Autodiff function for coefficient [add-ad-coeff] #4256
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Outdated suggestions cannot be applied.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.
This PR defines as helper class that can convert user provided functions in to first order derivatives, that can be used by the various
XXXFunctionCoefficient
classes.The functions need to to be provided as either
void vectorTDFun(const Vector& param, const ADVector& x, ADVector& f)
or
void vectorTDFun(const Vector& param, const ADVector& x, const ADFloat t, ADVector& f)
where
ADVector
is of typestd:vector<ADFloat>
andADFloat
is equivalent toreal_t
with additional autodiff functionality.A full suite of unit tests is provided. This also gives a usage examples.
Envisioned usages: convergence tests based on norms including derivatives.
Functionality could - in principle -be extended to Hessian, Laplacian and other higher order derivatives.
Credit were credit is due: large inspiration has been taken from the autodiff miniapp.