Remarks on the Grothendieck norm

Abstract

The goal of this short note is to point out three observations around the Grothendieck norm and semidefinite programming. The first is that the Grothendieck norm captures the difficulty of relating the off-diagonal entries of a real, symmetric matrix to a probabilistic correlation, the second is that there is an interesting ``Fourier''-type duality between the Schur and Grothendieck norms of a real matrix, and the third and last centers around the difficulty of finding an efficiently computable noncommutative analog to the Grothendieck norm.