Replies: 1 comment
-
We've merge a PR(s) to main to make the Fortran ordering more consistent when we return numpy arrays, print out the F ordering with the tensors, and print based on frontal slices. Took a little time to get things cleaned up. I'll close this for now but we can re-open if there are other conceptual things to discuss. Otherwise can open issues for any edge cases we may have missed. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
I have two issues (#338 & #340) that perhaps are better at this point as discussions. I think there is also some subtle connection here to the underlying ordering. I really dislike the numpy way of displaying arrays because it doesn't align the columns. But perhaps a bigger concern for me is that the bracketing gives the impression of a C-style ordering rather than an F-style ordering. I don't know that it's clear enough in the documentation that pyttb uses a F order under the hood. For anyone doing linear indexing into the tensor, this will be really critical to understand. I could even imagine stating this explicitly every time a tensor is printed.
Beta Was this translation helpful? Give feedback.
All reactions