Failed to run the pennyShapedToughnessDominated_benchmark.xml in develop branch. about geos HOT 5 CLOSED

Hi @liyangrock, this is a known issue. It is related to a change in the linear solver strategy that has produced some undesirable results. if you change this

GEOS/src/coreComponents/linearAlgebra/interfaces/hypre/mgrStrategies/Hydrofracture.hpp

Line 95 in 5d09e66

GEOS_LAI_CHECK_ERROR( HYPRE_BoomerAMGSetMaxLevels( mgrData.coarseSolver.ptr, 25 ) );

back to 1 the problem should run fine.

Hi @CusiniM , I'm facing the same 'hypre error' as before. I have tried to modify maxLevels to 1, but it doesn't solve the problem. Do you have any other suggestions?

The attached is input and output files. Here are the execution parameters that I applied for the analysis export OMP_NUM_THREADS=2 mpirun -np 150 geos -i ./input.xml -x 3 -y 5 -z 10

input_output.zip

I don't think it is the same issue. The original error was occuring at the very beginning. This one may be related to the choice of the parameters that somehow result in a NaN solution at some point.

You are not solving the benchmark problem, are you?

from geos.

Hi @liyangrock, this is a known issue. It is related to a change in the linear solver strategy that has produced some undesirable results. if you change this

from geos.

@CusiniM Thanks for your reply.

from geos.

Hi @liyangrock, this is a known issue. It is related to a change in the linear solver strategy that has produced some undesirable results. if you change this

GEOS/src/coreComponents/linearAlgebra/interfaces/hypre/mgrStrategies/Hydrofracture.hpp

Line 95 in 5d09e66

GEOS_LAI_CHECK_ERROR( HYPRE_BoomerAMGSetMaxLevels( mgrData.coarseSolver.ptr, 25 ) );

back to 1 the problem should run fine.

Hi @CusiniM , I'm facing the same 'hypre error' as before. I have tried to modify maxLevels to 1, but it doesn't solve the problem. Do you have any other suggestions?

The attached is input and output files. Here are the execution parameters that I applied for the analysis
export OMP_NUM_THREADS=2 mpirun -np 150 geos -i ./input.xml -x 3 -y 5 -z 10

input_output.zip

from geos.

Hi @liyangrock, this is a known issue. It is related to a change in the linear solver strategy that has produced some undesirable results. if you change this

GEOS/src/coreComponents/linearAlgebra/interfaces/hypre/mgrStrategies/Hydrofracture.hpp

Line 95 in 5d09e66

GEOS_LAI_CHECK_ERROR( HYPRE_BoomerAMGSetMaxLevels( mgrData.coarseSolver.ptr, 25 ) );

back to 1 the problem should run fine.

Hi @CusiniM , I'm facing the same 'hypre error' as before. I have tried to modify maxLevels to 1, but it doesn't solve the problem. Do you have any other suggestions?
The attached is input and output files. Here are the execution parameters that I applied for the analysis export OMP_NUM_THREADS=2 mpirun -np 150 geos -i ./input.xml -x 3 -y 5 -z 10
input_output.zip

I don't think it is the same issue. The original error was occuring at the very beginning. This one may be related to the choice of the parameters that somehow result in a NaN solution at some point.

You are not solving the benchmark problem, are you?

No, I'm not working on the benchmark problem. I will check the input file and close this issue. I appreciate your help.

from geos.