Translated sums of quasi-primitive sequences (Q6607830)

A sequence \( \mathcal{A}=(a_{i})_{i\geq 0}\) of strictly positive integers is said to be quasi-primitive if there are no three distinct terms \(a_{i},a_{j}\) and \(a_{k}\in \mathcal{A}\) such that \((a_{i},a_{j})=a_{k}.\) Erdős conjectured that the sum \(f(\mathcal{A},0) \leq f(\mathcal{Q},0),\) where \(\mathcal{Q}\) is the sequence of all powers of prime numbers and \(f(\mathcal{A},x) = \sum_{a\in \mathcal{A}}\frac{1}{a (\log a+x)}.\) \textit{P. Erdős} and \textit{Z. Zhang} [Proc. Am. Math. Soc. 117, No. 4, 891--895 (1993; Zbl 0776.11013)] gave the following conjecture: For any quasi-primitive sequence \(\mathcal{A}\), if \(\mathcal{A}\) is primitive then \( f(\mathcal{A}) \leq f(\mathcal{P}),\) where \(f(\mathcal{A})=f(\mathcal{A},0)\) and \(\mathcal{P}\) is the sequence of prime numbers, otherwise \( f(\mathcal{A}) \leq f(\mathcal{Q}).\) Erdős and Zhang showed that when the sequence \(\mathcal{ A}\) is primitive, the sum \(f(\mathcal{A})\) is bounded above by 1.84. However, \textit{J. D. Lichtman} and \textit{C. Pomerance} [Proc. Am. Math. Soc., Ser. B 6, 1--14 (2019; Zbl 1481.11026)] presented an even tighter upper bound, namely \(f(\mathcal{A}) <1.781\cdots.\) Recently the Erdős primitive set conjecture was proven by \textit{J. D. Lichtman} [Forum Math. Pi 11, Paper No. e18, 21 p. (2023; Zbl 1535.11033)]. \textit{I. Laib} et al. [C. R., Math., Acad. Sci. Paris 357, No. 5, 413--417 (2019; Zbl 1472.11052)] proved that the analogue of Erdős conjecture is false for \(x \geq 81,\) and later \textit{I. Laib} [Notes Number Theory Discrete Math. 27(3), 39--43 (2021)] improved this result to \(x \geq 60.\) \textit{J. D. Lichtman} [C. R., Math., Acad. Sci. Paris 360, 409--414 (2022; Zbl 1500.11067)] established this minimum value \(x=0.0277\cdots,\) then \textit{I. Laib} [Comptes Rendus Mathematique 362(G5), 481--486 (2024)] corrected this result to \(x=0.8035236546387282\cdots,\) where the best lower bound of \(x\) still remains an open question. Here it is shown that for \(x\) large enough there are an infinite number of primitive sequences such that \(f(\mathcal{A},x) >f(\mathcal{Q}).\) Furthermore, when \(\mathcal{A}\) is a sequence of semiprimes, the last inequality holds for any real number \(x\geq 4.92.\)