Accelerating convergence of the globalized Newton method to critical solutions of nonlinear equations

In the case of singular (and possibly even nonisolated) solutions of nonlinear equations, while superlinear convergence of the Newton method cannot be guaranteed, local linear convergence from large domains of starting points still holds under certain reasonable assumptions. We consider a linesearch globalization of the Newton method, combined with extrapolation and over-relaxation accelerating techniques, aiming at a speed up of convergence to critical solutions (a certain class of singular solutions). Numerical results indicate that an acceleration is observed indeed.

[1]  Ya-Xiang Yuan,et al.  On the Quadratic Convergence of the Levenberg-Marquardt Method without Nonsingularity Assumption , 2005, Computing.

[2]  Andreas Fischer,et al.  Unit stepsize for the Newton method close to critical solutions , 2021, Math. Program..

[3]  Jorge J. Moré,et al.  Testing Unconstrained Optimization Software , 1981, TOMS.

[4]  Francisco Facchinei,et al.  An LP-Newton method: nonsmooth equations, KKT systems, and nonisolated solutions , 2013, Mathematical Programming.

[5]  Alexey F. Izmailov,et al.  Critical solutions of nonlinear equations: local attraction for Newton-type methods , 2018, Math. Program..

[6]  Alexey F. Izmailov,et al.  Stabilized SQP revisited , 2012, Math. Program..

[7]  Stephen J. Wright.  Superlinear Convergence of a Stabilized SQP Method to a Degenerate Solution , 1998, Comput. Optim. Appl..

[8]  Jorge J. Moré,et al.  Benchmarking optimization software with performance profiles , 2001, Math. Program..

[9]  Alexey F. Izmailov,et al.  Critical solutions of nonlinear equations: stability issues , 2018, Math. Program..

[10]  Andreas Fischer,et al.  Local behavior of an iterative framework for generalized equations with nonisolated solutions , 2002, Math. Program..

[11]  D. Marquardt.  An Algorithm for Least-Squares Estimation of Nonlinear Parameters , 1963 .

[12]  Dimitri P. Bertsekas,et al.  Nonlinear Programming , 1997 .

[13]  R. Schnabel,et al.  Tensor Methods for Nonlinear Equations. , 1984 .

[14]  Mikhail V. Solodov,et al.  Stabilized sequential quadratic programming for optimization and a stabilized Newton-type method for variational problems , 2010, Math. Program..

[15]  A. Griewank.  Starlike domains of convergence for Newton's method at singularities , 1980 .

[16]  M. Fukushima,et al.  On the Rate of Convergence of the Levenberg-Marquardt Method , 2001 .

[17]  Kenneth Levenberg.  A METHOD FOR THE SOLUTION OF CERTAIN NON – LINEAR PROBLEMS IN LEAST SQUARES , 1944 .

[18]  A. Griewank.  On Solving Nonlinear Equations with Simple Singularities or Nearly Singular Solutions , 1985 .

[19]  A. Griewank.  Analysis and modification of Newton's method at singularities , 1980 .

[20]  Francisco Facchinei,et al.  A family of Newton methods for nonsmooth constrained systems with nonisolated solutions , 2013, Math. Methods Oper. Res..

[21]  A. F. Izmailov,et al.  Newton-Type Methods for Optimization and Variational Problems , 2014 .

[22]  FanJin-yan,et al.  On the quadratic convergence of the Levenberg-Marquardt method without nonsingularity assumption , 2005 .