作者: Helmut Jarausch , Wolfgang Mackens
DOI: 10.1007/978-3-0348-6256-1_20
关键词:
摘要: We report on the numerical computation of Solution branches (u,λ)∈ℝn×ℝ large finite dimensional nonlinear systems (1) by use a certain condensation technique, Condensed Newton with Supported Picard iteration ([4]). Making efficient given (fast) solver for linear involving fixed positive definite n × n-matrix A CNSP-approach adaptively reduces problem (1) to decoupled pair equations. One these is high but controllable by modified while other one low and may thus be treated expensive methods such as Newton’s iteration.