The Hadamard square of any square matrix A is bounded above and below
by some doubly stochastic matrices times the square of the largest and
the smallest singular values of A. Applications to graphs, permanents
, and eigenvalue perturbations are discussed.