CONGRUENCES FOR BROKEN k-diamond PARTITIONS

Transcription

1 CONGRUENCES FOR BROKEN k-diamond PARTITIONS MARIE JAMESON Abstract. We prove two conjectures of Paule and Radu from their recent paper on broken k-diamond partitions. 1. Introduction and Statement of Results In [1], Paule and Andrews constructed a class of directed graphs called broken k-diamonds, and defined k n) to be the number of broken k-diamond partitions of n. They noted that the generating function for k n) is essentially a modular form. More precisely, if k 1, then 1.1) k n)q n k+1)/12 η2z)η2k + 1)z) = q ηz) 3 η4k + 2)z), where q := e 2πiz and ηz) is Dedekind s eta function ηz) = q 1/24 1 q n ). One can show various congruences for k n) for n in certain arithmetic progressions. For example, Xiong [4] proved congruences for 3 n) and 5 n) which had been conjectured by Paule and Radu in [3]. In particular, he showed that 1.2) 1 q n ) 4 1 q 2n ) n + 5)q n mod 7) In this note, we prove the remaining two conjectures in [3]. First, we use 1.2) to prove the following statement which is denoted Conjecture 3.2 in [3]). Theorem 1.1. For all n N, we have that n + 82) n + 229) n + 278) n + 327) 0 mod 7). Now, recall that the weight k Eisenstein series where k 4 is even) are given by E k z) := 1 2k σ k n)q n, B k where B k is the kth Bernoulli number, and σ k n) := d n dk. Also define 1.3) cn)q n := E 4 2z) 1 q n ) 8 1 q 2n ) 2 = q /2 E 4 2z)ηz) 8 η2z) Mathematics Subject Classification. 05A15, 05A17, 11P83. 1

2 2 MARIE JAMESON The coefficients cn) are of interest here because they are related to broken k-diamond partitions in the following way as conjectured in [3] and proved in [4]): 1.4) cn) n + 6) mod 11). Here we prove the last remaining conjecture of Paule and Radu which is Conjecture 3.4 of [3]). More precisely, we have the following theorem. Theorem 1.2. For every prime p 1 mod 4), there exists an integer yp) such that c pn + p 1 ) ) n p 1)/2 + p 8 c = yp)cn) 2 p for all n N. Remark 1. Theorem 1.2 follows from a more technical result see Theorem 3.1 which is proved in Section 3). Remark 2. As noted in [3], one can combine 1.4) with Theorem 1.2 to see that for every prime p 1 mod 4) and n N we have 5 11n + 6)p p 1 ) 11n p p 1 ) yp) 5 11n + 6) mod 11). 2 p 2p To prove Theorems 1.1 and 1.2, we make use of the theory of modular forms. In particular, we shall make use of the U-operator, Hecke operators, the theory of twists, and a theorem of Sturm. These results are described in [2]. We shall freely assume standard definitions and notation which may be found there. 2. Proof of Theorem 1.1 First we consider the form η3z) 4 η6z) 6. By Theorems 1.64 and 1.65 in [2], we have that η3z) 4 η6z) 6 S 5 Γ0 72), )). Note from 1.2) that η3z) 4 η6z) n + 5)q 3n+2 mod 7). It follows that fz) := η3z) 4 η6z) 6 U n + 33)q 3n+2 mod 7). Here, U d denotes Atkin s U-operator, which is defined by an)q n U d = adn)q n for d a positive integer. By the theory of the U-operator see Proposition 2.22 and Remark 2.23 in [2]), it follows that fz) S 5 Γ0 504), )). Now if we define bn) by bn)q n := fz), then our goal is to show that b21n + 5) b21n + 14) b21n + 17) b21n + 20) 0 mod 7).

3 CONGRUENCES FOR BROKEN k-diamond PARTITIONS 3 In order to prove the desired congruence, consider the Dirichlet character ψ defined by ψ) := 7). We may consider the ψ-twist of f, which is given by f ψ z) := ψn)bn)q n. By Proposition 2.8 of [2], we have that f ψ z) S 5 Γ ), )). Then consider )) )) n fz) f ψ z) = 1 bn)q n S 5 Γ ),. 7 In fact, fz) f ψ z) 0 mod 7). This follows from a theorem of Sturm see Theorem 2.58 in [2]), which states that fz) f ψ z) 0 mod 7) if its first coefficients are 0 mod 7) which was verified using Maple). Thus we have that for all n, and thus 1 n 7 )) bn) 0 mod 7) b21n + 5) b21n + 14) b21n + 17) b21n + 20) 0 mod 7) for all n N, as desired. 3. Proof of Theorem Preliminaries. Let us first recall the Hecke operators and their properties. If fz) = an)qn M k Γ 0 N), χ) and p is prime, the Hecke operator T p,k,χ or simply T p if the weight and character are known from context) is defined by fz) T p := apn) + χp)p k an/p) ) q n, where we set an/p) = 0 if p n. It is important to note that fz) T p M k Γ 0 N), χ). In order to prove the final statement of Theorem 1.2, define )) gz) = c 0 n)q n := E 4 4z)η2z) 8 η4z) 2 S 9 Γ 0 16), and note that cn) = c 0 2n + 1). Thus we wish to show that for every prime p 1 mod 4) there exists an integer yp) such that ) 2n + 1 c 0 p2n + 1)) + p 8 c 0 = yp)c 0 2n + 1) p for all n N. By summing and noting that c 0 n) = 0 when n is even) we see that this is equivalent to the statement that gz) T p = yp)gz). That is, we need only show that gz) is an eigenform of the Hecke operator T p for all p 1 mod 4).

