CONGRUENCES RELATED TO THE RAMANUJAN/WATSON MOCK THETA FUNCTIONS ω(q) AND ν(q)

Transcription

1 CONGRUENCES RELATED TO THE RAMANUJAN/WATSON MOCK THETA FUNCTIONS ωq) AND νq) GEORGE E. ANDREWS, DONNY PASSARY, JAMES A. SELLERS, AND AE JA YEE Abstract. Recently, Andrews, Dixit, and Yee introduced partition functions associated with the Ramanujan/Watson mock theta functions ωq) and νq). Based on one of their results, mod 2 congruences for these mock theta functions are obtained. In addition, infinite families of mod 4 and mod 8 congruences are presented. Lastly, an elementary proof of the first explicit examples of congruences for ωq) given by Waldherr is presented. 1. Introduction In his last letter to Hardy in 1920, Ramanujan introduced the notion of a mock theta function along with a number of examples of order, 5, and 7. Since then, mock theta functions have been the subject of intense study. Recently, the first and fourth authors with A. Dixit found a new partition function p ω n) that is associated with the third order mock theta function ωq) [1]: where ωq) q 2n2 +n) q; q 2 ) 2, n+1 a; q) n : 1 a)1 aq) 1 aq n 1 ). One of the results in [1] yields a mod 2 congruence of the coefficients of ωq), which led us to a further search for congruences. S. Garthwaite and D. Penniston [5] showed that the coefficients of ωq) satisfy infinitely many congruences of the similar type as Ramanujan s partition congruences, and M. Waldherr [6] found the first explicit examples of congruences, suggested by some computations done by J. Lovejoy. We define p ω n) by p ω n)q n q n 1 q n )q n+1 ; q) n q 2n+2 ; q 2 ), 1) where a; q) : lim n a; q) n. From its generating function definition, we see that p ω n) counts the number of partitions of n in which each odd part is less that twice the smallest part. In [1], it is shown that The main result of this paper is: p ω n)q n qωq). 1

2 2 GEORGE E. ANDREWS, DONNY PASSARY, JAMES A. SELLERS, AND AE JA YEE Theorem 1.1. For nonnegative integers n and k, p w 2 2k+ n + 11 ) 22k mod 4), p w 2 2k+ n + 17 ) 22k mod 8), p w 2 2k+4 n + 8 ) 22k mod 4). In [1], another partition function p ν n) is defined. Namely, p ν n)q n q n q n+1 ; q) n q 2n+2 ; q). It is shown that p ν n)q n ν q), where νq) is a third order mock theta function, q nn+1) νq) q; q 2. ) n+1 This mock theta function νq) is related to ωq) as follows [4, p. 62, Equation 26.88)]: where ν q) qωq 2 ) + q 2 ; q 2 ) ψq 2 ), 2) ψq) q nn+1)/2. By 2), we can derive congruences of p ν n) from p ω n), which will be given in Section 4. This paper is organized as follows. In Section 2, we present mod 2 congruences of p ω n) and p ν n). In Section, we prove Theorems 1.1. In Section 4, we prove congruences of p ν n). As noted earlier, Waldherr [6] provided the first explicit congruences for ωq): p ω 40n + 28) 0 mod 5), p ω 40n + 6) 0 mod 5). In Section 5, we provide an elementary proof of the above congruences. 2. Mod 2 congruences We recall the following results from [1]: q n q n ; q) n+1 q 2n+2 ; q 2 1) j q 6j2 +4j q 4j+2 ), ) ) and q n q n+1 ; q) n q 2n+2 ; q 2 ) j0 1) j q jj+2) 1 + q 2j+1 ). 4) j0

3 CONGRUENCES RELATED TO THE RAMANUJAN/WATSON MOCK THETA FUNCTIONS ωq) AND νq) Let p ω,o n) and p ω,e n) be the number of partitions of n counted by p ω n) into an odd number of parts and an even number of parts, respectively. Then it follows from ) that { 1) j, if n 6j 2 + 4j + 1 for some j Z, p ω,o n) p ω,e n) 0, otherwise. This yields the following obvious result. Theorem 2.1. We have p ω n) { 1 mod 2), if n 6j 2 + 4j + 1 for some j Z, 0 mod 2), otherwise. Corollary 2.2. Let p 5 be prime, and let r be such that 0 r p 1, and r 1)p + 1)/2 is a quadratic nonresidue mod p. Then, for all n 0, p ω pn + r) 0 mod 2). Proof. Assume that there exists an integer j such that Then This yields pn + r 6j 2 + 4j + 1. pn + r) 1 6j 2 + 4j + 1) 1 18j j j 2 + 6j + 1) 2j + 1) 2. r 1 2j + 1) 2 mod p) or r 1) p + 1) j + 1) 2 mod p). 2 But, we have chosen r such that r 1) p+1) 2 is a quadratic nonresidue mod p, so it cannot be congruent to a square. Similarly, 4) yields the following mod 2 result. Theorem 2.. We have p ν n) Corollary 2.4. For all n 0 and r 2,, { 1 mod 2), if n j 2 + 2j for some j Z, 0 mod 2), otherwise. p ν 4n + r) 0 mod 2). Proof. If j is even, then j 2 + 2j 0 mod 4). If j is odd, then j 2 + 2j 1 mod 4). So 4n + 2 and 4n + can never be represented as j 2 + 2j. Corollary 2.5. Let p 5 be prime, and let r be such that 0 r p 1, and r + 1) is a quadratic nonresidue mod p. Then, for all n 0, p ν pn + r) 0 mod 2).

