Adding DVGeometryMulti to MPhys

[bernardopacini]

Purpose

Adding DVGeometryMulti to MPhys. I added pass-through functions and modified the compute and compute_jacvec_product calls as needed.

Expected time until merged

1 week.

Type of change

• Bugfix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (non-backwards-compatible fix or feature)
• Code style update (formatting, renaming)
• Refactoring (no functional changes, no API changes)
• Documentation update
• Maintenance update
• Other (please describe)

Testing

I tried this on my optimization and it appears to work (local analysis and derivatives on HPC). I would appreciate others verifying this on their cases.

Checklist

• I have run flake8 and black to make sure the Python code adheres to PEP-8 and is consistently formatted
• I have formatted the Fortran code with fprettify or C/C++ code with clang-format as applicable
• I have run unit and regression tests which pass locally with my changes
• I have added new tests that prove my fix is effective or that my feature works
• I have added necessary documentation

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 38 lines in your changes missing coverage. Please review.

Project coverage is 65.35%. Comparing base (3404b51) to head (216832e).
Report is 3 commits behind head on main.

Files with missing lines	Patch %	Lines
pygeo/mphys/mphys_dvgeo.py	0.00%	38 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #230      +/-   ##
==========================================
- Coverage   65.47%   65.35%   -0.12%     
==========================================
  Files          47       47              
  Lines       12265    12286      +21     
==========================================
  Hits         8030     8030              
- Misses       4235     4256      +21     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[bernardopacini]

I didn't add the addIntersection() method to the MPhys wrapper. Let me know if you think it should be here or if it should be left as a direct reference to the underlying class.

[bernardopacini]

Additional note: when adding a global DV the input was being added twice to OpenMDAO. This bug is fixed here too.

[bernardopacini]

It turns out PR #201 broke the naming convention for child FFDs by prepending the child name by default. This meant that MPhys would not connect the OM variable to the DVGeo variable because the names would not match. I fixed that in this PR by setting prependName=False for all FFD-based design variables.

[sseraj]

Looks pretty good. I just have a few minor comments

[anilyil]

Most of this look good, but one big picture comment: it looks like the logical check to find the correct dvgeo is repeated in several places. Instead, I think you can reduce duplication by using "nom_getDVGeo" before each DV addition. That way you wont need to duplicate the code that finds the right DVGeo. Is this correct? If you also think this makes sense, I say go for it. If not, I am fine with merging as is.

[hajdik]

LGTM, just a few comments

[bernardopacini]

There is some weird behavior in the derivatives: the first check totals is fine, but a second check totals after a design perturbation makes the second function derivative wrong. This only happens when there is an intersection. I can't exclude this being an issue with DAFoam (or even IDwarp), but I do not have the bandwidth to look more into this now. Converting to a draft.

[bernardopacini]

This is good to go. The issues with derivatives are coming from other code components, not DVGeo, and they are not unique to this PR. We should merge this in so it does not go stale again.

[anilyil]

Looks good overall. Is there a corresponding addIntersection call? If not, how are you handling it?

[bernardopacini]

Looks good overall. Is there a corresponding addIntersection call? If not, how are you handling it?

Instead of having a pass-through call in MPhys, I just call the DVGeo method directly. For example:

DVGeo = self.geometry.nom_getDVGeo()
self.geometry.nom_addComponent("hub", "FFD/hubFFD.xyz", triMesh="hubTri.cgns")
self.geometry.nom_addComponent("blade", "FFD/bladeFFD.xyz", triMesh="bladeTri.cgns")
DVGeo.addIntersection(
    "hub",
    "blade",
    dStarA=1.0,
    dStarB=0.05,
    featureCurves=featureCurves,
    project=True,
    includeCurves=True,
    curveEpsDict=curveEpsDict,
    anisotropy=[1.0, 1.0, 0.75],
    slidingCurves=slidingCurves,
    trackSurfaces=trackSurfaces,
)

Is this fine with you?

[anilyil]

Yeah this is fine I guess. The add intersection call itself does not modify openmdao i/o so thats ok