Lipschitz maps and primitives for continuous functions in quasi-Banach spaces (Q448518)

If \(X\) is a Banach space, every continuous map \({f:[a, b]\to X}\) is Riemann-integrable and the corresponding integral function \(F(t)=\int_a^t f(u)\, du\) is differentiable at every \(t\in [a, b]\) with derivative \(F'(t)=f(t)\), that is, \(F\) is a primitive of \(f\). However, when \(X\) is a nonlocally convex \(F\)-space, there exists a continuous \(X\)-valued function on \([a, b]\) failing to be integrable. \textit{M. M. Popov} investigated in [Stud. Math. 110, No. 3, 205--220 (1994; Zbl 0803.46004)] the properties of the Riemann integral for functions with values in \(F\)-spaces and formulated the following question: Does every continuous function \({f:[a, b]\to X}\) have a primitive? \textit{N. J. Kalton} [Atti Semin. Mat. Fis. Univ. Modena 44, No. 1, 113--117 (1996; Zbl 0866.46021)] provided an affirmative answer for quasi-Banach spaces \(X\) which, like the \(L_p\) spaces for \(0<p<1\), have trivial dual, but in the general case Popov's question remained unsolved. The authors of the paper under review give a negative answer. They show that, if the space \(l_p\) with \(0 < p < 1\) embeds isomorphically in a quasi-Banach space \(X\) with separating dual, then there exists an integrable continuous function \({f:[a, b]\to X}\) failing to have a primitive. Furthermore, the authors show that every nonlocally convex quasi-Banach space \(X\) with separating dual admits a continuously differentiable function \({f:[a, b]\to X}\) which is not Lipschitz.