The norm of sums of independent noncommutative random variables in \(L_{p}(\ell_1)\) (Q1779207): Difference between revisions

This paper deals with \(L_p\)-norm estimates for series of independent noncommutative random variables. The independence here is taken in the tensor product sense. The main result is the noncommutative version of the following inequality. Let \((f_k)\) be a finite sequence of (usual) independent random variables on a probability space and \(1\leq p<\infty\). Then \[ \big\| \sum| f_k| \,\big\| _p\approx \sum\| f_k\| _1 + \big(\sum\| f_k\| _p^p\big)^{1/p}\,. \] The formulation of the noncommutative version involves asymmetric noncommutative \(L_p\)-spaces, a subject studied in detail in a subsequent work by the same authors. As one can imagine, the arguments are of quantum probabilistic nature, and employ in an essential way column and row subspaces through the Haagerup tensor product. The authors then apply this result to obtain complete embeddings of the Schatten class \(S_q^n\) into the vector-valued Schatten class \(S_p[\ell_q^m]\) for \(1\leq q\leq p<\infty\). As a further application, they determined the sharp operator space type and cotype of \(L_p(0,1)\), which are \(\min(p,p')\) and \(\max(p,p')\), respectively, a result in strong contrast with the Banach space setting.