Title:On multi-ideals and polynomial ideals of Banach spaces

Abstract:Given an operator ideal, how to consider a natural extension to the polynomial and multilinear settings? How can we evaluate if a given extension is better than other? In the last years some effort have been done in the direction of finding abstract properties that a polynomial ideal must fulfill in order to be considered an adequate extension of a given operator ideal. The interesting concepts of coherent sequences of polynomial ideals and compatibility of a polynomial ideal with a given operator ideal were recently introduced by D. Carando, V. Dimant and S. Muro in order to answer these questions. However, for an approach that is not motivated just by holomorphy, it seems to exist no reason to omit the multi-ideals from these cycle of ideas. In the present paper we revisit these notions; more precisely, we propose that these concepts are considered for a pair $(\mathcal{U}_{k},\mathcal{M}_{k})_{k=1}^{\infty}$, where $(\mathcal{U}_{k})_{k=1}^{\infty}$ is a polynomial ideal and $(\mathcal{M}_{k})_{k=1}^{\infty}$ is a multi-ideal. We were mostly inspired by the important special case of absolutely summing operators (which is a good model for testing since this operator ideal has several different related nonlinear notions). Our approach, in particular, makes the pair $(\mathcal{P}%_{k},\mathcal{L}_{k})_{k=1}^{\infty}$ (composed by the ideals of continuous $n$-homogeneous polynomials and continuous $n$-linear operators, with the $\sup$ norm) coherent and compatible with the ideal of continuous linear operators. This result, albeit strongly expected, illustrates that the new definitions seem more adequate in the context of the theory of multi-ideals and polynomial ideals, since for the case of real scalars it is surprising that the sequence $(\mathcal{P}_{k})_{k=1}^{\infty}$ is not coherent according to the original definition of Carando, Dimant and Muro.