A Shamanskii-like self-adaptive Levenberg–Marquardt method for nonlinear equations

Abstract

In this paper, we are concerned with the Shamanskii-like self-adaptive Levenberg–Marquardt methods for nonlinear equations. We consider two choices of Levenberg–Marquardt parameter. One of them is the standard self-adaptive Levenberg–Marquardt parameter, the other is nonmonotone self-adaptive Levenberg–Marquardt parameter by using the nonmonotone technique of Grippo, Lampariello and Lucidi. Under the error bound condition which is weaker than nonsingularity, we show that the Shamanskii-like self-adaptive Levenberg–Marquardt methods converge with Q-order (m+1). Numerical experiments show the new algorithms are efficient and promising.