Theorem 1.6, Theorem 1.9.

Section snippets

Preliminaries

In this paper, we need the following identities given by Liu [23], [24].

Lemma 2.1

[23, Theorem 1.7], [24, p. 2089] For max{|uab/q|,|ua|,|ub|,|c|,|d|}<1,(uq,uab/q;q)(ua,ub;q)3ϕ2(q/a,q/b,vc,d;q,uabq)=n=0(1uq2n)(u,q/a,q/b;q)n(1u)(q,ua,ub;q)n(uab)nqn23n23ϕ2(qn,uqn,vc,d;q,q).

Lemma 2.2

[24, Eq. (3.14)] For any nonnegative integer n,ϕ23(qn,αqn,βc,d;q,q)=(c)nqn2n2(qα/c;q)n(c;q)n×3ϕ2(qn,αqn,d/βd,qα/c;q,qβ/c).

Chen and Wang [9] established the following identities.

Lemma 2.3

[9, Lemma 2.2] For any nonnegative integer

Proofs of Theorems 1.3-1.5

In this section, using Lemma 2.1, Lemma 2.4, and the properties of Appell-Lerch sums, we prove Theorem 1.3, Theorem 1.5.

Proof of Theorem 1.3

Setting (a,b,c,d,u,v,q)(q2,q,xq3,x1q3,q2,q2,q2) in (2.1), we deducen=0(1;q)2nq3n(xq,x1q;q2)n+1=2(1q)(1xq)(1x1q)J1,2n=0(1q2n+1)(1)nqn2+2n(1+q2n)(1+q2n+2)×3ϕ2(q2n,q2n+2,q2xq3,x1q3;q2,q2). Then applying (2.4) with (a,b,c,q)(q2,xq,x1q,q2) yields thatϕ23(q2n,q2n+2,q2xq3,x1q3;q2,q2)=(xq,x1q;q2)nq2n(xq3,x1q3;q2)n. Substituting (3.2) into (3.1), we haven=0(1;

Proofs of Theorems 1.6-1.9

In this section, by using the lemmas in Section 2, we prove Theorem 1.6, Theorem 1.9.

Proof of Theorem 1.6

Setting (a,b,c,d,u,v,q)(0,q1,q3,q3,q2,q2,q2) in (2.1), we findn=0qn2(q;q2)n+1=1(1q)J1,4n=0(1q4n+2)(q3;q2)n(1)nq2n2n(q;q2)n×3ϕ2(q2n,q2n+2,q2q3,q3;q2,q2)=1(1q)J1,4n=0(1q4n+2)(1)nq3n2+n×3ϕ2(q2n,q2n+2,qq3,q;q2,q), where the second equality follows from (2.2) with (α,β,c,d,q)(q2,q2,q3,q3,q2). Then setting (α,c,d,q)(1,q,q,q2) in (2.3) yields thatϕ23(q2n,q2n+2,qq3,q;q2,q)=qn(q;q2)n(q3;q2)

Acknowledgements

We would like to thank the referee for valuable comments and suggestions. This work was supported by the National Natural Science Foundation of China (Grant Nos. 11771222 and 12001309), the Fundamental Research Funds for the Central Universities (Nankai University), the Foreign Youth Exchange Program of China Association of Science and Technology, and the Natural Science Foundation Youth Fund of Qinghai Normal University.

References (34)

  • Y.-S. Choi

    Tenth order mock theta functions in Ramanujan's lost notebook

    Invent. Math.

    (1999)
  • Y.-S. Choi

    Tenth order mock theta functions in Ramanujan's lost notebook. IV

    Trans. Am. Math. Soc.

    (2002)
  • Y.-S. Choi

    Tenth order mock theta functions in Ramanujan's lost notebook. III

    Proc. Lond. Math. Soc. (3)

    (2007)
  • S.-P. Cui et al.

    On second and eighth order mock theta functions

    Ramanujan J.

    (2019)
  • F.G. Garvan

    Universal mock theta functions and two-variable Hecke-Rogers identities

    Ramanujan J.

    (2015)
  • G. Gasper et al.

    Basic Hypergeometric Series

    (2004)
  • B. Gordon et al.

    Some eighth order mock theta functions

    J. Lond. Math. Soc. (2)

    (2000)
  • Cited by (5)

    • New families of mock theta functions and partial fraction decomposition

      2022, Advances in Applied Mathematics
      Citation Excerpt :

      The aim of this paper is to obtain genuine mock theta functions and express them in term of Appell-Lerch sums via partial fraction decomposition. In particular, we prove some identities involving several families of two-parameter mock theta functions and Appell-Lerch sums which imply some results due to Cui and Gu [11]. The main results of the paper can be stated as follows.

    • New mock theta functions and formulas for basic hypergeometric series

      2023, Proceedings of the Edinburgh Mathematical Society
    • Generalizations of Mock Theta Functions and Appell–Lerch Sums

      2023, Bulletin of the Iranian Mathematical Society
    • Families of identities involving universal mock theta functions

      2023, International Journal of Number Theory
    View full text