Friday, October 11

8:30-9:30 AM

Kidd Island & North Cape Rooms

Chair: Åke Björck, Linkoping University, Sweden
## IP3

Sparsity, Structure, and Separability: Connections
with Large-Scale Optimization

Large-scale optimization is both a major consumer and a producer of sparse matrix techniques. This talk will focus on three main areas of interaction between these two roles, with special emphasis on recent developments: (1) solving the Newton equations in interior methods for constrained optimization; (2) exploiting structure that does not fit the classical definition of sparsity, but nonetheless allows large-scale problems to be represented and solved in a "sparse" form (much smaller than the total number of matrix elements); and (3) creating efficient linear algebraic techniques to process the structures produced by algebraic modeling languages that automatically recognize partially separable structure.
**Margaret H. Wright**

Bell Laboratories, Lucent Technologies

Back to SM96 Program-at-a-Glance

*MMD, 8/15/96*