455. Comparison of Interior-point and active set quadratic programming for regularised column generation
Invited abstract in session WB-11: Interior point methods and applications - Part I, stream Interior point methods and applications.
Wednesday, 10:30-12:30Room: B100/5017
Authors (first author is the speaker)
| 1. | Wim Vanroose
|
| mathematics, U. Antwerpen | |
| 2. | Sebastian Van Thienen
|
| Mathematics, UAntwerpen |
Abstract
Regularized column generation is equivalent to a bundle method that maximizes the dual objective. In each iteration, a cut/column is added, and the solution is updated by solving a quadratic programming problem.
In this talk, we compare the performance of a warm-started interior point method, where multiple columns are added simultaneously, with a hot-started active-set QP approach, where columns are added one by one while maintaining a factorization of the KKT matrix using rank-one updates.
We discuss the challenges in high-performance numerical linear algebra and present numerical results for large-scale planning problems in the airline industry.
Keywords
- Large-scale optimization
- Computational linear algebra
Status: accepted
Back to the list of papers