An affine scaling trust-region approach to bound-constrained nonlinear systems (Q1861959): Difference between revisions

The paper presents an iterative method for numerically finding zero points of nonlinear maps which lie within a given cell. The method combines ideas from trust-region Newton methods and an affine interior scaling approach for constrained optimization. Global and local convergence properties are derived. Extensive numerical tests on a large class of test problems from the literature are reported. The present method appears to compare well with an active set-type Newton method and an inexact Gauss-Newton-type method.