Closed briantoby closed 2 months ago
started on this.
Mathew suggests looking at Automatic differentiation (https://jax.readthedocs.io/en/latest/automatic-differentiation.html). See https://jax.readthedocs.io/en/latest/automatic-differentiation.html#taking-gradients-with-jax-grad.
Provide users with a mechanism to supply an arbitrary equation into GSAS-II that derives a value from the fitted parameters. The derivatives for that equation can then be determined numerically and the uncertainty for that parameter can then be evaluated.