A fast robust numerical continuation solver to a two‐dimensional spectral estimation problem

This paper presents a fast algorithm to solve spectral estimation problem for two-dimensional random fields. The latter is formulated as convex optimization with the Itakura-Saito pseudodistance objective function subject constraints of moment equations. structure Hessian dual exploited in order make possible Newton solver. Then solver incorporated predictor-corrector numerical continuation method which able produce parametrized family solutions Two sets simulations are performed test and estimator. on frequency show that our estimator outperforms classical windowed periodograms case two hidden frequencies has higher resolution. other set system identification indicates more robust than Newton's alone ill-conditioned instances.