Selection principle for pointwise bounded sequences of functions (Q2518060): Difference between revisions

Let \(\varepsilon >0\) and \(f:[a,b]\to\mathbb R\). The quantity \(N(\varepsilon, f, [a,b])\) is defined as the least upper bound in the set of indices \(n\) for which there is a family \(([a_i,b_i])_{i=1}^n\) of disjoint intervals such that \(|f(b_i)-f(a_i)|>\varepsilon\) for any \(i=1,2,\dots, n\). The main theorem reads as follows: Let \((f_j)\) be a pointwise bounded sequence of real functions on \([a,b]\) such that \(M_{\varepsilon}:=\limsup_{j\to\infty}N(\varepsilon, f_j,[a,b])<\infty\) for any \(\varepsilon>0\). Then \((f_j)\) contains a subsequence which converges everywhere on \([a,b]\) to some function \(f:[a,b]\to\mathbb R\) with \(N(\varepsilon,f,[a,b])\leq M_{\varepsilon}\) for any \(\varepsilon>0\). It is next shown that if \((f_n)\) converges uniformly on \([a,b]\) to an \(f:[a,b]\to\mathbb R\) with \(N(\varepsilon,f,[a,b])<\infty\) for any \(\varepsilon>0\), then \(\limsup_{j\to\infty}N(\varepsilon, f_j,[a,b])<\infty\) for any \(\varepsilon>0\). A similar result is also proven for almost everywhere convergence of a sequence \((f_j)\) of measurable functions on \([a,b]\). Examples are provided showing the sharpness of the above results. Finally, in the last section, the results are compared with known selection principles.