[WIP] Add ChainRules differentiation rules for expv #51
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
sethaxen
commented
Dec 20, 2020
| ∂b = Thunk() do | ||
| # using similar is necessary to ensure type-stability | ||
| b̄ = similar(b) | ||
| _copyto!(b̄, expv(t', A', Δw; kwargs...)) |
There was a problem hiding this comment.
adjoint is currently not part of the required interface for generic A. How should we handle this?
- Make it a required part of the interface
- Make it an optional requirement for
A(i.,e. "if you want to work with Zygote, you also need to defineadjoint") - restrict this
rruletoA::AbstractMatrix(this may be necessary anyways, as for non-matrixA, its differential type is unknown.).
There was a problem hiding this comment.
"if you want to work with Zygote, you also need to define adjoint"
I think that's entirely reasonable.
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Outdated suggestions cannot be applied.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.
Following #40, this PR implements forward- and reverse-mode rules for
expv, adding compatibility with Zygote and perhaps making ForwardDiff2's differentiation of these rules faster.This is still marked as a WIP until I figure out sensitivities of the matrix
A.