Interpolation of Fremlin tensor products and Schur factorization of matrices (Q418711): Difference between revisions

Within finite dimensional Banach lattices the authors prove interpolation formulas for the Fremlin tensor product [\textit{D.~H.\ Fremlin}, Am.\ J.\ Math.\ 94, 777--798 (1972; Zbl 0252.46094)] and spaces of regular multilinear forms and operators. For spaces of regular operators between Minkowski spaces, their results yield (with equality of norms) the complex interpolation formula \([\mathcal{L}^r(\ell_{p_0}^n),\mathcal{L}^r(\ell_{p_1}^n)]_\theta=\mathcal{L}^r(\ell_{p_\theta}^n)\), where \(0<\theta<1\), \(1 \leq p_0,p_1 \leq \infty\) and \(1/{p_\theta}=(1-\theta)/{p_0}+\theta/{p_1}\). The special case \(p_0=1\), \(p_1=\infty\) and \(\theta=1/2\) (hence \(p_\theta=2\)) in part is due to [\textit{I.\ Schur}, J.\ Reine\ Angew.\ Math.\ 140, 1--28 (1911; JFM 42.0367.01)], and for the same values of \(p_0, p_1\) and \(0<\theta<1\) arbitrary, this result is contained in the remarkable paper [\textit{G.\ Pisier}, Arch.\ Math.\ 62, No. 3, 261--269 (1994; Zbl 0991.46007)]. In the linear case, the authors' proof is based on the atomic description of Fremlin's tensor products due to [\textit{A.~R.\ Schep}, Ill.\ J.\ Math.\ 28, 579--591 (1984; Zbl 0599.47055)], whereas in the multilinear case, variants of the Maurey-Rosenthal factorization theorem for multilinear forms on Banach lattices proven by the first named author in [Positivity\ 5, No. 2, 153--175 (2001; Zbl 0994.47036)] are used. Some of the interpolation formulas are extended to the Calderón-Lozanovskii construction. Eventually, the authors provide applications to Schur type factorizations of matrices.