-
Notifications
You must be signed in to change notification settings - Fork 841
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Feature libROM for reduced order modelling #1310
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hi Jessie, thank you for this contribution.
It is not my specialty but I recall one or two interesting presentations.
Some general questions before we get into code details:
Are the results of this ready to use? Or do they need some additional post-processing steps to make sense? If so it would be good to have an example with any scripts needed.
Our current gold standard has been set by @TobiKattmann, see TestCases/incomp_navierstokes/streamwise_periodic.
If you were already planning to add a tutorial then nevermind the above.
Thanks Pedro for the suggestions! I'm working on the changes now. |
Co-authored-by: Pedro Gomes <[email protected]>
There are some post-processing steps involved, so I think it would be a good idea for me to include an example. The features in this PR are one step to a bigger projection-based ROM process. The code for the bigger process is in feature_MOR, and will take some time for me to prepare the code from "make it work" to "actually good code". However, this feature-PR can be used by itself by anyone who wants to efficiently compute the POD of SU2 simulation data, and then use the POD for other modeling strategies. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yup a short example would be nice, but if there is more work coming we can wait until then.
Co-authored-by: Pedro Gomes <[email protected]>
Co-authored-by: Pedro Gomes <[email protected]>
Co-authored-by: Pedro Gomes <[email protected]>
Co-authored-by: Pedro Gomes <[email protected]>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thank you for the changes, just a couple more things I noticed.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
Proposed Changes
Adding support to use the external library libROM (https://github.com/LLNL/libROM) for reduced order modeling with SU2.
libROM captures solution data and either computes the proper orthogonal decomposition (POD) over the data, or saves the solution "snapshot" in an hdf5 file for later POD computation. The POD matrix is also saved in an hdf5 file.
Related Work
There are a couple projects using libROM:
For projection-based model order reduction (PROM) in the branch feature_MOR
It is also being used with the NEMO solver to further test the PROM capabilities
libROM was used in a prior project that used the POD matrix as input to a neural network (to be published)
libROM also has other features, like help choosing the hyper-parameter nodes (a PROM step) and under development are black-box modeling methods like dynamic mode decomposition (DMD)
Please let me know if you have any questions / comments / recommendations!
I have tested this on macOS and linux.
PR Checklist
Put an X by all that apply. You can fill this out after submitting the PR. If you have any questions, don't hesitate to ask! We want to help. These are a guide for you to know what the reviewers will be looking for in your contribution.
Steps to Compile:
./meson.py build --prefix=/your/prefix -Denable-librom=true -Dlibrom_root=/path/to/libROM
export LD_LIBRARY_PATH=/path/to/libROM/build/:$LD_LIBRARY_PATH
SAVE_LIBROM=YES
to the Quickstart config