Hello Ram,
> *However the problem is i need to specify EPSProblemType to be > EPS_GHEP since only this option helps to get an M orthonormal eigen > vectors. The issue is i currently don't know of an option in the > slepc wrappers to specify this input. It will be of great help if you > can help me figuring this issue out. Currently there is no built-in option to symmetry the symmetry of the matrix from the SLEPcWrappers. This is something we would like (and will) to add in the near future. However, the EPSSetProblemType can be controlled on the command line; see the Slepc documentation: http://www.grycap.upv.es/slepc/documentation/current/docs/manualpages/EPS/EPSSetProblemType.html In any case, specifying GHEP (or a different EPSProblemType) *only makes use of symmetries internally within the Slepc solver*. Not specifying this option should still give you the same eigenvectors back, since they are a result of the matrices you solve (that is, assuming you use these flags correctly). The default with the SLEPcWrappers is not to exploit symmetries, which should always work. An alternative to specify the eigenproblem symmetry type is to investigate the SLEPcWrappers::SpectralTransformation class. Hint: More often than not (for Schroedinger eigenspectrum problems), KrylovSchur with a TransformationShiftInvert works exceptionally well. Hope that helps. Best, Toby -- Toby D. Young Assistant Professor Philosophy & Physics Polish Academy of Sciences www: http://www.ippt.gov.pl/~tyoung skype: stenografia _______________________________________________ dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
