Abstract
Recently Gordon and McIntosh introduced the third order mock theta function \(\xi (q)\) defined by
Our goal in this paper is to study arithmetic properties of the coefficients of this function. We present a number of such properties, including several infinite families of Ramanujan-like congruences.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
In his last letter to Hardy in 1920, Ramanujan introduced the notion of a mock theta function. He listed 17 such functions having orders 3, 5, and 7. Since then, other mock theta functions have been found. Gordon and McIntosh [8], for example, introduced many additional such functions, including the following of order 3:
where we use the standard q-series notation:
Arithmetic properties of the coefficients of mock theta functions have received a great deal of attention. For instance, Zhang and Shi [15] recently proved seven congruences satisfied by the coefficients of the mock theta function \(\beta (q)\) introduced by McIntosh. In a recent paper, Brietzke et al. [5] found a number of arithmetic properties satisfied by the coefficients of the mock theta function \(V_0(q)\), introduced by Gordon and McIntosh [7]. Andrews et al. [2] prove a number of congruences for the partition functions \(p_{\omega }(n)\) and \(p_{\nu }(n)\), introduced in [1], associated with the third order mock theta functions \(\omega (q)\) and \(\nu (q)\), where \(\omega (q)\) is defined below and
In a subsequent paper, Wang [14] presented some additional congruences for both \(p_{\omega }(n)\) and \(p_{\nu }(n)\).
This paper is devoted to exploring arithmetic properties of the coefficients \(p_{\xi }(n)\) defined by
It is clear from (1) that \(p_{\xi }(n)\) is even for all \(n \ge 1\). In Sects. 4 and 5, we present other arithmetic properties of \(p_{\xi }(n)\), including some infinite families of congruences.
2 Preliminaries
McIntosh [12, Theorem 3] proved a number of mock theta conjectures, including
where
and \(\omega (q)\) is the third order mock theta functions given by
It follows from (1), (3), and (4) that
Throughout the remainder of this paper, we define
in order to shorten the notation. Combining (5) and (2), we have
We recall Ramanujan’s theta functions
The function \(\phi (q)\) satisfies many identities, including (see [3, (22.4)])
In some of the proofs, we employ the classical Jacobi’s identity (see [4, Theorem 1.3.9])
We note the following identities which will be used below.
Lemma 1
The following 2-dissection identities hold:
Proof
By Entry 25 (i), (ii), (v), and (vi) in [3, p. 40], we have
Using (7) and (8) we can rewrite (19) in the form
from which we obtain (11) after multiplying both sides by \(\frac{f_4^2}{f_2^5}\). Identity (12) can be easily deduced from (11) using the procedure described in Section 30.10 of [9].
By (7) and (8) we can rewrite (20) in the form
from which we obtain (13).
Identities (14), (15), and (18) are equations (30.10.3), (30.9.9), and (30.12.3) of [9], respectively. Finally, for proofs of (16) and (17) see [13, Lemma 4]. \(\square \)
The next lemma exhibits the 3-dissections of \(\psi (q)\) and \(1/\phi (-q)\).
Lemma 2
We have
Proof
Identity (21) is Eq. (14.3.3) of [9]. A proof of (22) can be seen in [10]. \(\square \)
3 Dissections for \(p_{\xi }(n)\)
This section is devoted to proving the 2-, 3-, and 4-dissections of (2). We begin with the 2-dissection.
Theorem 1
We have
Proof
We start with equation (4) of [2]:
where f(q) is the mock theta function
and
Thus,
Using (5), it follows that
By (11), we have
which along with (11) allows us to rewrite (25) as
Thus,
Dividing (27) by q and replacing \(q^2\) by q in the resulting identity and in (26), we obtain (23) and (24). \(\square \)
The next theorem exhibits the 3-dissection of (2).
Theorem 2
We have
Proof
In view of (8), we rewrite (6) as
Using (21), we obtain
Extracting the terms of the form \(q^{3n+r}\) on both sides of (31), for \(r \in \{0,1,2\}\), dividing both sides of the resulting identity by \(q^r\) and then replacing \(q^3\) by q, we obtain the desired results. \(\square \)
We close this section with the 4-dissection of (2).
Theorem 3
We have
Proof
In order to prove (32), we use (13) and (18) to obtain the even part of (24), which is given by
Replacing \(q^2\) by q we obtain (32).
Using (13) and (18) we can extract the odd part of (23):
After simplifications we arrive at (33).
Next, extracting the odd part of (24) with the help of (13) and (18) yields
which, after simplifications, gives us (34).
