Parity of the coefficients of certain eta-quotients (Q2668923)

This is a fascinating and beautifully written paper on some very classical (and still very mysterious!) aspects of modular forms. For integers \(j\geq 1\), let \(f_j := (1-q^j)(1-q^{2j})(1-q^{3j})\cdots \in \mathbb{Z}[[q]]\). Consider ratios \(F(q)\) of the form \((f_{a_1}\cdots f_{a_m})/(f_{b_1}\cdots f_{b_n})\), with \(a_1,\dots,a_m\) and \(b_1,\dots,b_n\) being arbitrary sequences of positive integers. Say \(F(q)\) has odd density \(\delta\) if its odd coefficients occur with density exactly \(\delta\). Say \(F\bmod{2}\) is lacunary if it \(F\) has odd density \(0\). [\textit{T. R. Parkin} and \textit{D. Shanks}, Math. Comput. 21, 466--480 (1967; Zbl 0149.28501)] conjectured that \(1/f_1\), the generating series of the partition function, has odd density \(1/2\). The authors make a vast novel generalization (Conjecture~4) to all ratios \(F(q)\) above, conjecturing that (i) the odd density \(\delta_F\) always exists, and satisfies \(\delta_F\leq 1/2\); (ii) if \(\delta_F = 1/2\), then the coefficients of \(F\) must have odd density \(1/2\) when restricted to any nontrivial polynomial sequence (e.g.,~any non-constant arithmetic progression); and (iii) if \(\delta_F < 1/2\) (e.g.,~if \(F\bmod{2}\) is lacunary), then the coefficients of \(F\bmod{2}\) must vanish identically along some non-constant arithmetic progression. The conjecture highlights the significance of even progressions (i.e.,~non-constant arithmetic progressions along which the coefficients of \(F\bmod{2}\) vanish). It is thus of interest to try to find even progressions for various ratios \(F\), or to conjecture their nonexistence otherwise. This, among other things, is what the present authors, together with previous literature, do for the generating series \(B_m(q) := f_m/f_1\) for \(m\)-regular partitions (partitions with no parts divisible by \(m\)), for positive integers \(m\leq 28\). We refer the reader to p.~278 (especially the table within), and to \S2, for a description of the authors' very substantial contributions to this task; the authors prove (and give many further conjectures on) many results on lacunarity, self-similarity, and congruences. The tools used include a wide variety of identities (e.g.,~the following: Euler's Pentagonal Number Theorem, for \(f_1\bmod{2}\); the Frobenius congruence \(f_j^2\equiv f_{2j}\bmod{2}\); expansions, from [\textit{N. Robbins}, Fibonacci Q. 38, No. 1, 39--48 (2000; Zbl 0947.11030)] and [\textit{M. D. Hirschhorn}, Ramanujan J. 4, No. 2, 129--135 (2000; Zbl 0988.11049)], of the \(t\)-core series \(f_t^t/f_1\bmod{2}\); and certain dissection identities from [\textit{S. D. Judge} et al., Ann. Comb. 22, No. 3, 583--600 (2018; Zbl 1440.11195)]); auxiliary modular forms and Hecke operators; the Sturm bound for congruences between modular forms; and the lacunarity of values of positive-definite binary quadratic forms (which, in a quantitative form, goes back to Landau). Finally, it is worth quoting the following part of the abstract: ``All of our work is consistent with a new, overarching conjecture that we present for arbitrary eta-quotients, greatly extending Parkin-Shanks' classical conjecture for the partition function. We pose several other open questions throughout the paper, and conclude by suggesting a list of specific research directions for future investigations in this area.''