Stability properties of the class of Banach spaces in which all multilinear forms are weakly sequentially continuous (Q4531388)

A Banach space \(X\) is said to be an \(\mathcal M\)-space (resp \(\mathcal P\)-space) if every (continuous) multilinear (resp. polynomial) map on \(X\) is weakly sequentially continuous. The property is true for linear map but it is not longer true for bilinear maps, as it is shown, e.g., by the polynomial \(\|x\|^2 =\langle x,x\rangle\) on \(\ell _2\). The space \(X\) is called a hereditarily \(\mathcal M\)-space (\(\mathcal P\)-space) if every subspace of it is an \(\mathcal M\)-space (resp a \(\mathcal P\)-space). Obviously that an \(\mathcal M\)-space is \(\mathcal P\)-space, but it is unknown whether these two classes coincide. The classes of hereditarily \(\mathcal M\)-spaces and hereditarily \(\mathcal P\)-spaces agree. Banach spaces with Dunford-Pettis property are \(\mathcal M\)-spaces. A detailed study of \(\mathcal M\)-spaces has been done by the authors in [Stud. Math. 136, No. 2 , 121-145 (1999; Zbl 0948.46010)]. The present paper is concerned with stability properties of \(\mathcal M\)-spaces and \(\mathcal P\)-spaces. Among the positive stability results we mention: the sum \(E\oplus E\) of an \(\mathcal M\)-space is an \(\mathcal M\)-space (it is unknown if this property holds for \(\mathcal P\)-spaces), the quotient of an \(\mathcal M\)-space by a subspace not containing \(\ell _1\) is an \(\mathcal M\)-space, if \(E^{**}\) is an \(\mathcal M\)-space then \(E\) is an \(\mathcal M\)-space, but the converse is not true. Some negative results are: subspaces, quotients, duals or preduals of \(\mathcal M\)-spaces need not be \(\mathcal M\)-spaces, \(\mathcal M\)-spaces are not stable by products, showing that \(\mathcal M\)-space property is not a three-space property [see \textit{J. M. F. Castillo } and \textit{M. González}, Three-space problems in Banach space theory, Lect. Notes Math. 1667, Springer-Verlag (1997; Zbl 0914.46015)], the projective tensor product \(E\otimes_\pi E\) of an \(\mathcal M\)-space need not be an \(\mathcal M\)-space. The paper is written in a very clear and pleasant style.