pnkraemer / probfindiff

Probabilistic numerical finite differences. Compute finite difference weights and differentiation matrices on scattered data sites and with out-of-the-box uncertainty quantification.
https://probfindiff.readthedocs.io
MIT License
10 stars 2 forks source link

Vector-value differential operators #6

Open pnkraemer opened 2 years ago

pnkraemer commented 2 years ago

It would be quite incredible if one could make the finite difference coefficients handle vector-valued functions/operators well. For example, the Jacobian.

One difficulty will be that the number of dimensions could become quite awkward...

pnkraemer commented 2 years ago

I think some good API decision would be to matchthe shapes of input and outputs to what would've been returned by AD (https://jax.readthedocs.io/en/latest/notebooks/autodiff_cookbook.html).

See also https://findiff.readthedocs.io/en/latest/source/examples-vector-calculus.html