Description

SPRAL provides an implementation SSIDS of the multifrontal method for the solution of linear systems with sparse symmetric indefinite system matrices. Solutions of these type of systems is the main computational kernel in optimization and finite element computations.

The aim of the project is to write a python interface to the solver SPRAL and assess the performance of SPRAL in comparison with the sequential solvers MA57 and UMFPACK.

The project is to be completed by a report and an oral presentation in the Simulation and Optimization group seminar.

Focus

• Getting acquainted the usage of FORTRAN software.
• Getting acquainted GPU and CUDA.
• Learning how to interface FORTRAN code from python.
• Demonstrating the functionality of the algorithm on a test set of problems.

The practical can serve as a preparation project for a Bachelor, Master, or Diplom thesis in the Simulation and Optimization group.

Classification of the project

This project is suited for one student as an advanced software practical. Exceptional students can ask for completion as a beginners' practical.

Requirements

• Unix, FORTRAN and python programming skills

Contact

Felix Lenders
Interdisciplinary Center for Scientific Computing (IWR)
Im Neuenheimer Feld 205
Universität Heidelberg

e-mail: felix.lenders@iwr.uni-heidelberg.de
Office: INF 205 (Mathematikon), 2/407