PSPIKE Solver Project

   

Manual and Examples 1.0.0

   
A draft of the PSPIKE User Guide 1.0.0 is available. All feedback is welcome. These examples are for the latest PSPIKE version. You can request the latest PSPIKE Version 1.0.0 at the download form.

After downloading the mandatory software libraries they must be linked to your application in the following order (here shown for the Fortran example pspike_unsym.f90):
  • with GNU: mpif90 -O3 -o pspike_unsym pspike_unsym.f90 libpspike.so libfiedler.a libmc73.a libpatoh.a libpardiso.so libblas.so -fopenmp
  • with Intel: mpiifort -O3 -o pspike_unsym pspike_unsym.f90 libpspike.so libfiedler.a libmc73.a libpatoh.a libpardiso.so libblas.so -openmp

   

Mandatory Software Libraries

   
Download following software libraries and link them together with the PSPIKE library.
   

Interfaces and Option Files

   
Put following files into the same directory and control the behavior of
  • PSPIKE via the file pspike.opt.
  • Mondriaan via the file Mondriaan.options; you do not need to download the partitioner since a specific version of Mondriaan is already built-in the shared library.
You can choose PSPIKE as a linear solver in the open-source nonlinear optimizer Ipopt when choosing the parallel inexact variant of the framework. If you set following options in the option file ipopt.opt, PSPIKE is invoked.
   

Current Available Libraries Version 1.0.0

   
Architecture X86-64, 64-bit, Intel MPI 4.0 + Intel 10.1 Linux libpspike100-INTEL101-X86-64.so
Architecture X86-64, 64-bit, Intel MPI 4.0 + Intel 12.1.3 Linux libpspike100-INTEL121-X86-64.so
Architecture X86-64, 64-bit, MPICH2 1.4.1 + GNU 4.6.3 Linux libpspike100-GNU463-X86-64.so
Architecture X86-32, 32-bit, MPICH2 1.4.1 + GNU 4.6.3 Linux libpspike100-GNU463-X86-32.so

Other libraries can be compiled upon request.

   

References

   
Please cite the following references when using PSPIKE in your application:
  • M. Sathe, O. Schenk, B. Uçar, and A. H. Sameh, Towards A Scalable Hybrid Linear Solver Based On Combinatorial Algorithms. Book Chapter in Combinatorial Scientific Computing, Edited by U. Naumann and O. Schenk, Chapman-Hall CRC Computational Science, pp. 96-127, January 2012. bibtex
  • M. Manguoğlu, A. H. Sameh, and O. Schenk, PSPIKE: A Parallel Hybrid Sparse Linear System Solver, Lecture Notes in Computer Science, Volume 5704/2009, pp.797-808, 2008. bibtex
  • O. Schenk and K. Gärtner, Solving Unsymmetric Sparse Systems of Linear Equations with PARDISO, Journal of Future Generation Computer Systems, 20(3):475-487, 2004. bibtex