Iterative algorithms for multiscale state estimation. II: Numerical investigations (Q5956969)

This paper is the continuation of Part I [ibid. 111, 501-527 (2001; Zbl 1001.93076)] in which the main elements of the theory are described. Here the authors analyze the numerical feature of the approach and investigate several iterative schemes in the context of multiscale state estimation. It is shown that the so-called simplified Uzawa scheme with Jacobi preconditioning and suitable regularization parameter is the most efficient. Comparison of different system formulations, comparison of different iterative methods, and comparison of different preconditioners are performed.