Hello, Thank you for all your information and suggestions. You're right that I was using a LAPACK-based solver. As you recommended, I first tried to use another solver other than LAPACK, and I ended up with a convergence error in SCM.
So second, I performed a mesh convergence study to see if further reduction in the DOF (degree of freedom) is possible. As a result, I can get a model with a DOF of 36426, far less than the maximum DOF available in LAPACK (46340). By this, I finally avoided the overflow error. But the thing is, the new error occurred, and the error message is ambiguous this time. The exact error message is: killed Inspecting this, I found out that the further reduction of a DOF will not cause the error. But I'd want to know if there is another way to fix this error other than reducing the number of DOF. Could you possibly give me further advice on this? Thank you, all. Best regards, Dayoung Kang 2022년 4월 8일 (금) 오후 11:11, John Peterson <jwpeter...@gmail.com>님이 작성: > > > On Fri, Apr 8, 2022 at 8:25 AM 강다영 <dydy1...@pusan.ac.kr> wrote: > >> Dear all, >> >> I have a question about the overflow error I met during the SCM training >> in >> Libmesh. >> >> For the training, I used the mesh model with a degree of freedom of >> 318,021. >> And I got an error message as below: >> >> [0]PETSC ERROR: --------------------- Error Message >> -------------------------------------------------------------- >> [0]PETSC ERROR: No support for this operation for this object type >> [0]PETSC ERROR: Product of two integer 317065 317065 overflow, you must >> ./configure PETSc with --with-64-bit-indices for the case you are running >> [0]PETSC ERROR: See https://www.mcs.anl.gov/petsc/documentation/faq.html >> for trouble shooting. >> [0]PETSC ERROR: Petsc Release Version 3.12.2, unknown >> [0]PETSC ERROR: example-opt on a arch-linux2-c-debug named >> X570-AORUS-ELITE >> by dc308-ubuntu Thu Apr 7 12:46:34 2022 >> [0]PETSC ERROR: Configure options --with-cc=mpicc --with-cxx=mpicxx >> --with-fc=mpif90 --download-fblaslapack --download-parmetis >> --download-metis --download-scalapack --download-hdf5 --download-superlu >> --download-mumps >> [0]PETSC ERROR: #1 PetscIntMultError() line 2167 in >> /home/dc308-ubuntu/Packages/petsc/include/petscsys.h >> [0]PETSC ERROR: #2 BVCreate_Svec() line 461 in >> /home/dc308-ubuntu/Packages/slepc/src/sys/classes/bv/impls/svec/svec.c >> [0]PETSC ERROR: #3 BVSetSizesFromVec() line 186 in >> /home/dc308-ubuntu/Packages/slepc/src/sys/classes/bv/interface/bvbasic.c >> [0]PETSC ERROR: #4 EPSAllocateSolution() line 570 in >> /home/dc308-ubuntu/Packages/slepc/src/eps/interface/epssetup.c >> [0]PETSC ERROR: #5 EPSSetUp_LAPACK() line 36 in >> /home/dc308-ubuntu/Packages/slepc/src/eps/impls/lapack/lapack.c >> [0]PETSC ERROR: #6 EPSSetUp() line 173 in >> /home/dc308-ubuntu/Packages/slepc/src/eps/interface/epssetup.c >> [0]PETSC ERROR: #7 EPSSolve() line 136 in >> /home/dc308-ubuntu/Packages/slepc/src/eps/interface/epssolve.c >> >> In the 3rd line of the error message, I noticed that the number "317065" >> matches with (the number of degree of freedom of model - the number of >> constrained degree of freedom of the model). >> Therefore, I decreased the degree of freedom to 7,503 to deal with this >> error. >> With this model, it was possible to finish the SCM training without an >> error. >> So I guess a large degree of freedom caused this overflow error. >> >> However, I'd still like to fix this overflow error to use the original >> model with a degree of freedom of 318,021 (for more accurate solutions). >> >> Could you possibly give some information on how to finish SCM training for >> a model with a large degree of freedom without the overflow error? >> And I'd appreciate it if you could let me know the maximum degree of >> freedom that will not cause the overflow error. >> > > Hi Dayoung, > > I'm speculating here, but it looks like you are using a LAPACK-based > (hence dense) solver, so that will limit the size of problem that you can > solve. This solver apparently needs to build a single array of size > NDOFS**2, so the largest size problem you will be able to run is likely > approximately NODFS = sqrt(INT_MAX) = sqrt(2147483647) = 46340. I don't > know exactly how to fix this issue, but it would probably involve using a > different solver at the very least. Finally, our mailing list is not really > followed by many people, so you may also have more luck by starting a > GitHub Discussion (https://github.com/libMesh/libmesh/discussions) on > this topic. > > -- > John > _______________________________________________ Libmesh-users mailing list Libmesh-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-users