In order to obtain (35), we use (13) and (18) in (23) to extract its even part:
Replacing \(q^2\) by q in this identity, we obtain (35). \(\square \)
4 Arithmetic properties of \(p_{\xi }(n)\)
Our first observation provides a characterization of \(p_{\xi }(3n) \pmod {4}.\)
Theorem 4
For all \(n \ge 0\), we have
Proof
By (28), using (9) and the fact that \(f_k^4 \equiv f_{2k}^2 \pmod {4}\) for all \(k\ge 1,\) it follows that
By (7), we obtain
which completes the proof. \(\square \)
Theorem 4 yields an infinite family of Ramanujan-like congruences modulo 4.
Corollary 1
For all primes \(p \ge 3\) and all \(n\ge 0,\) we have
if r is a quadratic nonresidue modulo p.
Proof
If \(pn+r = k^2\), then \(r \equiv k^2 \pmod {p}\), which contradicts the fact that r is a quadratic nonresidue modulo p. \(\square \)
Since \(\gcd (3, p) = 1\), among the \(p-1\) residues modulo p, we have \(\frac{p-1}{2}\) residues r for which r is a quadratic nonresidue modulo p. Thus, for instance, the above corollary yields the following congruences:
Theorem 5
For all \(n \ge 0\), we have
Proof
From Theorem 2,
So we only need to consider the parity of
Note that
where \(a_3(n)\) is the number of 3-core partitions of n (see [11, Theorem 1]). Thanks to [6, Theorem 7], we know that
This completes the proof. \(\square \)
Theorem 5 yields an infinite family of congruences modulo 4.
Corollary 2
For all primes \(p > 3\) and all \(n\ge 0,\) we have
if \(3r+1\) is a quadratic nonresidue modulo p.
Proof
If \(3(pn+r)+1 = k^2\), then \(3r+1 \equiv k^2 \pmod {p}\), which would be a contradiction with \(3r+1\) being a quadratic nonresidue modulo p. \(\square \)
For example, the following congruences hold for all \(n\ge 0\):
We next turn our attention to the arithmetic progression \(4n+2\) to yield an additional infinite family of congruences.
Theorem 6
For all \(n \ge 0\), we have
Proof
From (34), we obtain
Using Euler’s identity (see [9, Eq. (1.6.1)])
we obtain
which concludes the proof. \(\square \)
Theorem 6 yields an infinite family of congruences modulo 4.
Corollary 3
Let \(p > 3\) be a prime and r an integer such that \(2r+1\) is a quadratic nonresidue modulo p. Then, for all \(n \ge 0\),
Proof
If \(pn+r = 6k(3k\pm 1)\), then \(r \equiv 18k^2 \pm 6k \pmod {p}\). Thus, \(2r+1 \equiv (6k \pm 1)^2 \pmod {p}\), which contradicts the fact that \(2r+1\) is a quadratic nonresidue modulo p. \(\square \)
Thanks to Corollary 3, the following example congruences hold for all \(n\ge 0\):
We now provide a mod 8 characterization for \(p_{\xi }(3n).\)
Theorem 7
For all \(n \ge 0\), we have
Proof
By (28), using (7) and (9), we have
which yields
Since
we have
Therefore
which completes the proof. \(\square \)
As with the prior results, Theorem 7 provides an effective way to yield an infinite family of congruences modulo 8.
Corollary 4
Let p be a prime such that \(p \equiv \pm 1 \pmod {24}\). Then
if r is a quadratic nonresidue modulo p.
Proof
Since \(p \equiv \pm 1 \pmod {8}\) and \(p \equiv \pm 1 \pmod {12}\), it follows that 2 and 3 are quadratic residues modulo p. Thus, r, 2r, 3r, and 6r are quadratic nonresidues modulo p. Indeed, according to the properties of Legendre’s symbol, for \(j\in \{1,2,3,6\}\), we have
It follows that we cannot have \(3(pn+r) = jk^2,\) for some \(k \in {\mathbb {N}}\) and \(j \in \{ 1, 2, 3, 6 \}\). In fact, \(3(pn+r) = jk^2\) would imply \(3(pn+r) \equiv 3r \equiv jk^2 \pmod {p}\). However, for \(j = 1, 2, 3, 6\), this would imply that 3r, 6r, r, or 2r, respectively, is a quadratic residue modulo p, which would be a contradiction since 2, 3, and 6 are quadratic residues modulo p. The result follows from Theorem 7. \(\square \)
As an example, we note that, for \(p=23\) and all \(n\ge 0,\) we have
Theorem 8
For all \(n \ge 0\), we have
Proof
Initially we use (14) to extract the odd part on both sides of (29). The resulting identity is
The result follows using (29). \(\square \)
Now we present complete characterizations of \(p_{\xi }(48n+4)\) and \(p_{\xi }(12n+1)\) modulo 8.
