[Nipy-devel] Affine orientation ordering - please review and test
Tue Jan 19 14:15:57 CST 2010
> - RS = RZS / zooms
> + RS = np.dot(RZS, np.diag(zooms))
Yes, it is easy to explain -- it's a bug and should be reversed. Sorry about
It is meant to be np.diag(1/zooms).
I used the np.dot
just because I wasn't sure that in the non-square case broadcasting would
work (it does, I've now checked).
Interesting that none of the tests picked up this bug, I suppose one would
have to exaggerate one row severely.
Dept. of Statistics
Sequoia Hall, 137
390 Serra Mall
Stanford, CA 94305
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Nipy-devel