Andrews-Beck type congruences for overpartitions (Q2121774)

An Andrews-Beck type congruence, as referred to in the title of the paper under review, is in general an arithmetic property enjoyed not by the partition function itself, but rather by a cumulative counting function wherein each partition under consideration contributes certain weight (e.g. the number of parts) instead of \(1\) to the summation. Since Andrews's 2021 paper proving Beck's congruence conjectures, we have seen around a dozen of papers discovering or conjecturing further congruences of Andrews-Beck type. The current paper proves three such theorems for overpartitions involving two kinds of ranks: the D-rank (D for Dyson, who defined the classical rank for ordinary partitions) and the \(M_2\)-rank. More precisely, let \(\overline{NT}(b,k,n)\) (resp. \(\overline{NT2}(b,k,n)\)) denote the total number of parts in the overpartitions of \(n\) with D-rank (resp. \(M_2\)-rank) congruent to \(b\) modulo \(k\), then Theorem 1 (resp. Theorem 2) gives Andrews-Beck type (mod 4 or 8) congruence on \(\overline{NT}(b,k,n)\) (resp. \(\overline{NT2}(b,k,n)\)), while Theorem 3 establishes a mod 8 congruence between \(\overline{NT}(b,k,n)\) and \(\overline{NT2}(b,k,n)\). The proofs of these three theorems build on previous works and are quite elementary. Let \(\overline{R}(x,z,q)\) (resp. \(\overline{R2}(x,z,q)\)) be the generating function of overpartitions with respect to the number of parts recorded by the power of \(x\), and the D-rank (resp. \(M_2\)-rank) recorded by the power of \(z\). The proof of Theorem 1 begins with a series expression for \(\overline{R}(x,z,q)\) that was implicit in \textit{J. Lovejoy}'s paper [Ann. Comb. 9, No. 3, 321--334 (2005; Zbl 1114.11081)], then with standard generating function manipulation, the author derives the same expression in two ways, so as to link the weighted D-rank moment with the 2nd order D-rank moment, then further with the 2nd order symmetrized D-rank moment. All these amount to produce another expression for \(\overline{R}(x,z,q)\), which implies in particular the congruence in Theorem 1. The proof of Theorem is highly analogous with the role of \(\overline{R}(x,z,q)\) replaced by \(\overline{R2}(x,z,q)\) and the starting expression coming from the 2nd paper [Ann. Comb. 12, No. 1, 101--113 (2008; Zbl 1147.11062)] of \textit{J. Lovejoy}. The prove of Theorem 3 requires certain generalized Lambert series identities from [\textit{S. H. Chan}, Proc. Lond. Math. Soc. (3) 91, No. 3, 598--622 (2005; Zbl 1089.33012)], as well as several identities involving two classical theta functions \(\varphi(q)\) and \(\psi(q)\). Finally, it seems to me that the third congruence in Theorem 6 could be strengthened as \[ \overline{NT}(2,4,8n+i)\equiv 0 \pmod 8, for i=1,2,3,4,5,6,7. \] It should also be noted that there are two typos. On page 7 line 3, \(0\le b\le k\) should be \(0\le b <k\). On the bottom line of page 9, the \(1-q^n\) in the denominator should be \(1-q^{2n}\).