Keywords: CFD, HPCN, stability analysis, parallel processing, matrices, singular value decomposition
Start Date: open / Duration: 18 months
[ participants / contact]
The partners will deal with the problem of determining when a physical system, structure or flow, is stable. Different situations will be investigated, the primary one being the study of flow originating from heat convection or concentration gradient of liquids. The goal is to obtain a clear picture of the linear stability analysis for the incompressible Navier-Stokes equations with respect to the parameters evolution.
This project is a continuation of CP94-682 and therefore builds on the results that the partners obtained in that project. The objective of CP94-682 was to design parallel solvers that construct spectral portraits of matrices, so that the spectral portrait is the depiction of the contour regions of variation of the eigenvalues with respect to a given threshold of perturbation.
The first part of the work will be devoted to testing the parallel codes that have been developed on different parallel computers.
In the second part of the work, new parallel algorithms for the singular value decomposition of dense matrices, as well as techniques for lowering the computational cost of portrait constructions will be investigated.
The third part of the work consists of a full study of the stability of a flow governed by the Navier-Stokes equation.
University of Patras, Laboratory of High Performance, GR
Russian Academy of Sciences, RU
Slovak Academy of Sciences, SK
Laboratory for Parallel and Distributed Systems, BG
Mr. Philippe Bernard
Tel: +33 2 99 84 73 38
Fax: +33 2 99 84 71 71
STABLE - CP96-237, May 1997
please address enquiries to the ESPRIT Information Desk
html version of synopsis by Nick Cook