We describe the results of a series of tests upon a class of new methods of trust region type for solving the nonlinear network optimization problem. The trust region technique considered is characterized by the use of the infinity norm and of inexact projections on the network constraints. The results are encouraging and show that this approach is particularly useful in solving large-scale nonlinear network optimization problems, especially when many bound constraints are expected to be active at the solution.
|Pages (de - à)||379-407|
|Nombre de pages||29|
|journal||SIAM Journal on Optimization|
|Numéro de publication||2|
|Etat de la publication||Publié - 1995|