Closed mrava87 closed 1 year ago
@cako i decided to tackle this first as initially I thought it would just mean swapping inputs and filters… didn’t realize that the adjoint is completely different as now the model lives in the filter space. Never mind, the pure python (with numba) code is ready for review. I prefer to wait to add the cuda version after I tackle point 3 in your review
Looks good! See my comment on the race condition. Up to you if you want to change it in this PR.
@cako let me know if you agree with these changes and I merge :)
This PR introduces two new operators as described in https://github.com/PyLops/pylops/issues/466 (point 4).
For a more detailed example usage see https://github.com/PyLops/pylops_notebooks/blob/master/developement/NonStationaryConvolveFilters.ipynb