Distributed simulation for large models

[chengumd]

Hi akaszynski, thanks for the beautiful work. with the current pyansys version, when establishing the ANSYS instance, is it possible to specify the distributed simulation settings such as number of cpus and memory space as one usually can do in batch mode?

I have a set of inp files which are really hard to maintain and update, due to the outdated nature of apdl. I was thinking to drive the simulation with pyansys. Thank you for any suggestions.

[akaszynski]

Hello,

You're able to pass additional switches to the Mapdl class on initialization. For example, if your command line looks like:

/ansys_inc/ansys/bin/ansys201 -machines 0.0.0.0:30:10.0.0.79:30

Your corresponding python command would be:

import pyansys
mapdl = pyansys.Mapdl(additional_switches='-machines 0.0.0.0:30:10.0.0.79:30')

[chengumd]

Hey Alex, thanks for the help. I was able to install pyansys on a server. However, when I try to establish an Ansys instance using pyansys.Mapdl(), ansys = pyansys.Mapdl(run_location= path) It returns the following error stating that the Mapdl() is missing.

Traceback (most recent call last): File "", line 1, in AttributeError: 'module' object has no attribute 'Mapdl'

The other modules such as pyansys.change_default_ansys_path() work just fine. Thanks you very much again for your help.

Best,

[akaszynski]

That's strange. Was pyansys installed via pip? If so, you might have another directory or module named "pyansys" and it's looking at that instead of at the actual installed module.

[akaszynski]

I'm guessing that this issue was related either to a bad install of pyansys or some other issue. Could you please post the script you used here? Also, please include:

import scooby
scooby.Report(['pyansys'])

import pyansys
print(pyansys.__file__)

[akaszynski]

Please let me know if this is still an issue. Otherwise, I'll be closing this soon.