On the \(\text{Ext}^2\)-problem for Hilbert spaces (Q2217513)

The authors study the second derivative of the functor Hom\((E,-)\) in the category of Banach spaces and continuous linear maps. Although this category is not abelian, the derivatives have been constructed by \textit{V. P. Palamodov} [Russ. Math. Surv. 26, No. 1, 1--64 (1972; Zbl 0247.46070); translation from Usp. Mat. Nauk 26, No. 1(157), 3--65 (1971)] using injective resolutions and they can also be conceived as Yoneda Ext-groups. The first derivative Ext\(^1(E,F)\) vanishes if every short exact sequence \(0\to F\to X\to E \to 0\) splits, and despite the simple fact that all such sequences with three Hilbert spaces split, a famous result of Enflo, Lindenstrauss and Pisier [\textit{P.~Enflo} et al. [Math. Scand. 36, 199--210 (1975; Zbl 0314.46015)] implies Ext\(^1(\ell_2,\ell_2)\neq 0\). Palamodov had asked whether the higher derivatives Ext\(^2(E,F)\) vanish for all Banach spaces. Whereas the negative answer to this question was comparatively easy, the question for \(E=F=\ell_2\) is much harder. More concretely, this problem asks whether every \(4\)-term exact sequence \(0\to \ell_2\to X\to Y\to \ell_2\to 0\) is equivalent to a trivial exact sequence or, equivalently, whether for every Banach space \(X\) containing \(\ell_2\) and satisfying Ext\(^1(\ell_2,X)=0\) also the quotient satifies Ext\(^1(\ell_2,X/\ell_2)=0\). Following an ingenious approach of \textit{N. J. Kalton} and \textit{N. T. Peck} [Trans. Am. Math. Soc. 255, 1--30 (1979; Zbl 0424.46004)], the authors mainly work in the category of quasi Banach spaces (where the triangle inequality is weakened to \(\|x+y\|\le c(\|x\|+\|y\|)\)) and construct a non-trivial \(4\)-term exact sequence as a concatenation of a classical non-splitting Kalton-Peck sequence \(\ell_2\to X\to\ell_2\) and a suitably constructed modification thereof. The main result Ext\(^2(\ell_2,\ell_2)\neq 0\) has several interesting consequences like Ext\(^2(\ell_1,\mathbb K)\neq 0\) in the category of quasi Banach spaces, which is a higher order version of Ext\(^1(\ell^1,\mathbb K)\neq 0\) (which was proved independently by \textit{N. J. Kalton} [Compos. Math. 37, 243--276 (1978; Zbl 0395.46003)] and \textit{M. Ribe} [Proc. Am. Math. Soc. 73, 351--355 (1979; Zbl 0397.46002)]).