Close #13

[anshgupta1234]

In m_variables_conversion in pre_process and post_process, added a pointer s_compute_pressure that points to 3 new subroutines:

• s_compute_pressure_from_energy (when model_eqns != 4 and no bubbles)
• s_compute_pressure_from_bubbles (when model_eqns != 4 and bubbles)
• s_compute_pressure_4eqns (when model_eqns == 4)

In m_data_output in pre_process and m_variables_conversion in pre and post-process, the EOS has been replaced by a call to s_compute_pressure

[sbryngelson]

Does this work on GPUs? Can you verify that the post-processing .silo files are the same before and after this change (they aren't part of the CI).

[sbryngelson]

You also did not modify the simulation code, which is part of this issue

[henryleberre]

@anshgupta1234 You might also want to consider placing these subroutines in a file in src/common

[anshgupta1234]

@henryleberre Actually, the subroutines are slightly different in pre_process and post_process so I probably can't do that 😅

[sbryngelson]

@henryleberre Actually, the subroutines are slightly different in pre_process and post_process so I probably can't do that 😅

Consider changing and refactoring them so that they can all be the same.

[anshgupta1234]

Sure! I think I can refactor them and try to bring them into src/common. Is there any specific file I should bring it into or create a new module?

I'll also work on simulation, but that might take a while since it has to be compatible with OpenACC.

Apart from that, I actually don't have any GPUs on my laptop so I'm unsure if I can do GPU testing for this code.

[sbryngelson]

It should probably be a new module. Probably m_variables_convert.f90. I added you to a GPU cluster, you will receive instructions for that via email.

[anshgupta1234]

Great! I'll test it out on a GPU whenever I have the chance. Till then, I did make sure to run examples/2D_shockbubble and examples/2D_axisym_shockbubble on this branch and master, and they both looked identical in ParaView.

In this latest commit, I did name the module m_compute_pressure because naming it m_variables_convert might be more confusing when importing in other files, and could look like:

use m_variables_conversion    ! From local dir
use m_variables_convert       ! From src/common

If this is not an issue, I can definitely change the file name :)

[sbryngelson]

I think this PR should also try to move as much of m_variables_conversion.f90 to a common location as possible (perhaps all of it!). Great job so far.

[sbryngelson]

@anshgupta1234 what is the status of this? I cannot tell.

[anshgupta1234]

@sbryngelson the status of this PR is incomplete. It actually represents the branch for just moving all the pressure subroutines to a new module, but then we decided to pivot to moving all of m_variables_conversion to common, which I'm working on in a different branch that's based off of this one. Perhaps it might be better to close this PR and I can just create a new one when I'm done with the new task?

[sbryngelson]

It's fine how it is I was just curious.