Nested AD for getindex
#944
Closed
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
This adds a bunch of definitions to make nested AD work and adds tests for second-order AD. Unfortunately by the time we get to third-order AD, the types get so large that base decides to go on vacation while it thinks about whether or not it might be willing to compile a function with a type of such complexity. Additionally, Zygote introduces some unnecessary stacks, which then prevent higher order AD. I plan to work on both of those issues, but in the meantime, here are the changes to Zygote required to make this work.
|
Thanks for looking into this. We might need to be a bit careful here to assess performance and memory issues. We should also have good amount of testing to ensure we can handle the use cases we care for |
| gradient(x) do x | ||
| sin(x[1]) | ||
| end[1][1] | ||
| end[1][1] == -sin(1.0) |
There was a problem hiding this comment.
can we put all these under a "Nested AD" testset?
|
closing as stale |
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Outdated suggestions cannot be applied.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.
This incorporates parts of #77 and fixes #820 by implementing the adjoint for the adjoint, i.e. nested derivative, for
getindex, thus facilitating higher order AD.I cannot asses which parts of #77 that I left out are further necessary, but this works for
getindexand other methods depending on it (e.g.sum).