4 4 GEORGE E. ANDREWS, DONNY PASSARY, JAMES A. SELLERS, AND AE JA YEE Proof. For some j, let which is equivalent to Replacing n by pn + r yields and this gives n j 2 + 2j, n + 1 j + 1) 2 pn + r) + 1 j + 1) 2, r + 1 j + 1) 2 mod p). But, we have chosen r such that r+1 is a quadratic nonresidue mod p, so it cannot be congruent to a square. We start with a formula from page 6 of [7]: where. Proof of Theorem 1.1 fq 8 ) + 2qωq) + 2q ω q 4 ) φq)φq2 ) 2 φq) ψq) By [, p.40, Entry 25 v), vi)], we have Lemma.1. Let Then q 4 ; q 4 ) 2 : F q), 5) q n2 q; q 2 ) 2 q 2 ; q 2 ), 6) q n+1 2 ) q 2 ; q 2 ) q; q 2 ). 7) φq) 2 φq 2 ) 2 + 4qψq 4 ) 2. 8) F q) F 0 q 4 ) + qf 1 q 4 ) + q 2 F 2 q 4 ) + q F q 4 ). F 0 q) φq) q; q) 2, 9) F 1 q) 2ψq2 )φq) 2 q; q) 2, 10) F 2 q) 4φq)φq2 )ψq 4 ) q; q) 2, 11) F q) 8ψq2 )φq 2 )ψq 4 ) q; q) 2. 12) Proof. Separating even n s and odd n s, we obtain φq) q 4n2 + q 4n2 +4n+1.

5 CONGRUENCES RELATED TO THE RAMANUJAN/WATSON MOCK THETA FUNCTIONS ωq) AND νq) 5 Thus F q) 1 q 4 ; q 4 ) 2 q 4n2 + q 1 q 4 ; q 4 ) 2 q 4n2 + q ) q 8n q 2 q 4n2 +4n q 4n2 +4n ) ) q 8n2 q 8n2 + q 2 q 8n2 +8n + q 4 ) 2 q 8n2 +8n q 8n2 +8n ) ) 2 Consequently extracting the portion for F 0 q 4 ), we have F 0 q 4 1 ) q 4 ; q 4 ) 2 where the last equality follows from q 4n2 φq4 ) φq 8 q 4 ; q 4 ) 2 ) 2 + 4q 4 ψq 16 ) 2), q n2 +n 2 q 8n2 ) 2 + q 4 q n2 +n 2ψq 2 ). q 8n2 +8n ) 2 ) Finally, by 8), we obtain the identity 9). Similarly, the other identities can be proved. We omit the details. The following lemmas can easily be proved by the binomial theorem and induction so we state without proof. Lemma.2. For any positive integer n, Lemma.. For any prime p, 1 + x) 2n 1 + x 2 ) 2n 1 mod 2 n ). 1 + x) p 1 + x p ) mod p). We first prove the case k 0 in Theorem.4 in a separate theorem. Theorem.4. For any nonnegative integer n, p ω 8n + 4) 0 mod 4), p ω 8n + 6) 0 mod 8), p ω 16n + 1) 0 mod 4). Proof. Examining the left hand side of 5), we see that the only terms producing q n for n 0,, 7 mod 8) come from 2qωq). Consequently, if we can prove that in F q) the terms for q 8n+4, q 8n+6, q 16n+1, respectively, have coefficients divisible by 8, we are done.