4 4 MARIE JAMESON To see this, we let F be the weight 2 Eisenstein series see 1.18) of [2]) given by F z) := η4z)8 η2z) = σ 4 1 2n + 1)q 2n+1 M 2 Γ 0 4)), let θ 0 z) be the theta-function given by θ 0 z) := n= q n2 M 1/2 Γ 0 4)), and let hz) be the normalized cusp form )) hz) := η4z) 6 = an)q n = q 6q 5 + 9q 9 + S 3 Γ 0 16),. Then hz) is a modular form with complex multiplication, and for primes p we have see Section of [2]) { 2x 2 2y 2 p = x 2 + y 2 with x, y Z and x odd ap) = 0 p 2, 3 mod 4). Then we may define f 1, f 2, f S 9 Γ0 16), )) by f 1 z) = d 1 n)q n := E 4 4z)F z) [ 4θ04z) 6 θ02z) 6 + 4θ02z)θ 4 04z) 2 6θ02z)θ 2 04z) ] 4 f 2 z) = fz) = d 2 n)q n := E 4 4z)F 2z)hz) dn)q n := f 1 z) + 8i 3f 2 z). We prove the following theorem involving these forms. Theorem 3.1. The forms fz) and fz) are eigenforms of the Hecke operator T p for all primes p. Furthermore we have that T g = f, f, where T g is the subspace of S 9 Γ0 16), )) spanned by g together with g Tp for all primes p. Proof. First note that f and f are eigenforms of the Hecke operator T p for all primes p. To see this, note that there is a basis of Hecke eigenforms of the space S 9 Γ0 16), )). Also, both f and f are eigenforms of T 5 with eigenvalue 258, and one can compute that this eigenspace ker T 5 258) is 2-dimensional this can be done, for example, by computing the characteristic polynomial of T 5 using Sage). Finally, both f and f are eigenforms of the Hecke operator T 7, and they have different eigenvalues.

5 CONGRUENCES FOR BROKEN k-diamond PARTITIONS 5 Now, note that 1 g = 2 + i ) 1 2 f i ) 2 f 3 and thus T g is a two-dimensional subspace of f, f. Thus T g = f, f, as desired Proof of Theorem 1.2. Suppose p is a prime with p 1 mod 4). Then we need only check that f and f are eigenforms of T p with the same eigenvalue. Since these eigenvalues are the coefficients of q p in the expansions of f and f see Proposition 2.6 of [2]), we need only show that dp) = dp), i.e., dp) R. Now, note that the coefficients of E 4 4z) are only supported on indices that are congruent to 0 mod 4 by construction. Also, the coefficients of F 2z) are supported on indices which are 2 mod 4), and the coefficients of hz) are supported on indices which are 1 mod 4). Thus the coefficients of f 2 are only supported on indices that are congruent to 3 mod 4, so we have that d 2 p) = 0, and thus dp) = d 1 p) R, as desired. Acknowledgements The author thanks Peter Paule and Silviu Radu for suggesting these problems and mentioning Xinhua Xiong s work, and also thanks Ken Ono and the referees for their helpful comments on earlier versions of this work. References [1] George E. Andrews and Peter Paule. MacMahon s partition analysis. XI. Broken diamonds and modular forms. Acta Arith., 1263): , [2] Ken Ono. The web of modularity: arithmetic of the coefficients of modular forms and q-series, volume 102 of CBMS Regional Conference Series in Mathematics. Published for the Conference Board of the Mathematical Sciences, Washington, DC, [3] Peter Paule and Silviu Radu. Infinite families of strange partition congruences for broken 2-diamonds. Ramanujan J., 231-3): , [4] Xinhua Xiong. Two congruences involving Andrews-Paule s broken 3-diamond partitions and 5-diamond partitions. Proc. Japan Acad. Ser. A Math. Sci., 875):65 68, Department of Mathematics and Computer Science, Emory University, Atlanta, Georgia address: mjames7@emory.edu