| |
|
Description: |
We are interested in solving nonlinear programming problems by a Newton-like method without the need for second derivatives. The talk will start out by motivating the well-known SQP method for equality-constrained optimization and by discussing quasi-SQP variants that arise from approximating necessary parts of a second derivative matrix. Globally convergent variants and good practical performance on standard testproblems show that the methods hold much promise. Finally we point out how these techniques can be extended in a nice way to allow for the treatment of inequality constraints, at least for the case where the solution is nondegenerate and strictly complementary.
Area(s):
|
|
Date: |
|
| Start Time: |
14:30 |
|
Speaker: |
Michael Wagner (Old Dominion University, Norfolk, USA)
|
|
Place: |
Room 5.5
|
| Research Groups: |
-Numerical Analysis and Optimization
|
|
See more:
|
<Main>
|
|
