Recognizing linear structure in noisy matrices (Q556892)

A Wigner-noise is a random \(n\times n\) matrix that is symmetric, with the entries being independent mean zero, with uniformly bounded variance, and such that they are either uniformly bounded or Gaussian. The author [Linear Algebra Appl. 377, 219-240 (2004; Zbl 1042.15014)] studied the asymptotic behaviour of the eigenvalues of a symmetric block matrix perturbed by a Wigner-noise. In the paper under review, he extends the analysis to the case of blow-up matrices. Similar estimates are obtained for the perturbed eigenvalues, and for noisy weighted graphs. Cases where the blow-up matrix can be recognized from the perturbed matrix are analyzed.