### Tuesday, July 11

## MS21

Combinatorial Algorithms in Scientific Computing

10:30 AM-12:30 PM

*Pelican *

Although scientific computing deals primarily with problems of continuous mathematics, discrete mathematics and
combinatorial algorithms play an important role in finding efficient solutions to many scientific computing problems.
Seymour Parter introduced a graph model for Gaussian elimination in 1961, and since then graph models have been
used to design efficient algorithms for sparse linear algebra. Topics from graph theory such as matchings,
colorings, and partitioning have found wide use in scientific computing. Conversely tools from continuous mathematics such as eigenvalue solvers have found use in graph partitioning and graph sequencing problems. The speakers in this
minisymposium will describe recent applications of combinatorial techniques to preconditioning, sparse orthogonal factorizations, and mesh generation.

**Organizers: Alex Pothen**

*Old Dominion University and ICASE, USA*

**Suely Oliveira**

*University of Iowa, USA*
**10:30-10:55 Robust Parallel Preconditioning via Graph Model of Incomplete Matrix Factorizations**
- David Hysom, Old Dominion University and ICASE, USA; and
* Alex Pothen*, Organizer
**11:00-11:25 Graph Embeddings and Preconditioning**
- Bruce A. Hendrickson, Sandia National Laboratories, USA
**11:30-11:55 A Parallel Mesh Improvement Algorithm**
- Mark T. Jones, Virginia Polytechnic Institute and State University, USA; and
*Paul Plassmann*, The Pennsylvania State University, USA
**12:00-12:25 An Algorithm for Predicting QR Fill-in Exactly**
- Suely Oliveira, Organizer