[CIG-SEISMO] Problems with running SPECFEM3D

Dimitri Komatitsch komatitsch at lma.cnrs-mrs.fr
Fri Jul 7 11:56:04 PDT 2017


Hi XiangTu,

Thanks for your message. I do not know the answer to the CUBIT/TRELIS 
question (have you tried the last one that Emanuele Casarotti committed 
to Git / devel branch last week?), but regarding MPI it is an MPI 
installation problem on your machine, unrelated to SPECFEM; your system 
administrator should be able to fix it.

Best regards,
Dimitri.

On 07/06/2017 12:39 PM, tu xiang wrote:
> Hello, everyone
> 
> I want to do fault spontaneous rupture with the SPECFEM3D software, so I 
> start my learning from running the fault examples provided by SPECFEM3D.
> First, I install the softwares required by the simulation.
> I purchased the Trelis/CUBIT 16.0 software and installed it. I 
> downloaded openMPI software and installed it.
> 1.Problems with Installation of SPECFEM3D
> Then, I need to install the SPECFEM3D software. I downloaded the 
> software from GitHub.
> I run the configure shell script as follow:./configureFC=gfortran CC=gcc 
> MPIFC=mpif90 --with-mpi
> I have successfully compiled the xcheck_mesh_quality, 
> xconvolve_source_timefunction, 
> xdecompose_mesh,xgenerate_databases,xmeshfem3D, xspecfem3D code in 
> SPECFEM3D.
> If I want to construct the mesh model of simulation with Trelis/CUBIT 
> and output in the SPECFEM3D format, I need to install GEOCUBIT python 
> script library.
> The SPECFEM3D software provide a CUBIT_GEOCUBIT directory.
> According to the README.md file in the CUBIT_GEOCUBIT directory, 
> GEOCUBIT can be install by the command `sudo python setup.py install`. 
> Thus, I run this command and restart my computer. Then I run ` import 
> cubit2specfem3d ` in the Trelis/CUBIT command line panel. But it reveals 
> that Trelis cannot find this script.
> In this case, I tried the second way of install GEOCUBIT python script 
> library, that is, added the path to my .bashrc file:
> export 
> PYTHONPATH=$PYTHONPATH:/home/soft/specfem3d/CUBIT_GEOCUBIT/geocubitlib
> export PATH=$PATH:/home/soft/specfem3d/CUBIT_GEOCUBIT/geocubitlib
> I restart my computer. Then I run ` import cubit2specfem3d ` in the 
> Trelis/CUBIT command line panel. It does not output any warnings or errors.
> 2.Problems with Running Mesh Generating Script
> So I think I have successfully installed all the required softwares.
> Next, I want to run the fault examples provided in the 
> specfem3d/EXAMPLES/fault_examples/ directory.
> I start from the tpv5 example.
> I run the mesh constructing python script ‘TPV5.py’ with Trelis/CUBIT 
> software. But it output errors:
> Trelis>
> Traceback (most recent call last):
> File "<string>", line 1, in <module>
> File 
> "/home/bei/specfem3d-master/CUBIT_GEOCUBIT/geocubitlib/cubit2specfem3d.py", 
> line 1464, in export2SPECFEM3D
> sem_mesh = mesh(hex27, cpml, cpml_size, top_absorbing)
> File 
> "/home/bei/specfem3d-master/CUBIT_GEOCUBIT/geocubitlib/cubit2specfem3d.py", 
> line 467, in __init__
> self.block_definition()
> File 
> "/home/bei/specfem3d-master/CUBIT_GEOCUBIT/geocubitlib/cubit2specfem3d.py", 
> line 554, in block_definition
> if qk < 0 or qmu < 0:
> UnboundLocalError: local variable 'qk' referenced before assignment
> Trelis>
> I have tried again and again.
> 3.Problems with mpirun
> Luckily, SPECFEM3D provided the tpv5 mesh model with the example. Thus, 
> I can continue the simulation.
> First, I run the tpv5 example with only one processor, it can run 
> successfully. But when I run this example with 6 processors, it output 
> errors.
> My operation is as follows:
> I make OUTPUT_FILES directory:
> mkdir -p OUTPUT_FILES
> I make bin directory:
> mkdir -p bin
> I link the compiled code to the bin directory:
> ln -s ../../../../bin/xdecompose_mesh
> ln -s ../../../../bin/xgenerate_databases
> ln -s ../../../../bin/xspecfem3D
> I change the NPROC parameter to 6 in Par_file file.
> I copy the Par_file, CMTSOLUTION , and STATIONS files to the OUTPUT_FILE
> I run the command 
> ‘./bin/xdecompose_mesh6./MESH./OUTPUT_FILES/DATABASES_MPI’to decompose 
> the mesh model.
> It can run successfully
> I run the command ‘pirun -np 6 ./bin/xgenerate_databa
> ses’ to generate databases.
> It output errors:
> $ mpirun -np 6 ./bin/xgenerate_databa
> ses
> --------------------------------------------------------------------------
> [[2895,1],1]: A high-performance Open MPI point-to-point messaging module
> was unable to find any relevant network interfaces:
> 
> Module: OpenFabrics (openib)
>   Host: ubuntu
> 
> Another transport will be used instead, although this may result in
> lower performance.
> --------------------------------------------------------------------------
> [ubuntu:02332] 5 more processes have sent help message 
> help-mpi-btl-base.txt / btl:no-nic
> s
> [ubuntu:02332] Set MCA parameter "orte_base_help_aggregate" to 0 to see 
> all help / error
> messages
> I do not know why.
> 
> 
> 
> _______________________________________________
> CIG-SEISMO mailing list
> CIG-SEISMO at geodynamics.org
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-seismo
> 

-- 
Dimitri Komatitsch, CNRS Research Director (DR CNRS)
Laboratory of Mechanics and Acoustics, Marseille, France
http://komatitsch.free.fr


More information about the CIG-SEISMO mailing list