Theorem 9
For all \(n \ge 0\), we have
Proof
The first congruence follows directly from Theorem 8. Replacing (14) in (29), we obtain
Extracting the terms of the form \(q^{2n}\), we have
which, after replacing \(q^2\) by q, yields
Now we use (15) to obtain
which completes the proof. \(\square \)
Theorem 9 also provides an effective way to yield an infinite family of congruences modulo 8.
Corollary 5
For all primes \(p > 3\) and all \(n\ge 0\), we have
if \(12r+1\) is a quadratic nonresidue modulo p.
Proof
Let \(p>3\) be a prime and \(12r+1\) a quadratic nonresidue modulo p. If \(pn+r = k(3k\pm 1)\), then \(r \equiv 3k^2\pm k \pmod {p}\), which implies that \(12r+1 \equiv (6k\pm 1)^2 \pmod {p}\), a contradiction. The result follows from Theorem 9. \(\square \)
5 Additional congruences
In this section, we prove several additional Ramanujan-like congruences that are not included in the results of the previous section.
Theorem 10
For all \(n \ge 0\), we have
Proof
Using (15) we can now 2-dissect (40) to obtain
from which we have
Now, dividing both sides of the above expression by q and replacing \(q^2\) by q, we obtain
Taking the odd parts on both sides of the last equation, we are left with
which proves (41).
In order to prove (42), we use (22) to extract the terms of the form \(q^{3n}\) of (28). The resulting identity is
which, after replacing \(q^3\) by q and using (8), yields
By (8), we have
Since \(n(n+1)/2 \not \equiv 2 \pmod {3}\) for all \(n\ge 0,\) all terms of the form \(q^{3n+2}\) in the last expression have coefficients congruent to \(0 \pmod {3}\), which proves (42).
We now prove (43). Replacing (22) in (28) and extracting the terms of the form \(q^{3n+2}\), we obtain
Dividing both sides of (45) by \(q^2\) and replacing \(q^3\) by q, we have
Now we use (11) to extract the odd part of (46) and obtain
Since \(f_1^3 \equiv f_3 \pmod {3}\), we have
Using (15) we obtain
Since the odd part of (17) is divisible by 3, then the coefficients of the terms of the form \(q^{2n+1}\) in \(\sum _{n=0}^{\infty } p_{\xi }(36n+15)q^{n}\) are congruent to 0 modulo 3. This completes the proof of (43). \(\square \)
We now prove a pair of unexpected congruences modulo 5 satisfied by \(p_{\xi }(n).\)
Theorem 11
For all \(n \ge 0\), we have
Proof
By (46), we have
Thanks to Jacobi’s identity (10) we know
Note that, for all integers j and k, \(3j(j+1)\) and \(k(k+1)/2\) are congruent to either 0, 1 or 3 modulo 5. The only way to obtain \(3j(j+1)+k(k+1)/2 = 5n+3\) is the following:
-
\(3j(j+1) \equiv 0 \pmod {5}\) and \(k(k+1)/2 \equiv 3 \pmod {5}\), or
-
\(3j(j+1) \equiv 3 \pmod {5}\) and \(k(k+1)/2 \equiv 0 \pmod {5}\).
Thus, \(j \equiv 2 \pmod {5}\) or \(k \equiv 2 \pmod {5}\) in all possible cases, and this means
Therefore, for all \(n \ge 0\), \(p_{\xi }(45n+33) = p_{\xi }(9(5n+3)+6) \equiv 0 \pmod {5}\), which is (47).
In order to complete the proof of (48), we want to see when
Four possible cases arise:
-
\(k \equiv 1 \pmod {5}\) and \(j \equiv 2 \pmod {5}\),
-
\(k \equiv 3 \pmod {5}\) and \(j \equiv 2 \pmod {5}\),
-
\(j \equiv 1 \pmod {5}\) and \(k \equiv 2 \pmod {5}\), or
-
\(j \equiv 3 \pmod {5}\) and \(k \equiv 2 \pmod {5}\).
In all four cases above, either \(j \equiv 2 \pmod {5}\) or \(k \equiv 2 \pmod {5}\). So
in all these cases. Therefore,
which completes the proof of (48). \(\square \)
Next, we prove three congruences modulo 8 which are not covered by the above results.
