Skip to content

Conversation

oxinabox
Copy link
Member

closes #151

@oxinabox
Copy link
Member Author

oxinabox commented May 28, 2021

Added two more changes.

Firstly to stop testing exacty what type + returned.
Because Adjoint + Adjoint = Matrix so it isn't trivially consistent, but per my last comments in #151 I don't really care that much, we have to handle that case because we have to handle that case.
Better to be pragmatic.
It's also out of the power of this package to fix.
That will be fixed with projection/rerepresentation which will be defined else where.

Instead for more complicated nested objects where we might screw-up, I test that.

Secondly, this really relaxed a lot about how difference works.
I think the new behavour is intuitive but it feels more dubious.
But I want to delete difference anyway, so i don't think it matters (#166)

@mzgubic mzgubic changed the title =Make rand_tangent on adjoint an transpose return natural Make rand_tangent on adjoint an transpose return natural May 28, 2021
@mzgubic
Copy link
Member

mzgubic commented May 28, 2021

LGTM after difference is deleted

@oxinabox oxinabox merged commit f70a651 into master May 28, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

rand_tangent(adjoint(array)) returns Composite even though Adjoint is a natural differential

3 participants