6 6 GEORGE E. ANDREWS, DONNY PASSARY, JAMES A. SELLERS, AND AE JA YEE For the first congruence, it suffices to prove that the odd powers in F 0 q) vanish mod 8, which implies that the coefficient of q 8n+4 in F q) is divisible by 8. By 9) in Lemma.1, F 0 q) φq) q; q) 2 q2 ; q 2 ) q; q 2 ) 6 q; q) 2 q 2 ; q 2 ) q; q 2 ) 8 q 2 ; q 2 ) 2 q; q 2 ) 2 q; q 2 ) 2 q2 ; q 2 ) q 2 ; q 4 ) 4 q 2 ; q 2 ) 2 q 2 ; q 4 ) 2 mod 8), where the last congruence follows from Lemma.2. Thus mod 8 F 0 q) is an even function, and returning to the observations at the beginning of this section, we see that p w 8n + 4) is divisible by 4. We now prove the second congruence, which is equivalent to that the odd powers in F 2 q) vanish mod 16, which implies that the coefficient of q 8n+6 in F q) is divisible by 16. By 11) in Lemma.1, 1 4 F 1 2q) q; q) 2 φq)φq 2 )ψq 4 ) 1 q; q) 2 ψq) 2 ψq 2 ) 1 q 2 ; q 2 ) 2 q 4 ; q 4 ) q; q) 2 q; q 2 ) 2 q 2 ; q 4 ) 1 q 4 ; q 4 ) q; q 2 ) 4 q 2 ; q 4 ) 1 q 4 ; q 4 ) q 2 ; q 4 ) 2 q 2 ; q 4 ) mod 4). By the analysis at the beginning of this section, we obtain that p w 8n + 6) is divisible by 8. Lastly, for p w 16n + 1), we begin by considering the powers q 16n+1 in F q) which still only come from the term 2qωq). By looking at the dissection of F q) in Lemma.1, the terms q 16n+1 in F q) must come from qf 1 q 4 ). Next, from 16n n + ) + 1, if we can proof that the

7 CONGRUENCES RELATED TO THE RAMANUJAN/WATSON MOCK THETA FUNCTIONS ωq) AND νq) 7 powers of the form q 4n+ in 1 2 F 1q) vanish mod 4, we would be done. By 10) in Lemma.1, 1 2 F 1q) ψq2 )φq) 2 q; q) 2 ψq2 )q 2 ; q 2 ) 2 q; q 2 ) 4 q; q) 2 q4 ; q 4 ) q 2 ; q 4 ) q; q) 2 q; q 2 ) 4 q4 ; q 4 ) q; q) 2 q; q 2 ) 4 q; q 2 ) q; q 2 ) q4 ; q 4 ) q; q 2 ) 4 q; q 2 ) q; q 2 ) q4 ; q 4 ) q; q 2 ) 4 q; q 2 ) q; q 2 ) mod 4) q4 ; q 4 ) q; q 2 ) 2 q 2 ; q 4 ) q4 ; q 4 ) φ q) q 2 ; q 4 ) q 2 ; q 2 ) q2 ; q 2 ) q 2 ; q 2 ) q 2 ; q 4 ) q 2 ; q 2 ) φ q) φ q) q 2 ; q 2 ) 2 ) ) n q n2 q 2 ; q 2 ) 2 q 2 ; q 2 ) ) n q n2 q 2 ; q 2 ) 2. Since q 2 ; q 2 ) 2 does not contain any term of the form q 4n+, it suffices to show that the coefficient of q 4n+ vanishes mod 2 in 1)n q n2 q 2 ; q 2 ) 2. Now 1) n q n2 q 2 ; q 2 ) 2 1) n q n2 q 4 ; q 4 ) mod 2). But is not a quadratic residue modulo 4, so there is no term of q 4n+ in this sum. And so we are done. We now prove Theorem 1.1 by induction on k. For k 0, the congruences are given in Theorem.4. To prove the congruence for any positive integer k, we consider the following sequence g k defined by g k 4g k 1 1, 1) which has the solution g k 2 2k g 0 22k 1.

8 8 GEORGE E. ANDREWS, DONNY PASSARY, JAMES A. SELLERS, AND AE JA YEE In particular, for any nonnegative integer n, 2 2k+ n k +1 if g 0 8n + 4, g k 2 2k+ n k +1 if g 0 8n + 6, 2 2k+4 n k +1 if g 0 16n + 1. From 1), we see that g k mod 4) for k > 0. Thus, on the left hand side of 5), the coefficient of q g k comes from 2qωq) + 2q ω q 4 ). By the definition of p w n), we have 2qωq) + q ω q 4 ) 2qωq) 2q 1 q 4 ω q 4 ) ) ) 2 p w n)q n p w n) 1) n q 4n 1, and the coefficient of q g k in 2qωq) + 2q ω q 4 ) is 2 p ω g k ) 1) g k 1 p ω g k 1 ) ). By the induction hypothesis, we know that p ω g k 1 ) 0 mod 4). Therefore, to show that p ω g k ) 0 mod 4), it suffices to show that the coefficient of q g k in F q) in 5) vanishes mod 8, which is indeed true as seen in 12) in Lemma Congruences for p ν n) The formula 2) in the Introduction is the key to the results in this section. Theorem 4.1. For any nonnegative integer n, Proof. Note that p ν 8n + 6) 0 mod 4). q 2 ; q 2 ) ψq 2 ) q2 ; q 2 ) q 4 ; q 4 ) q 2 ; q 4 ) q 2 ; q 2 ) 2 q 4 ; q 4 ) q 4 ; q 4 ) 2 q 2 ; q 4 ) 2 q 4 ; q 4 ) q 4 ; q 4 ) 2 q 2 ; q 4 ) 2 q 4 ; q 4 ) mod 4) 14) q 8 ; q 8 ) 2 q 4 ; q 8 ) 2 q 2 ; q 4 ) 2 q 4 ; q 4 ) q 8 ; q 8 ) 2 q 4 ; q 8 ) 2 q 2 ; q 4 ) 2 q 4 ; q 4 ) mod 4) q 8 ; q 8 [ ) q 8 ; q 8 ) q 4 ; q 8 ) 2 ][ q 2 ; q 4 ) 2 q 4 ; q 4 ] ) ] ] q 8 ; q 8 ) [1 + 2 q [1 4n2 + 2 q 2m2 q 8 ; q 8 ) [1 + 2 q 4n2 + 2 m1 in which the terms for q 8n+6 vanish. This completes the proof. m1 ] q 2m2 mod 4). Again, by 2), we see that p ν 2n 1) p ω n). Thus the following congruences immediately follow from Theorem 1.1.

