From: Steffen P. <ste...@tu...> - 2006-02-16 19:16:43
|
The problems you reported are not solely due to the slepc eigensolver. libMesh simply does not support the petsc/slepc uni processor version. Or did anyone get this version running together with libmesh recently? Steffen ________________________________ Von: lib...@li... [mailto:lib...@li...] Im Auftrag von Dong Xu Gesendet: Donnerstag, 16. Februar 2006 09:27 An: Ondrej Certik Cc: John Peterson; lib...@li... Betreff: Re: [Libmesh-users] Help with LibMesh Compilation It does seem to have something to do with mpi althought the error message didn't hint that. I recomplied petsc and slepc with mpi on and libmesh compilation didn't complain again at this point. Thanks! On 2/15/06, Ondrej Certik <on...@ce...> wrote: Could be the problem with mpi switched off. See http://libmesh.sourceforge.net/wiki/index.php/Installation for more info. Ondrej On Wed, Feb 15, 2006 at 11:20:12AM -0600, John Peterson wrote: > If this is still the bug you're having: > > > > I'm stuck at the final compilation of libmesh. It stoped at > > eigen_solver.C and reported the following errors. I'm using gcc 4.0.1 > > on kubuntu distro. PETSc and SLEPc are both version 2.3.0 and both > > were compiled successfully with mpi option turned off. The problem > > doesn't seem to be mpi related. Any help will be greatly appreciated. > > > > David > > > > ===================================================== > > > > Compiling C++ (in optimized mode) src/numerics/eigen_solver.C... > > /home/dxu/libmesh/include/numerics/eigen_solver.h:72: error: > > 'SLEPC_SOLVERS' was not declared in this scope > > src/numerics/eigen_solver.C: In static member function 'static > > AutoPtr<EigenSolver<T> > > > EigenSolver<T>::build(libMeshEnums::SolverPackage)': > > src/numerics/eigen_solver.C:45: error: 'SLEPC_SOLVERS' was not > > declared in this scope > > src/numerics/eigen_solver.C:47: error: expected type-specifier before > > 'SlepcEigenSolver' > > src/numerics/eigen_solver.C:47: error: expected `)' before 'SlepcEigenSolver' > > src/numerics/eigen_solver.C: In static member function 'static > > AutoPtr<EigenSolver<T> > > > EigenSolver<T>::build(libMeshEnums::SolverPackage) [with T = Number]': > > src/numerics/eigen_solver.C:68: instantiated from here > > src/numerics/eigen_solver.C:47: error: no matching function for call > > to 'AutoPtr<EigenSolver<Number> >::AutoPtr(int*)' > > /home/dxu/libmesh/include/base/auto_ptr.h:278: note: candidates are: > > AutoPtr<T>::AutoPtr(AutoPtrRef<Tp>) [with Tp = EigenSolver<Number>] > > /home/dxu/libmesh/include/base/auto_ptr.h:135: note: > > AutoPtr<T>::AutoPtr(AutoPtr<T>&) [with Tp = EigenSolver<Number>] > > /home/dxu/libmesh/include/base/auto_ptr.h:125: note: > > AutoPtr<T>::AutoPtr(Tp*) [with Tp = EigenSolver<Number>] > > make: *** [src/numerics/eigen_solver.i686-pc-linux-gnu.opt.o] Error 1 > > > then it looks like there might be a bug in the eigensolver code which > prevents it from working possibly when MPI is not installed. Is it > possible for you to install mpich? It might make your life a bit easier. > You can follow the directions on the wiki: > > http://libmesh.sourceforge.net/wiki/index.php/Installation > > I think folks have successfully installed libmesh under at least > Fedora and Debian, but probably most others should work as well. > As far as I know, we have supported every version of GCC since 2.95. > > -J > > > ------------------------------------------------------- > This SF.net email is sponsored by: Splunk Inc. Do you grep through log files > for problems? Stop! Download the new AJAX search engine that makes > searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642 > _______________________________________________ > Libmesh-users mailing list > Lib...@li... > https://lists.sourceforge.net/lists/listinfo/libmesh-users <https://lists.sourceforge.net/lists/listinfo/libmesh-users> |