Add non-energy boundary adjoint example#63
Open
krystophny wants to merge 35 commits into
Open
Conversation
65c5d37 to
c543810
Compare
c543810 to
22c5608
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Chains the non-energy objective gradient through the equilibrium boundary adjoint.
simsopt_gvec_objective.pycomposesGvecEnergyProblem's cached interior re-solve withState.objective_state_gradientandadjoint_boundary_gradient, giving an analytic boundary gradient forbsq_variancewith no finite-difference re-solves.OptimizablewrapperGvecObjectivewithJ()/dJ()and an L-BFGS-Boptimize_boundaryloop.Closes #27.
Verification
Test fails before
Test passes after
Enzyme-on isolated build (fixed RZ, LLVM/Flang 22.1.6):
test_non_energy_boundary_gradient_sampledre-solves the interior for every boundary DOF and requires cosine similarity above 0.99 andrtol=5e-3between the adjoint gradient and the finite-difference truth.Enzyme-off isolated build: