search

eitcom / pyEIT

Python based toolkit for Electrical Impedance Tomography
Other
169 stars 96 forks source link

add jac_normalized in setup parameter for jac and remove init flag from solve_eit,... #51

Closed DavidMetzIMT closed 2 years ago

DavidMetzIMT commented 2 years ago

Hy @liubenyuan,

I saw that we forget the jac_normalized option for jac. is it also needed for greit? If yes I can added it!

By the way I just forget to ask you why you introduced the init flag on solve_eit, compute_b_matrix, compute_jac_matrix. I think the init flag can be removed! each time you set it to true you pass a perm vector and viceversa, so it make it the flag redondant.
I would suggest we recalculate kg only when perm is not None. It is maybe more intuitive, I was confused by using it in my own sofware. I made made some modifications (which made fem even more lighter!)

liubenyuan commented 2 years ago

Yes, this parameter jac_normalized should be kept for the analysis of the sensitivity distributions.

I added a note on using perm = None or user_perm != perm to trigger a recalculation of se and kg. I think your modification dropping init=True is truly intuitive to use!

The warning message on passing perm=None can be suppressed, as it is the desired behavior (default) of calculating se from mesh.perm. We can add a note on this parameter rather than throw a warning message.

DavidMetzIMT commented 2 years ago

Ok I deleted the warning and added also jac_normalized to the setup parameters in greit solver!