On generalizations of a conjecture of Kang and Park (Q6175150)

Let \(\Delta_d(n):=q_d(n)-Q_d(n)\), where \(q_d(n)\) is the number of partitions of \(n\) into parts differing by at least \(d\) and \(Q_d(n)\) is the number of partitions of \(n\) into parts \(\equiv \pm 1 \pmod{d+3}\). \textit{H. L. Alder} [Bull. Am. Math. Soc. 54, 712--722 (1948; Zbl 0035.31201)] conjectured that \(\Delta_d(n)\geq 0\) for all positive \(d\) and \(n\). This conjecture generalizes Euler's partition identity (\(d=1\)), the first Rogers-Ramanujan identity (\(d=2\)), and Schur's identity (\(d=3\)). Alder's conjecture has now been proven completely through the efforts of \textit{G. E. Andrews} [Pac. J. Math. 36, 279--284 (1971; Zbl 0195.31201)], \textit{A. J. Yee} [J. Reine Angew. Math. 616, 67--88 (2008; Zbl 1221.11205)] and \textit{C. Alfres} et al. [Proc. Am. Math. Soc. 139, No. 1, 63--78 (2011; Zbl 1242.11075)]. \textit{S.-Y. Kang} and \textit{E. Y. Park} [Discrete Math. 343, No. 7, Article ID 111882, 5 p. (2020; Zbl 1440.05041)] constructed an analog of Alder's conjecture that incorporates the second Rogers-Ramanujan identity. Let \(q_d^{(a)}(n)\) be the number of partitions of \(n\) into parts \(\geq a\) differing by at least \(d\) and \(Q_d^{(a)}(n)\) be the number of partitions of \(n\) into parts \(\equiv \pm a \pmod{d+3}\), and denote \(\Delta_d^{(a)}(n):=q_d^{(a)}(n)-Q_d^{(a)}(n).\) S.-Y. Kang and E. Y. Park found that \(\Delta_d^{(2)}(n)\) is negative for some \(d\) and \(n\). By modifying \(Q_d^{(a)}(n)\) to be \(Q_d^{(a,-)}(n)\), where \(Q_d^{(a,-)}(n)\) is the number of partitions of \(n\) into parts \(\equiv \pm a \pmod{d+3}\) excluding the part \(d+3-a\), and denote \(\Delta_d^{(a,-)}(n):=q_d^{(a)}(n)-Q_d^{(a,-)}(n)\), S.-Y. Kang and E. Y. Park conjectured that \(\Delta_d^{(2,-)}(n)\) is always nonnegative for all positive \(d\) and \(n\). Kang and Park's conjecture was proven for all but finitely many \(d\) by \textit{A. L. Duncan} et al. [Res. Number Theory 7, No. 1, Paper No. 11, 26 p. (2021; Zbl 1456.05012)]. Explicitly, they proved that \(\Delta_d^{(2,-)}(n)\) is nonnegative for all \(d\geq 62\) and \(n\geq 1\). For the \(a=3\) case, they further proposed the following conjecture. Conjecture 1. For all \(d,n\geq 1\), \(\Delta_d^{(3,-)}(n)\geq 0\). For \(a\geq 4\), A. L. Duncan et al. found that \(\Delta_d^{(a,-)}(n)\geq 0\) does not hold for all positive \(d\) and \(n\). By considering the functions \(\Delta_d^{(a,-,-)}(n):=q_d^{(a)}(n)-Q_d^{(a,-,-)}(n)\), where \(Q_d^{(a,-,-)}(n)\) denotes the number of partitions of \(n\) into parts \(\equiv \pm a \pmod{d+3}\) excluding the parts \(a\) and \(d+3-a\), A. L. Duncan et al. proposed the following analog of Kang and Park's conjecture and proved infinitely many cases. Conjecture 2. For \(d\), \(a\), \(n\geq 1\) with \(1\leq a\leq d+2\), \(\Delta_d^{(a,-,-)}(n)\geq 0\). The first result of the paper under review is to confirm Conjecture 1 for \(n\geq 1\) and \(d=1,2,91,92,93\) or \(d\geq 187\), the methods of A. L. Duncan et al. are employed. By using the partition counting functions from A. J. Yee, the paper also proves Conjecture 2 for \(d,n\geq 1\) and \(a\geq 5\) such that \(\lceil\frac{d}{a}\rceil \geq 2^{a+3}-1\). Then the paper presents the following conjectured Alder-type inequality. Conjecture 3. For \(d,n\geq 12\) such that \(n\geq d+2\), \(q_d^{(1)}(n)-Q_{d-4}^{(1,-)}(n)\geq 0\). It is known that Conjecture 3 has been proven recently by \textit{L. Armstrong} et al. [Electron. J. Comb. 30, No. 2, Research Paper P2.36, 16 p. (2023; Zbl 1516.05008)] for all but finitely many cases. The paper further proves conditionally that \(\Delta_d^{(a)}(n)\geq 0\) for \(a\geq 1\), \(\lceil\frac{d}{a}\rceil\geq 12\) and \(n\geq 5\lceil\frac{d}{a}\rceil+1\) assuming Conjecture 3 holds. Moreover, with more restrictive bounds on \(n\) and \(d\), the results \(\Delta_d^{(a)}(n)\geq 0\) are proven unconditionally.