Accessible operators on ultraproducts of Banach spaces
Félix Cabello Sánchez
TL;DR
This work builds a bridge between representing linear operators on ultraproducts of Banach spaces via nonlinear maps and the theory of twisted sums through quasilinear maps. A central result is that an accessible operator on an ultraproduct is not the ultraproduct of linear operators precisely when a corresponding short exact sequence fails to split, tying operator accessibility to the three-space problem. The paper then provides explicit, concrete constructions on sequence spaces such as $\ell_p$ (via Ribe and Kalton–Peck maps) to produce accessible functionals and endomorphisms, and develops general procedures to generate accessible operators from a single quasilinear map, including restriction/truncation techniques and considerations of ultrasummands. The discussion culminates with a detailed account of when accessible operators must be ultraproducts, links to $\mathscr{K}$-spaces, and a survey of examples in nonlocally convex and exotic spaces, offering both methodological tools and concrete instances for further exploration in ultraproduct and twisted-sum theory.
Abstract
We address a question by Henry Towsner about the possibility of representing linear operators between ultraproducts of Banach spaces by means of ultraproducts of nonlinear maps. We provide a bridge between these "accessible" operators and the theory of twisted sums through the so-called quasilinear maps. Thus, for many pairs of Banach spaces $X$ and $Y$, there is an "accessible" operator $X_U\to Y_U$ that is not the ultraproduct of a family of operators $X\to Y$ if and only if there is a short exact sequence of quasi-Banach spaces and operators $0\to Y\to Z\to X\to 0$ that does not split. We then adapt classical work by Ribe and Kalton--Peck to exhibit pretty concrete examples of accessible functionals and endomorphisms for the sequence spaces $\ell_p$. The paper is organized so that the main ideas are accessible to readers working on ultraproducts and requires only a rustic knowledge of Banach space theory.
