Adding features for release of eFlows4HPC European Project #10081
Conversation
| @@ -2,21 +2,36 @@ | |||
| import KratosMultiphysics.FluidDynamicsApplication | |||
There was a problem hiding this comment.
These changes go against the latest modifications. Implementations must be done in the rom_analysis. Note that this class is deprecated and is only keep for retrocompatibility.
Is there any technical limitation that avoids implementing these new features in the rom_analysis?
There was a problem hiding this comment.
The rom_analysis that we have now works well for testing small cases, since it allows to train a ROM quite automatically. However, this automatic workflow was not inmmediatly usable in a larger parallel training campain.
That is, the rom_analysis builds the ROMs using a single time-dependent, or a single quasi-static simulation(with multiple load-steps). If more parameter variations are to be studied, one needs to compute the SVD of the snapshots for the whole set of simulations. The scripts for such a larger parallel training are not included in this PR, since they involve the external library COMPSs. They are being place in the Kratos/Examples repository.
Incorporating an expanded workflow for the training of a ROM in the rom_analysis will be done in another PR.
Notice that the modifications on the fluid_dynamic_analysis_rom add but a single extra functionality in comparison to the rom_analysis. The PATH is accepted as an extra parameter, in order to allow to run in a cluster. All the real handling is done externally.
There was a problem hiding this comment.
Incorporating an expanded workflow for the training of a ROM in the
rom_analysiswill be done in another PR.
If there is no technical limitation, can't we just simply do this?. From the sake of code sustainability it makes no sense to invest effort on developing and maintaining deprecated code.
Notice that the modifications on the
fluid_dynamic_analysis_romadd but a single extra functionality in comparison to therom_analysis. The PATH is accepted as an extra parameter, in order to allow to run in a cluster. All the real handling is done externally.
This can be easily added to current rom_analysis.
There was a problem hiding this comment.
We had a brief discussion with @Rbravo555 about this. As the due date of the eFlows4HPC release is close, we decided to do the release with the old stage version to ensure everything is working fine. This comes with the compromise all these developments to the current rom_analysis after the deliverable submission.
applications/RomApplication/tests/test_randomized_singular_value_decomposition_mpi.py
Outdated
Show resolved
Hide resolved
📝 Description
Adds features for parallel computation of SVD and Simulations
🆕 Changelog
fluid_dynamic_analysis_romto ease its launching in clusters (no extra tests added, since it depends on external library COMPSs)