Optimal average approximations for functions mapping in quasi-Banach spaces (Q2444479)

By the well known result of \textit{S. Mazur} and \textit{W. Orlicz} [Stud. Math. 10, 184--208 (1948; Zbl 0036.07801)], that a continuous function \(f: [a,b] \to X\) need not be Riemann integrable if \(X\) is a non-locally convex F-space. Even continuous Riemann integrable function may have ``bad'' properties, once \(X\) is a non-locally convex quasi-Banach space. So, in [\textit{F. Albiac} and \textit{J. L. Ansorena}, Nonlinear Anal., Theory Methods Appl., Ser. A, Theory Methods 75, No. 16, 6108--6119 (2012; Zbl 1266.46002)], the authors of the paper under review showed the existence of a continuous Riemann integrable function \(f: [a,b] \to X\) such that the average function \(\text{Ave} [f]\) defined by \[ \text{Ave} [f] (s,t) = \begin{cases} \frac{1}{t-s} \int_s^t f(u) \, du, & \text{if} \,\, a \leq s < t \leq b; \\ f(s), & \text{if} \,\, a \leq s = t \leq b; \\ \frac{1}{s-t} \int_t^s f(u) \, du, & \text{if} \,\, a \leq t < s \leq b \end{cases} \] is separately continuous and unbounded on \([a,b]^2\). It was then natural to ask, what are in general the connections between boundedness, separate and joint continuity of the average function for a continuous Riemann integrable function. The main results of the paper assert that, given a non-locally convex quasi-Banach space \(X\), there exist continuous Riemann integrable functions \(f_1, f_2, f_3: [a,b] \to X\) such that: {\parindent=6mm \begin{itemize} \item[(1)] \(\text{Ave} [f_1]\) is bounded and separately continuous but fails to be jointly continuous (Theorem 1.3); \item [(2)] \(\text{Ave} [f_2]\) is bounded but fails to be separately continuous (Theorem 1.4); \item [(3)] \(\text{Ave} [f_3]\) is neither bounded nor separately continuous (for \(X = \ell_p\) with \(0<p<1\), the latter result was proved earlier in [\textit{M. M. Popov}, Stud. Math. 110, No. 3, 205--220 (1994; Zbl 0803.46004), Theorem 2.1]) (Theorem 1.5). \end{itemize}}