dstndstn / tractor / 8ae6087f-b248-4fdc-94c7-bb5b46da2df3 / 1

46%

main: 46%

push

circleci

Now seems to be fully correct (fingers crossed).
Commented out intermediate save files but still have debug info printed.
Bug fixes:
	- indexing of npos when fit_pos is False
	- use X = cp.einsum("ijk,ik->ij", cp.linalg.pinv(A), B) instead
	  of cp.linalg.solve of A.T dot A, A.T dot B to avoid NaNs
	- Shift ci index when fit_pos is False to properly index priors
	- Use entire conv_mog.mean not just first row in calculation of
	  mog_g.
	- sourceOut = use CPU for now

702 of 2177 branches covered (32.25%)

3370 of 7394 relevant lines covered (45.58%)

0.46 hits per line