Numerical Analysis in Nonlinear Least Squares Methods and Applications
The approximate greatest descent (AGD) method and a two-phase AGD method (AGDN) are proposed as new methods for a nonlinear least squares problem. Numerical experiments show that these methods outperform existing methods including the Levenberg-Marquardt method. However, the AGDN method outperforms...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
Curtin University
2017
|
| Online Access: | http://hdl.handle.net/20.500.11937/70491 |
| Summary: | The approximate greatest descent (AGD) method and a two-phase AGD method (AGDN) are proposed as new methods for a nonlinear least squares problem. Numerical experiments show that these methods outperform existing methods including the Levenberg-Marquardt method. However, the AGDN method outperforms the AGD method with a faster convergence. If the AGDN method fails due to singularity of the Hessian matrix, the AGD method should be used. |
|---|