Linear support for the prime number sequence and the first and second Hardy-Littlewood conjectures (Q6568010)

Let \(\sigma(n)\) be the sum of the positive divisors of \(n\). A number \(n\) is said to be perfect if \(\sigma(n)=2n\) and is said to be abundant if \(\sigma(n)>2n\). The question of whether any odd number is perfect is an old open problem, dating back to at least Descartes. In the last two centuries, many results have been proven giving inequalities that an odd perfect must satisfy. The paper under review considers bounds relating the smallest prime factor of an odd perfect number to its total number of prime factors.\N\N\textit{Cl. Servais} [Mathesis 8, 92--93, 135 (1888; JFM 20.0174.01)] proved that if \(n\) is an odd perfect number expressed as \(n=p_1^{a_1}p_2^{a_2} \cdots p_k^{a_t}\) where \(p_1, p_2, \dots p_t\) are distinct primes with \(p_1 < p_2 \cdots < p_t \), then \N\[\Np_1 \leq t.\N\]\NThis result was improved by \textit{O. Grün} [Math. Z. 55, 353--354 (1952; Zbl 0046.27107)] who showed\N\[\Np_1 < \frac{2}{3}t +2.\N\]\NThe author of this review in a later paper [Integers 21, Paper A76, 55 p. (2021; Zbl 1487.11007)] proved that \N\[\Np_1 \leq \frac{1}{2}t -1.\N\]\NA similar, tighter inequality due to Andrew Stone will be published soon. While results of this sort are frequently phrased just for odd perfect numbers, they also apply to odd numbers \(n\) where \(\sigma(n)>2n\) (called abundant numbers). In the remainder of this review we will assume that \(n\) is an odd perfect number or an odd abundant number with smallest prime factor \(p_1\).\N\NThe starting point for all these results is as follows. Let \(H(n) = \prod_{p|n} \frac{p}{p-1}\) where \(p\) ranges over the prime divisors of \(n\). For any \(n\), we have\N\[\N\frac{\sigma(n)}{n} \leq H(n), \tag{1}\N\]\Nwith equality if and only if \(n=1\). Thus, if \(n\) is perfect or abundant, then Inequality (1) implies that \(2 < H(n)\). We can then estimate how many primes are needed to allow \(H(n) >2\) if the smallest prime of \(n\) is fixed.\N\NIn 1979, \textit{G. L. Cohen} and \textit{M. D. Hendy} [Math. Chron. 9, 120--136 (1980; Zbl 0435.10007)] proved that for any positive integer \(m\), there exists a \(b_m\) such that\N\[\Np_1 < \frac{m}{2^m -1}t + b_m. \tag{2}\N\]\NCohen and Hendy showed that one may take \(b_4=12\) and \(b_5= 35\).\N\NThe product defining \(H(n)\) is closely related to the product in Mertens' theorem, which was used by \textit{K. K. Norton} [Acta Arith. 6, 365--374 (1961; Zbl 0102.03203)] to give estimates which are better than linear, and in a certain sense asymptotically best. A slightly weaker but similar result was proved in 1953 by \textit{H. Salié} [Math. Nachr. 9, 217--220 (1953; Zbl 0050.04201)]. The existence of the \(b_i\) also follows from Norton's better than linear bounds. Norton's results were slightly tightened in the earlier mentioned paper by the author of this review, but interest in linear versions of these inequalities continues.\N\NThe paper under review gives a new proof of Cohen and Hendy's results, as well as simplified proofs that one may take \(b_4=12\) and \(b_5= 35\). They also obtain results for other values of \(b_m\) and tighten some of the results of Cohen and Hendy. For example, they show that one may take \(b_4=8\), and \(b_5=21\). They also show that some of their results are in a certain sense optimal. Although they do not phrase the results this manner, a natural way of interpreting a choice of \(b_m\) as optimal is that it is optimal given that one wants the result to apply not just to odd perfect numbers (which likely do not exist) but also odd abundant numbers (which do exist).\N\NThey also prove related results which are dependent on specific conjectures. One curious aspect is that some of their results depend on first Hardy-Littlewood conjecture, but others on the second Hardy-Littlewood conjecture. The authors do note that these two conjectures are known to be incompatible, and they are careful not to use both conjectures simultaneously at any point. The author of this review suspects that all their results which are implied by either conjecture are in fact simply true independent of any conjectures.\N\NThe literature contains some results relating \(t\) to \(p_i\) for other small values of \(i\). See for example results by \textit{M. Kishore} [Math. Comput. 36, 583--586 (1981; Zbl 0472.10007)] and the reviewer [Integers 23, Paper A13, 18 p. (2023; Zbl 1520.11008)]. It is plausible that the methods used in this paper can tighten those results.