A high-order modified Levenberg–Marquardt method for systems of nonlinear equations with fourth-order convergence

作者:

Highlights:

摘要

Fan (2014) presented an accelerated modified Levenberg–Marquardt method for nonlinear equations. At every iteration, the accelerated modified LM method computed not only a LM trial step, but also an additional approximate LM step which employed a line search. In this paper, based on the accelerated modified LM method, we compute the approximate LM step one more time at every iteration, and obtain a high-order accelerating modified Levenberg–Marquardt method. Under the local error bound condition which is weaker than nonsingularity, the convergence order of this new method is shown to be fourth. A globally convergence is also given by the trust region technique. Numerical results show that the new method is efficient and could save many calculations of the Jacobian.

论文关键词:Unconstrained optimization,Trust region,Systems of nonlinear equations,Levenberg–Marquardt method,Local error bound

论文评审过程:Received 11 April 2015, Revised 26 December 2015, Accepted 21 March 2016, Available online 9 April 2016, Version of Record 9 April 2016.

论文官网地址:https://doi.org/10.1016/j.amc.2016.03.031