topopt / topopt_in_petsc

Topology optimization for linear elastic minimum compliance with volume constraints on cartesian grids in 3D using PETSc.

TopOpt_in_PETSc

A 3D large-scale topology optimization code using PETSc

The code (or framework) presented on this page is a fully parallel framework for conducting very large scale topology optimziation on structured grids. For more details see www.topopt.dtu.dk/PETSc.

Updated and refactored to remove dependence of TopOpt.cc/h in all other classe, June, 2019, Niels Aage

To clone repository:

git clone https://github.com/topopt/TopOpt_in_PETSc.git

NOTE: The code requires PETSc version 3.11.0 or newer ! Also note that the code is not tested against the development branch on git.

This code has been tested on:

• Linux systems including: Ubuntu 18.04, Red hat enterprise linux 8

This code requires the following external software to work:

• PETSc version 3.11.4 or earlier (though never than 3.8.x)
• Requires LAPACK/BLAS
• Requires MPI

Compile following rules in makefile_ref

Normal compilation time of framework, e.g. 4s: "make topopt -j"

Run the base example by typing e.g.: "mpirun -np 4 ./topopt"

Postprocess results using Python 2.6: "bin2vtu #" where # refers to the iteration number

Visualize using ParaView (version 5.7 or earlier)

The expected result of the base code is the (but on a coarse mesh!) cantilever beam from: Aage, N., Andreassen, E., & Lazarov, B. S. (2015). Topology optimization using PETSc: An easy-to-use, fully parallel, open source topology optimization framework. Structural and Multidisciplinary Optimization, 51(3), 565–572. https://doi.org/10.1007/s00158-014-1157-0

Extensions:

An extension of the code including manufacturing filters/constraints can be found here: https://github.com/edofersan/MaximumSize_on_TopOpt_in_PETSc

topopt_in_petsc's Issues

Inlcude problem PETSc 12.xx

The latest build of PETSc requires the inclusion of headers to be changed to "" from <> for all local files.

Restart files are not written to '-workdir'

Currently the code dumps restart files to the "run" directory. On certain clusters this is not appropriate and should be fixed.

Restart from previous design - without resuming MMA

Add bool and input parameter to TopOpt.cc/h to indicate whether or not the restart should include resume in MMA.

Retrieve vtu file with bin2vtu results in 'Could not read header format'

I cannot retrieve the vtu files from the output.dat file. I tried for several iterations. E.g.:
python2.6 bin2vtu.py 440

Results in:

iter: 440 
rm: cannot remove 'output*.vtu': No such file or directory 
Reading in mesh information 
Could not read header format... exiting

Any suggestions?

xold in TopOpt.h/cc

The xold vector in the TopOpt class should be removed, since it is a leftover from a previous version with an OC design update.

Fix: Remove from TopOpt.h/cc and modify the DesignChange method in MMA.h/cc

What does ux、x stand for?

I ran the sample program and found the variables ux, uy, and uz in the paraview. I want to know what this represents

PETSc 3.11.3 error: cannot convert ‘PetscBool’ to ‘PCMGGalerkinType’ for argument ‘2’ to ‘PetscErrorCode PCMGSetGalerkin(PC, PCMGGalerkinType)’

Hi all,
I am using the newest version of PETSc for my work.
However I got an error while compiling with a newer version of PETSc3.11.3 as following:
"LinearElasticity.cc:1811:32: error: cannot convert ‘PetscBool’ to ‘PCMGGalerkinType’ for argument ‘2’ to ‘PetscErrorCode PCMGSetGalerkin(PC, PCMGGalerkinType)’
PCMGSetGalerkin(pc,PETSC_TRUE);"
Can you help me to solve this problem?

Thank you very much,
Best Regards,
Tran Quang Dat.

PETSc 3.7.5 error: cannot convert ‘PetscBool’ to ‘PCMGGalerkinType’ in PCMGSetGalerkin(PC, PCMGGalerkinType)

Hi,

It seems that there is an issue while compiling with a newer version of PETSc:

/TopOpt_in_PETSc/LinearElasticity.cc:613:32: error: cannot convert ‘PetscBool’ to ‘PCMGGalerkinType’ for argument ‘2’ to ‘PetscErrorCode PCMGSetGalerkin(PC, PCMGGalerkinType)’ PCMGSetGalerkin(pc,PETSC_TRUE);

PETSC ERROR: Caught signal number 11 SEGV: Segmentation Violation, probably memory access out of range