Theorem 12
For all \(n \ge 0\), we have
Proof
Initially we prove (49). From (34) and (7) we have
Now we can use (11), (12), and (20) to extract the terms involving \(q^{2n+1}\) from both sides of the previous congruence:
After dividing both sides by q and then replacing \(q^2\) by q, we are left with
whose odd part is congruent to 0 modulo 8, which implies (49).
In order to prove (50), we use (15) to obtain the even part of identity (40), which is
Now, employing (13), we obtain the odd part of the last identity, which is
which implies (50).
Now we prove (51). We employ (15) in (39) to obtain
By (12) and (13), we rewrite (52) in the form
from which we obtain
Dividing both sides by q and replacing \(q^2\) by q, we are left with
which implies (51). \(\square \)
We close this section by proving a congruence modulo 9.
Theorem 13
For all \(n \ge 0\), we have
Proof
We use (21) to extract the terms of the form \(q^{3n+1}\) from (32). The resulting identity is
which, after dividing by q and replacing \(q^3\) by q, yields
Using (13) and (14), we extract the even part on both sides of the above identity to obtain
Now we employ (18) and (16) to extract the odd part on both sides of the last congruence:
which implies (53). \(\square \)
6 Concluding remarks
Computational evidence indicates that \(p_{\xi }(n)\) satisfies many other congruences. The interested reader may wish to consider the following two conjectures.
Conjecture 1
Conjecture 2
Clearly, once proven, Conjectures 1 and 2 would immediately lead to infinite families of Ramanujan-like congruences. Morever, Conjecture 2 would immediately imply Theorem 13 since \(96n+76 = 32(3n+2)+12\) while the right-hand side of Conjecture 2 is clearly a function of \(q^3.\) The same argument would imply that, for all \(n\ge 0,\)
since \(96n+44 = 32(3n+1)+12.\)
References
Andrews, G.E., Dixit, A., Yee, A.J.: Partitions associated with the Ramanujan/Watson mock theta functions \(\omega (q)\) and \(\phi (q)\). Res. Number Theory 1, 1–19 (2015)
Andrews, G.E., Passary, D., Sellers, J.A., Yee, A.J.: Congruences related to the Ramanujan/Watson mock theta functions \(\omega (q)\) and \(\nu (q)\). Ramanujan J. 43, 347–357 (2017)
Berndt, B.C.: Ramanujan’s Notebooks. Part III, Springer, New York (1991)
Berndt, B.C.: Number Theory in the Spirit of Ramanujan. American Mathematical Society, Providence (2006)
Brietzke, E.H.M., da Silva, R., Sellers, J.A.: Congruences related to an eighth order mock theta function of Gordon and McIntosh. J. Math. Anal. Appl. 479, 62–89 (2019)
da Silva, R., Sellers, J.A.: Parity Considerations for the Mex-Related Partition Functions of Andrews and Newman, J. Integer Seq. 23, Article 20.5.7 (2020)
Gordon, B., McIntosh, R.J.: Some eight order mock theta functions. J. Lond. Math. Soc. 62, 321–335 (2000)
Gordon, B., McIntosh, R.J.: Modular transformations of Ramanujan’s fifth and seventh order mock theta functions. Ramanujan J. 7, 193–222 (2003)
Hirschhorn, M.D.: The Power of \(q\), a Personal Journey. Developments in Mathematics, vol. 49. Springer, Cham (2017)
Hirschhorn, M.D., Sellers, J.A.: Arithmetic relations for overpartitions. J. Combin. Math. Combin. Comput. 53, 65–73 (2005)
Hirschhorn, M.D., Sellers, J.A.: Elementary proofs of various facts about 3-cores. Bull. Aust. Math. Soc. 79, 507–512 (2009)
McIntosh, R.J.: New mock theta conjectures Part I. Ramanujan J. 46, 593–604 (2018)
Naika, M.S.M., Gireesh, D.S.: Congruences for \(3\)-regular partitions with designated summands. Integers 16, A25 (2016)
Wang, L.: New congruences for partitions related to mock theta functions. J. Number Theory 175, 51–65 (2017)
Zhang, W., Shi, J.: Congruences for the coefficients of the mock theta function \(\beta (q)\). Ramanujan J. 49, 257–267 (2019)
Acknowledgements
The authors thank the anonymous referee for carefully reading the manuscript and his/her helpful comments and suggestions.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Robson da Silva was supported by São Paulo Research Foundation (FAPESP) (Grant No. 2019/14796-8)
Rights and permissions
About this article
Cite this article
da Silva, R., Sellers, J.A. Congruences for the coefficients of the Gordon and McIntosh mock theta function \(\xi (q)\). Ramanujan J 58, 815–834 (2022). https://doi.org/10.1007/s11139-021-00479-8
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11139-021-00479-8