Article ID: | iaor2013602 |
Volume: | 62 |
Issue: | 1 |
Start Page Number: | 149 |
End Page Number: | 162 |
Publication Date: | Jan 2013 |
Journal: | Numerical Algorithms |
Authors: | Cheng Wanyou, Chen Zixin |
Keywords: | programming: nonlinear |
In this paper, by the use of the residual vector and an approximation to the steepest descent direction of the norm function, we develop a norm descent spectral method for solving symmetric nonlinear equations. The method based on the nomonotone line search techniques is showed to be globally convergent. A specific implementation of the method is given which exploits the recently developed cyclic Barzilai–Borwein (CBB) for unconstrained optimization. Preliminary numerical results indicate that the method is promising.