bempp/bempp-rs

Array3D shape given in different format to numpy/scipy

skailasa opened this issue · 3 comments

We currently return shape as (x, y, z), however it's standard in python science stack to give in (z, y, x) - i.e. from outer most to inner most axis.

May be useful to keep it consistent to what users may be expecting?

I thought the way I had it was consistent with how a Numpy ndarray would work, which seemed the most logical to me?

(As we're planning to replace Array3D with a tensor from RLST anyway, this might not be worth fixing but instead we should get it right when implementing these tensors.)

Yeah absolutely 👍

Closing this now that we're using RLST everywhere