Tps init #7
Merged
Tps init #7
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
… apart from tps_coefs. I might need the bending mat in semilmks relaxation. Here I'm just returning the pxp submat of L inverse. In the 89 bookstein paper, I think he returns Linv*K*Linv, but I don't think that definition is used elsewhere. Not sure.
I've really not tested things too thoroughly. For eg, not doing any checks on how the non-affine part of the warp behaves.
I'm doing this to be able to use `solve` or `lstsq` next (instead of calculating the `inv`).
I'm requiring mkl cuz the intermitten fail on travis is very likely due to something wrong with OpenBLAS. See statsmodels/statsmodels#5396. I'm removing py3.5 because requiring mkl in install trips up tox, (or I guess more specifically python setup.py, and it throws a DistUtilsError about not finding numpy>=1.10.0).
Removing mkl cuz it does not seem to help with the intermittent NaN issue on travis. Trying to just test on osx.
I think this might be better because of the problem is really only when I have 5 or more lmks, and this way I can have more oses.
The README_for_docs also contains a contents local directive at the beginning, as well as links for the high-level ops.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I had great trouble in trying to run travis tests for the tps warping with 5 landmarks (and I imagine more).
The failure would occur intermittently, and always at the L matrix inversion step. Somewhere during the step the numpy internal algorithm would encounter NaN values (although the 5x2 matrix going in does not have any) and give up. Changing from just
L_inv * Y0tonp.linalg.lstsq(L, Y0)seemed to give a more informative error -ValueError: On entry to DLASCL parameter number 4 had an illegal value.Some googling reveals the above error to possibly be at the OpenBLAS level. I'm not sure at the moment, but if it is true, there is little I can do about it.
It could be that restricting travis to just py 3.6 might help (although I have only a small sample of tests I'm basing this on). But I've decided instead to restrict the tps warp dataset to squares for now, so that I can keep testing on py3.5, 3.6 and 3.7.