| 摘要 |
An Augmented Lagrangian Pattern Search (ALPS) algorithm that attempts to solve a non-linear optimization problem subject to non-linear, linear, and, bound constraints is discussed. The present invention utilizes information from the linear and bound constraints, formulates sub-problems using Lagrange parameter estimates and appropriate penalty parameters (using a log barrier), and provides a robust update formulae for parameters which guides the algorithm towards a minimum. The present invention solves a general non-linear optimization problem without using any slack variables to convert the inequality constraints to equality constraints or equality constraints to inequality constraints. |
