Hey, as indicated here: https://www.mcs.anl.gov/petsc/petsc-dev/docs/manualpages/Mat/MatSTRUMPACKSetHSSMinSize.html
there is an example in src/ksp/ksp/examples/tutorials/ex52.c.html <https://www.mcs.anl.gov/petsc/petsc-dev/src/ksp/ksp/examples/tutorials/ex52.c.html> However, in my 3.7.6 distribution ex52.c does not have lines 29-31: 29: #if defined(PETSC_HAVE_STRUMPACK) 30: PetscBool <https://www.mcs.anl.gov/petsc/petsc-dev/docs/manualpages/Sys/PetscBool.html#PetscBool> flg_strumpack=PETSC_FALSE <https://www.mcs.anl.gov/petsc/petsc-dev/docs/manualpages/Sys/PETSC_FALSE.html#PETSC_FALSE>; 31: #endif On Mon, Aug 28, 2017 at 4:01 PM Matthew Knepley <[email protected]> wrote: > On Mon, Aug 28, 2017 at 9:17 AM, Toon Weyens <[email protected]> wrote: > >> Very interesting! I might try this afterwards, with H2lib. >> >> Also, I must say that the strumpack interface looks promising. What did >> you mean by "only for the sparse part", by the way? Also, any idea when it >> is going to be released in an official petsc release, preferably with some >> documentation? >> > > I thought STRUMPACK was in 3.7.6 > > Matt > > >> Thanks! >> >> On Mon, Aug 28, 2017 at 12:15 PM Jose E. Roman <[email protected]> >> wrote: >> >>> >>> > El 28 ago 2017, a las 10:15, Toon Weyens <[email protected]> escribió: >>> > >>> > Thank you Barry, that explains why I couldn't find information about >>> it. I am now going to implement this straight-forward implementation as a >>> first step. >>> > >>> > In the long term it would in any case be useful to have a solver that >>> uses H-matrices, such as H2lib. >>> > >>> > Is there any chance Petsc is thinking about moving to this kind of >>> matrices as well in the future? >>> > >>> > >>> >>> You can use H2lib in a PETSc program by wrapping it in a shell matrix. >>> We did this some time ago https://doi.org/10.1016/j.cam.2012.07.021 >>> >>> Also, PETSc has an interface to STRUMPACK, but only to the sparse part >>> and I guess what you need is the dense part. >>> >>> Jose >>> >>> > > > -- > What most experimenters take for granted before they begin their > experiments is infinitely more interesting than any results to which their > experiments lead. > -- Norbert Wiener > > http://www.caam.rice.edu/~mk51/ >
