Hi, in the SLEPc EPS documentation [1] you can see that the initial vectors have to be set explicitly, otherwise the initial vector is initialized with random values. You have to retrieve the EPS object somehow and set the initial starting vector manually, as there is no method implemented for that, yet. (No call to EPSSetInitialSpace in the libmesh sources)
[1] http://slepc.upv.es/documentation/current/docs/manualpages/EPS/EPSSetInitialSpace.html#EPSSetInitialSpace On Fri, Oct 21, 2016 at 8:47 AM, Hubert Weissmann <[email protected]> wrote: > Dear all, > > I wander whether one can set an initial guess to a (generalised) > eigenproblem in LibMesh. > > In the mailing list I found this question posed twice, where Roy said he > 'believes' that System::solution is used. > > However, I didn't any difference in performance whether I set > System::solution somehow before solving or not (which may be due to a > bad guess of course). However, I also didn't find any reference to it in > the EigenSolver class (except in get_eigenpair() function where it is > copied in the other direction). > > So my question is: Did I just seek in the wrong place or misunderstand > the interface there or is it not possible to set an initial guess to > slepc at this point? > > Any help is appreciated, > > Best, > > Hubert > > > > ------------------------------------------------------------------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, SlashDot.org! http://sdm.link/slashdot > _______________________________________________ > Libmesh-users mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/libmesh-users ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot _______________________________________________ Libmesh-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/libmesh-users
