[gina07]

Hi~Everyone~
I was using the su2-3.2.3 during the last month. And I tested the optimization_euler/steady_naca0012 and steady_oneram6 case for shape optimization purpose. However I received IOError:no such file or directory :mesh_oneram6_inv_FFD_1.su2. What can I do to correct such error?
Somebody said I could install SU2-3.2.7 ,the newest version and Testcases-master to ignore the errors. However,when I installed 3.2.7 in the same operation environment of Centos-6.5 system. I found that even the CFD solver could not receive the correct result just as seen with tecplot in the attachment.
And the shape optimization python script received another DivergenceFailure. I think it is owing to the wrong fluid and adjoint results.
I am using the CentOS 6.5 system and openmpi and as you have told me, the canopy for python ,scipy and numpy. I want to use SU2 to do the shape optimization project. I hope one version of SU2 can successfully run on my linux PC.Could you help me?

[hlk]

Originally Posted by gina07

Hi~Everyone~
I was using the su2-3.2.3 during the last month. And I tested the optimization_euler/steady_naca0012 and steady_oneram6 case for shape optimization purpose. However I received IOError:no such file or directory :mesh_oneram6_inv_FFD_1.su2. What can I do to correct such error?
Somebody said I could install SU2-3.2.7 ,the newest version and Testcases-master to ignore the errors. However,when I installed 3.2.7 in the same operation environment of Centos-6.5 system. I found that even the CFD solver could not receive the correct result just as seen with tecplot in the attachment.
And the shape optimization python script received another DivergenceFailure. I think it is owing to the wrong fluid and adjoint results.
I am using the CentOS 6.5 system and openmpi and as you have told me, the canopy for python ,scipy and numpy. I want to use SU2 to do the shape optimization project. I hope one version of SU2 can successfully run on my linux PC.Could you help me?

My apologies for not getting back to you sooner on your previous post;
I have not been able to replicate these issues however:
For the first error message (which look like its from parallel_compuation.py?) SU2_PRT is not a part of SU2 3.2.7 (this functionality is now taken care of elsewhere), since the output shows that SU2_PRT is being called, there is something going wrong with matching up the versions.
- make sure that all previous versions of SU2 are cleaned out (or at least removed from your $PATH) and run source ~/.bashrc (or equivalent if not using bash) to update the paths.
- make sure that you are using the most updated version of the python script - one way to double check that you are using the script you intend to is to copy the python script into your current directory and call it with "./parallel_computation.py" (the "./" in front makes it call from the current directory rather than whichever version is found in the path). Normally it is enough to have the SU2/bin directory included in your path and pythonpath, however this will at least eliminate that as the problem.
- check the config file to make sure that the mesh specified is in the current folder.


For the second error message (shape_optimization.py)
I was able to run this case for several optimization steps, so unfortunately I was not able to replicate this issue. If you are able to run the parallel_computation.py script and still have issues with diverging solutions for the optimization, the first troubleshooting step is to check the design geometry: Set VISUALIZE_DEFORMATION= YES in the config file. This allows you to look at the deformed geometry as the optimization is going, if it deforming a ridiculous amount you can reduce the objective function scale and/or set constraints. This can happen sometimes with optimization on the first step where it tries to take too large of a step in the design.

[gina07]

Originally Posted by hlk

My apologies for not getting back to you sooner on your previous post;
I have not been able to replicate these issues however:
For the first error message (which look like its from parallel_compuation.py?) SU2_PRT is not a part of SU2 3.2.7 (this functionality is now taken care of elsewhere), since the output shows that SU2_PRT is being called, there is something going wrong with matching up the versions.
- make sure that all previous versions of SU2 are cleaned out (or at least removed from your $PATH) and run source ~/.bashrc (or equivalent if not using bash) to update the paths.
- make sure that you are using the most updated version of the python script - one way to double check that you are using the script you intend to is to copy the python script into your current directory and call it with "./parallel_computation.py" (the "./" in front makes it call from the current directory rather than whichever version is found in the path). Normally it is enough to have the SU2/bin directory included in your path and pythonpath, however this will at least eliminate that as the problem.
- check the config file to make sure that the mesh specified is in the current folder.


For the second error message (shape_optimization.py)
I was able to run this case for several optimization steps, so unfortunately I was not able to replicate this issue. If you are able to run the parallel_computation.py script and still have issues with diverging solutions for the optimization, the first troubleshooting step is to check the design geometry: Set VISUALIZE_DEFORMATION= YES in the config file. This allows you to look at the deformed geometry as the optimization is going, if it deforming a ridiculous amount you can reduce the objective function scale and/or set constraints. This can happen sometimes with optimization on the first step where it tries to take too large of a step in the design.

Thank you very much,Kline~
I finally succeeded in doing the optimization after I switched user to ordinary user instead of root. I am new in using the Linux OS. All the configures of the operating environment were the same ,however, the root user could not complete the optimization. I am confused about the cause of that.
Now I can do the shape optimization testcases.However,in the naca0012 case the process stopped unexpectedly with a suddenly enlarged objfun and GNORM. Why could this happen? And in the oneram6 case the process went on for 38 iterations and stopped with an EvaluationFailure."SU2 process returned error 1". Could you help me? Thks~

Gina~

[imamwfgf]

Originally Posted by gina07

Thank you very much,Kline~
I finally succeeded in doing the optimization after I switched user to ordinary user instead of root. I am new in using the Linux OS. All the configures of the operating environment were the same ,however, the root user could not complete the optimization. I am confused about the cause of that.
Now I can do the shape optimization testcases.However,in the naca0012 case the process stopped unexpectedly with a suddenly enlarged objfun and GNORM. Why could this happen? And in the oneram6 case the process went on for 38 iterations and stopped with an EvaluationFailure."SU2 process returned error 1". Could you help me? Thks~

Gina~

Hi Gina,

I am running into the exact same situation with oneram6 stopping with an EvaluationFailure. "SU2 process returned error 1". Did you figure out how to fix it?

Best,

Will