On Computing Inverse Entries of a Sparse Matrix in an Out-of-Core Environment

Decanato - Facoltà di scienze informatiche

Data d'inizio: 12 Marzo 2013

Data di fine: 13 Marzo 2013

The Faculty of Informatics is pleased to announce a seminar given by Bora Uçar

DATE: Tuesday, March 12th, 2013
PLACE: USI Università della Svizzera italiana, room SI 008, Informatics building (Via G. Buffi 13)
TIME: 16.30

Given a factorization of the sparse matrix A held in out-of-core storage, we show how to compute a specified set of entries of the inverse of A efficiently.
When there are many inverse entries to compute, we need to guarantee that the overall computation scheme has reasonable memory requirements, while minimizing the volume of communication (data transferred) between disk and main memory. This leads to an NP-complete partitioning problem. In this talk, we will investigate this problem and practical algorithmic solutions. We will also illustrate the performance of our algorithms in practice using the MUMPS software package on a set of real-life problems.

This is joint work with P. R. Amestoy, I. S. Duff, J.-Y. L'Excellent, Y. Robert, and F.-H. Rouet.

Bora Uçar received the PhD degree (2005) in Computer Engineering from Bilkent University, Ankara, Turkey.
He has worked as post-doctoral researcher at the Mathematics and Computer Science Department of Emory University, Atlanta, USA and at the Parallel Algorithms Project, CERFACS, Toulouse, France.
Since January 2009, he is conducting research activities at ENS-Lyon, France as a CNRS research scientist.
His research interests include combinatorial scientific computing, parallel and high performance computing, and sparse matrix computations.

HOST: Prof. Olaf Schenk