Speed up DVJacobian computation

[marcomangano]

Purpose

Opening this PR on behalf of @anilyil.

pyGeo currently recomputes the Jacobian of the control points w.r.t. the user-defined DVs in computeDVJacobian every time we call computeTotalJacobian. If the DVs are not updated (e.g. we are calculating derivatives on different pointSets on the same design) this step is unnecessary (and expensive) as the Jacobian remains the same. Anil added a flag self.JTempUpdated to monitor if the Jacobian is up to date with the current design. If that is the case, we skip the CS-based calculation of the DVJacobian and immediately return the current matrix. The modifications do not change the code behavior but will improve code performance, especially for more complex geometry setups.

Expected time until merged

2-3 days

Type of change

• [x] 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

Checklist

• [x] 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
• [x] 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[bot]]

Codecov Report

Merging #189 (92b4286) into main (41ffb0b) will increase coverage by 0.01%. The diff coverage is 91.66%.

@@            Coverage Diff             @@
##             main     #189      +/-   ##
==========================================
+ Coverage   64.77%   64.79%   +0.01%     
==========================================
  Files          47       47              
  Lines       11951    11957       +6     
==========================================
+ Hits         7741     7747       +6     
  Misses       4210     4210              

Impacted Files	Coverage Δ
pygeo/parameterization/DVGeo.py	68.77% <91.66%> (+0.09%)	:arrow_up:

:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more

[marcomangano]

The implementation looks good. My only comment is about the variable name. J_temp is fine for a temporary local variable, but I would prefer calling it something like J_DV if it is going to be a class variable.

Agreed! Pushing soon