CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   SU2 Shape Design (http://www.cfd-online.com/Forums/su2-shape-design/)
-   -   Another shape optimization tutorial error (http://www.cfd-online.com/Forums/su2-shape-design/112704-another-shape-optimization-tutorial-error.html)

grjmpower February 3, 2013 11:08

Another shape optimization tutorial error
 
I have run the Tutorial 9 case on two different Linux systems: my personal 2 core OpenSUSE system (gnu compilers) and a large CRAY HPC system (PGI compilers). I am following the new tutorial instructions, except that I do not copy the files into the SU2Py directory, but rely on the fact that the executables and Python scripts are in my path.

I made sure I had the most recent version of the code and test cases. The SU2_MDC code ran successfully and I made the changes to the input file to point to the new mesh file.

I can run shape_optimization.py with -p 1 and the CFD solution part runs successfully.

If I run shape_optimization.py with the -p option, then the mesh files do not exist. Apparently the SU2_DDC code is not being called in this case since multiple mesh files do not exist. If I run SU2_DDC first, then the mesh files are there and the first CFD solution is run successfully.

I can also run parallel_computation.py to get the initial solution and change the input to RESTART = YES

However, when running the shape_optimization.py the following error occurs after successfully running the CFD portion whether running with 1 or multiple processors:

------------------------- Exit Success (SU2_CFD) ------------------------

Application 5022852 resources: utime ~4593s, stime ~82s


-------------------------------------------------------------------------
| SU2 Suite (Shape Optimization Script) |
-------------------------------------------------------------------------


DRAG objective function with a scale factor of 100.0

Traceback (most recent call last):
File "/lustre/scratch/grjmpower/local/bin/shape_optimization.py", line 676, in <module>
acc=1e-12, iprint=2, full_output=1, epsilon=1.0e-10)
File "/sw/xt/python/2.7.1/sles11.1_gnu4.3.2/lib/python2.7/site-packages/scipy/optimize/slsqp.py", line 237, in fmin_slsqp
mieq = len(f_ieqcons(x))
File "/sw/xt/python/2.7.1/sles11.1_gnu4.3.2/lib/python2.7/site-packages/scipy/optimize/optimize.py", line 176, in function_wrapper
return function(x, *args)
File "/lustre/scratch/grjmpower/local/bin/shape_optimization.py", line 463, in ieqcons
logfile = 'SU2_CFD.out' )
File "/lustre/scratch/grjmpower/local/bin/continuous_adjoint.py", line 203, in continuous_adjoint
ObjFun_Dict = libSU2.get_ObjFunVals( History_filename_CFD+plotfile_ext , special_cases )
File "/lustre/scratch/grjmpower/local/bin/libSU2.py", line 495, in get_ObjFunVals
history_data = read_History(History_filename)
File "/lustre/scratch/grjmpower/local/bin/libSU2.py", line 411, in read_History
line = line.split("=")[1].strip().split(",")
IndexError: list index out of range

The Python version is 2.6.8 on both platforms.

rktchip February 5, 2013 01:42

would you mind posting the full console output (including the command line call) that generates this, as well as the config file? thanks!

grjmpower February 5, 2013 19:05

5 Attachment(s)
The commands, configuration file and output from the MDC and shape_optimization.py are attached.

Thanks!

EMolina February 5, 2013 19:10

1 Attachment(s)
Hi all,

I am facing almost the same problem running the inviscid Onera Optimization Test case. I installed Numpy and Scipy from "apt-get install" and I got the following error. After I removed the old Numpy and Scipy and install the updated version of both softwares from source forge and the same error persists. I am running Ubuntu 12.04 LTS, Python 2.7.3, numpy 1.6.2 and Scipy 0.11.0.

One intersting thing is that the version 1.1 had run ok on the same tutorial. Just a question, I remember seeing someone using Ubuntu in the SU2 V2.2 workshop last month, which version he was using? It was you, Trent?

I'm breaking my head in the last three days and I do not understand what is happening. I have attached the prompt output with the error and the config file.

Any help is welcome.

Thank you in advance.

Eduardo



molina@Desktop-Molina:~/Documentos/Programas/SU2v2.0/TestCases/cont_adj_euler/oneram6$ python shape_optimization.py -f inv_ONERAM6.cfg -p 1 >> log.txt
Traceback (most recent call last):
File "shape_optimization.py", line 676, in <module>
acc=1e-12, iprint=2, full_output=1, epsilon=1.0e-10)
File "/usr/local/lib/python2.7/dist-packages/scipy/optimize/slsqp.py", line 196, in fmin_slsqp
constraints=cons, **opts)
File "/usr/local/lib/python2.7/dist-packages/scipy/optimize/slsqp.py", line 351, in _minimize_slsqp
fx = func(x)
File "/usr/local/lib/python2.7/dist-packages/scipy/optimize/optimize.py", line 268, in function_wrapper
return function(x, *args)
File "shape_optimization.py", line 245, in f
logfile = 'SU2_CFD.out' )
File "/home/molina/Documentos/Programas/SU2v2.0/TestCases/cont_adj_euler/oneram6/continuous_adjoint.py", line 203, in continuous_adjoint
ObjFun_Dict = libSU2.get_ObjFunVals( History_filename_CFD+plotfile_ext , special_cases )
File "/home/molina/Documentos/Programas/SU2v2.0/TestCases/cont_adj_euler/oneram6/libSU2.py", line 495, in get_ObjFunVals
history_data = read_History(History_filename)
File "/home/molina/Documentos/Programas/SU2v2.0/TestCases/cont_adj_euler/oneram6/libSU2.py", line 411, in read_History
line = line.split("=")[1].strip().split(",")
IndexError: list index out of range

rktchip February 6, 2013 08:02

1 Attachment(s)
Thanks alot for posting your files, they were very helpful.

It looks like we missed the paraview plot format in the read_history() function in libSU2.py. Would you please try replacing the file with this? <Attachment 18810>

Best,
- Trent

EMolina February 6, 2013 20:25

Thanks a lot....

It is working now...

rktchip February 7, 2013 04:12

Great! Sorry that caused so much headache...


All times are GMT -4. The time now is 11:07.