Typical behaviour of integrable functions at infinity (Q307723)

The article is motivated by \textit{E. Lesigne}'s result describing the behaviour of any summable function \(f\) on \([0,\infty)\) [Am. Math. Mon. 117, No. 2, 175--181 (2010; Zbl 1210.26006)]. Lesigne showed that for almost all \(x\in[0,\infty)\) there must be \(f(nx)\to0\) as \(n\to\infty\). This can be improved neither in connection with \(x\) (the above condition does not hold for all \(x\) even provided \(f\) is continuous), nor in connection with the rate of \(f(nx)\): for every sequence \(a_n\to\infty\) there is a \textit{continuous} summable \(f\) such that NEWLINE\[NEWLINE \limsup_{n\to\infty}\,a_nf(nx)=\infty NEWLINE\]NEWLINE holds \textit{everywhere} in \([0,\infty)\) [\textit{A.~Komisarski}, ``On the behavior of integrable functions at infinity'', Preprint, \url{arXiv:1601.01241}]. The article under review shows that such situation is typical. Namely, all pairs \((b,f)\), \(b=(b_n)_{n=1}^\infty\), such that NEWLINE\[NEWLINE \limsup_{n\to\infty}\frac{f(nx)}{b_n}=\infty, NEWLINE\]NEWLINE form comeagre subsets of \(c_0^+\times L_1[0,\infty)\) and \(c_0^+\times C[0,\infty)\); here \(c_0^+=\{b\in c_0:b_n>0\;\text{for all}\;n\}\).