Joshua J. Cogliati

I created a pull request to use pyDOE3 directly. #2298 I don't have time at the moment to switch the smt stuff to pyDOE, but the above pull request is...

I think this still has more to be written (it is a WIP after all). I am curious what else needs to be done. Also, a test case might be...

For what it is worth, here is some more example code that triggers the NaN after model.prune() in case anyone needs to test a fix. ```python import kan kan.torch.set_default_dtype(kan.torch.float64) device...

I continue to try and learn a bit more about it. Whatever prune does still allows forward pass to work, but not the fit. I am suspicious of `optimizer.step(closure) `...

What is the difference between this and #2489 ?

> What is the difference between this and [#2489](https://github.com/idaholab/raven/issues/2489) ? FYI: @JunyungKim and I discussed this. (This issue could use a bit more explaining why Barracuda will be useful.)

I believe this uses features in sparsesensing that are not yet in a released version, so besides any other work that needs to be done, this should not be merged...

> Job [Test mac](https://civet.inl.gov/job/2362449/) on [d2dc13b](https://github.com/idaholab/raven/commit/d2dc13b6646626ddab157e90c46c807a2cb9b7fd) : invalidated by @joshua-cogliati-inl > > failed in fetch Opps, actually failed in set python environment.