Hi all,

I am using PETSc 3.7.3 for my problem.
I got an error when my program ran some iterations:
[0]PETSC ERROR: ------------------------------------------------------------------------
[0]PETSC ERROR: Caught signal number 11 SEGV: Segmentation Violation, probably memory access out of range
[0]PETSC ERROR: Try option -start_in_debugger or -on_error_attach_debugger
[0]PETSC ERROR: or see http://www.mcs.anl.gov/petsc/documentation/faq.html#valgrind
[0]PETSC ERROR: or try http://valgrind.org on GNU/linux and Apple Mac OS X to find memory corruption errors
[0]PETSC ERROR: likely location of problem given in stack below
[0]PETSC ERROR: --------------------- Stack Frames ------------------------------------
[0]PETSC ERROR: Note: The EXACT line numbers in the stack are not available,
[0]PETSC ERROR: INSTEAD the line number of the start of the function
[0]PETSC ERROR: is given.
[0]PETSC ERROR: --------------------- Error Message --------------------------------------------------------------
[0]PETSC ERROR: Signal received
[0]PETSC ERROR: See http://www.mcs.anl.gov/petsc/documentation/faq.html for trouble shooting.
[0]PETSC ERROR: Petsc Release Version 3.7.7, Sep, 25, 2017
[0]PETSC ERROR: ./topopt on a arch-linux2-c-debug named tranquangdat-System-Product-Name by tranquangdat Sat Aug 17 11:31:46 2019
[0]PETSC ERROR: Configure options --with-cc=gcc --with-cxx=g++ --download-mpich –download-f2cblaslapack=1
[0]PETSC ERROR: #1 User provided function() line 0 in unknown file
application called MPI_Abort(MPI_COMM_WORLD, 59) - process 0
[cli_0]: aborting job:
application called MPI_Abort(MPI_COMM_WORLD, 59) - process 0

Could you please help me to fix this problem?

Thank you very much,

Best Regards,
Tran Quang Dat.

Hardcoded nu & nlvls

In LinearElasticity.cc line 22-23 nu and nlvls are hardcoded

LinearElasticity.cc error

Dear all,

are there anyone have the problem when run make topopt with the error like this?, can you help me to see the problem.

/Documents/TopOpt_in_PETSc/LinearElasticity.cc:613:32: error: cannot convert ‘PetscBool’ to ‘PCMGGalerkinType’ for argument ‘2’ to ‘PetscErrorCode PCMGSetGalerkin(PC, PCMGGalerkinType)’
PCMGSetGalerkin(pc,PETSC_TRUE);

Parallel speedup

I ran the catilever example in the documentation: mpiexec -np 2 ./topopt.
It ran fine, but the -np argument to mpiexec does not seem to do anything useful.
It just runs duplicate copies of the same simulation with no speedup.
Is there something I need to do to take advantage of multiple CPUs?
I am running on an Apple M1 chip.

Is that possible to use the PETSc-MMA with the overlapping Schwarz domain decomposition (DM) algorithm.

Thank you for the development of the PETSc-MMA, and making it available on Github. I am using  the overlapping Schwarz domain decomposition (DM) algorithm, and I am trying to build your MMA as a dynamic library and link it to my code. However, I have an issue. My understanding is that in your parallel framework the domain decomposition method results in several sub-domains without an overlapping area. But in the overlapping Schwarz domain decomposition algorithm there are overlapping areas. When I tested your MMA, it behaves well with 1 thread under DM algorithm, while the more threads yields wrong results.
I was wondering if it is possible to use your PETSc-MMA with DM algorithm. Thank you in advance for helping me with this question.

Memory use for big models

Hi,

When I am testing the standard problem with large meshes: 4 to 40 million elements, I see that the memory requirements go up into more than a 100GB's. For these testes I used between 16 and 48 cores and a density filter with a very small filter radius.

Is this memory usage normal or is there a way to decrease the amount of memory required?

Best, Thijs

Problem with plotting

Dear all,

I am trying to reproduce the example. I ran:

mpiexec -np 4 ./topopt -filter 2 -rmin 0.2 -volfrac 0.3 -maxItr 160

then

python bin2vtu.py 18.

However, when opening the file output_00018.vtu, I get an empty box:

Can you give me a hint please?

THank you for your help,

Romain

Terminate called after throwing an instance of 'std::bad_array_new_length'

Hi all,
I am using the PETSc 3.7.7 and want to update to new version of the PETSc 3.11.3.
I got an error while compiling with a newer version of PETSc 3.11.3 as following:
" terminate called after throwing an instance of 'std::bad_array_new_length'
terminate called after throwing an instance of 'std::bad_array_new_length'
what(): std::bad_array_new_length
terminate called after throwing an instance of 'std::bad_array_new_length'
what(): std::bad_array_new_length
terminate called after throwing an instance of 'std::bad_array_new_length'
what(): std::bad_array_new_length
terminate called after throwing an instance of 'std::bad_array_new_length'
what(): std::bad_array_new_length
terminate called after throwing an instance of 'std::bad_array_new_length'
what(): std::bad_array_new_length
what(): std::bad_array_new_length"
There was no error when I used the older version PETSc 3.7.7.
Can you help me to solve this problem?

Thank you very much,
Best Regards,
Tran Quang Dat.

TopOpt.h:4:10: fatal error: petsc.h:No such file or directory

I can't make TopOpt successfully. When I do make TopOpt, it says "TopOpt.h:4:10: fatal error: petsc.h:No such file or directory". Could you please tell me how to do with it?

makefile_ref

Hey! I'm getting
./makefile_ref: line 2: all: command not found
./makefile_ref: line 3: CFLAGS: command not found
./makefile_ref: line 15: include: command not found
./makefile_ref: line 16: include: command not found
./makefile_ref: line 17: include: command not found
./makefile_ref: line 19: topopt:: command not found
./makefile_ref: line 21: -: command not found
./makefile_ref: line 22: main.o: command not found
./makefile_ref: line 25: myclean:: command not found

when I'm running makefile_ref.
I've installed PETSc 3.13, MPI and BLAS/LAPACK
any idea in what am I doing wrong?
Thanks!!

bin2vtu problem

the bin2vtuy converter is not compatible with newest paraview versions. Can only be used with PV 5.7 or earlier