MATH 7664 ITERATIVE METHODS IN NUMERICAL LINEAR ALGEBRA

Spring 2002. University of Colorado at Denver

Project 3. Applications of Preconditioned Iterative Solvers in PETSc with BlockSolve95 and Hypre.


HOURS: TR 0530PM-0645PM CU-Dravo, 626.

INSTRUCTOR:
Prof. Andrew Knyazev
Office: CU (Dravo) 644. Phone: 556-8102.
Office hours: by appointment
WWW: http://www-math.cudenver.edu/~aknyazev/
Email: aknyazev@math.cudenver.edu

Read in the following matrix, pe3k.mtx.gz, stored here in the Matrix Market format and kindly provided by Chao Yang, see Large-Scale Normal Coordinate Analysis for Molecular Structures, Chao Yang, Barry W. Peyton, Donald W. Noid, Bobby G. Sumpter, Robert E. Tuzun, SIAM Journal on Scientific Computing Volume 23, Number 2 pp. 563-582. Solve the system Ax=b, where A = pe3k.mtx.gz + I, b is a vector of ones, using the zero initial guess in the PCG method with the incomplete Cholesky preconditioning for A by calling using departmental high-performance parallel Beowulf Cluster, supported by the NSF Award DMS MRI 0079719.

Run the code on one processor first and compare your results with those obtained by MATLAB's PCG preconditioned by CHOLINC. Finally, run on multiple processors and analize the scalability. The final report should be in HTML format and should include the code itself as well as discussions of comparisons with MATLAB and scalability results. In addition, there should be 30 min class presentations with live demostrations of the software runs.