Ethan Peterson

Thanks for taking a look @drewejohnson! All good points. > Now, the question becomes does this change make the results more correct, and thus the failure is because the reference...

FWIW the tests in `tests/regression_tests/deplete_with_transfer_rates/` do pass locally for me on this branch since adding the `direct_solve` kwarg to keep the behavior the same as before when `transfer_rates` is set.

I have updated the profile results for this patch (difference between current develop and this branch) since some of the other improvements have been made. This is the same depletion...

> I personally don't like the inclusion of the conditionals in the CRAM solve. Is the rationale behind not using LU for transfer rate problems we can't get them to...

It should definitely be feasible to apply the weight cutoff relative to the local weight window lower bound to get the benefits of both, no?

mmmmm this is probably my fault sorry. Although I think the `inplace` only means without changing the ID of that surface since it doesn't make sense to return the same...

I think the original intent was to avoid the situation like the following `cone = ZCone(surface_id=5).rotate((0, 90, 0)` giving you a cone back with a different surface ID. I'm not...

Hi @FusionSandwich, thanks for raising the issue. The problem here is that the input points need to be ordered by the user. They can be ordered either clockwise or counter-clockwise,...

Generally with a call to a constructor like `mat = nmm.Material(**kwargs)` there shouldn't be any side effects that aren't clear to the user. The ability to create a material with...

@Shimwell I think that makes sense although the `from_library` and `from_mixture` functions should be class methods so you don't need the parentheses i.e. `mat_3 = nmm.Material.from_mixture(materials=[mat_1, mat_2], fracts=[0.5, 0.5]) #...