9 CONGRUENCES RELATED TO THE RAMANUJAN/WATSON MOCK THETA FUNCTIONS ωq) AND νq) 9 Theorem 4.2. For any nonnegative integers n and k, p v 2 2k+4 n + 11 ) 22k mod 4), p v 2 2k+4 n + 17 ) 22k mod 4), p v 2 2k+5 n + 8 ) 22k mod 4). 5. Waldherr s congruences In this section, we provide an elementary proof of the first explicit examples of congruences for p ω n). Theorem 5.1 Waldherr [6]). p ω 40n + 28) 0 mod 5), p ω 40n + 6) 0 mod 5). Proof. First note that the terms for q 40n+28 and q 40n+6 in F q) in 5) come from 2qωq) only. Also, since 40n n + 7) and 40n n + 9), it suffices to show that in F 0 q) the terms for q 10n+7 and q 10n+9 vanish mod 5. Since by Lemma.1, we have φq) q; q 2 ) 2 q 2 ; q 2 ) q 2 ; q 2 ) 5 q; q) 2 q 4 ; q 4 ) 2, From [2, Equation 5.11)], 1 q; q) 8 F 0 q) φq) q; q) 2 q 4 ; q 4 ) 14 q 2 ; q 2 ) 14 q 8 ; q 8 ) 4 q 2 ; q 2 ) 15 q; q) 8 q 4 ; q 4 ) q q4 ; q 4 ) 2 q 8 ; q 8 ) 4 ) 2 q 2 ; q 2 ) 10 q 8 ; q 8 ) 4 Since 10n + 7 and 10n + 9 are odd, we consider odd powers of q only in F 0 q), which appear in 8q q4 ; q 4 ) 16 q 2 ; q 2 ) 15 q 2 ; q 2 ) 24 q 4 ; q 4 ) 6 8q q4 ; q 4 ) 10 q 2 ; q 2 ) q 2 ; q 2 ) 10 8q q20 ; q 20 ) 2 q 10 ; q 10 ) 2 q nn 1) mod 5), where the congruence follows from Lemma. and the Euler s pentagonal number theorem. Now we can easily check that nn 1) 0, 2, 4 mod 10, from which it follows that the coefficients of q 10n+7 and q 10n+9 are divisible by 5. This completes the proof. Acknowledgements The fourth author was partially supported by a grant #28090) from the Simons Foundation.

10 10 GEORGE E. ANDREWS, DONNY PASSARY, JAMES A. SELLERS, AND AE JA YEE References [1] G. E. Andrews, A. Dixit, and A. J. Yee, Partitions associated with the Ramanujan/Watson mock theta functions ωq), νq) and φq), submitted for publication. [2] N. D. Baruah and B. K. Sarmah, Identities and congruences for the general partition and Ramanujan s tau functions, Indian J. Pure Appl. Math., ), [] B. C. Berndt, Ramanujan s Notebooks, PartIII, Springer-Verlag, New York, [4] N. J. Fine, Basic hypergeometric series and applications, Amer. Math. Soc., Providence, [5] S. A. Garthwaite and D. Penniston, p-adic properties of Maass forms arising from theta series, Math. Res. Lett., ), [6] M. Waldherr, On certain explicit congruences for mock theta functions, Proc. Amer. Math. Soc., ), [7] G. N. Watson, The final problem: an account of the mock theta functions, J. London Math. Soc., ), Department of Mathematics, The Pennsylvania State University, University Park, PA 16802, USA address: gea1@psu.edu address: dxp299@psu.edu address: sellersj@psu.edu address: auy2@psu.edu