Add Fastmath operators #388
Merged
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
MikeInnes
reviewed
Nov 1, 2019
src/lib/number.jl
Outdated
|
|
||
| fastf = fast_op[f] | ||
|
|
||
| if unarydx != nothing |
There was a problem hiding this comment.
This seems like a strange way to write it. Why not just test hasdiffrule directly here?
MikeInnes
reviewed
Nov 1, 2019
src/lib/number.jl
Outdated
|
|
||
| DiffRules._abs_deriv(x::Complex) = x/abs(x) | ||
| fastmath() |
There was a problem hiding this comment.
Why split this out rather than using a top-level loop as above?
|
This is very useful, thanks! There are quite a few whitespace changes unrelated to the main issue being fixed. While many are useful it would be better to have them as a separate PR, to make review easier. |
|
I have updated the code with given suggestions |
MikeInnes
reviewed
Nov 4, 2019
src/lib/number.jl
Outdated
|
|
||
| # Function add adjoint for Fastmath operations | ||
| # DiffRules is used to find derivatives for each operation | ||
| for f in keys(fast_op) |
There was a problem hiding this comment.
Suggested change
| for f in keys(fast_op) | |
| for (f, fastf) in fast_op |
|
👍 Thanks! bors r+ |
bors bot
added a commit
that referenced
this pull request
Nov 4, 2019
388: Add Fastmath operators r=MikeInnes a=Satya758 Fixes issue #90 1. Basic idea is to apply @adjoint macro to all defined fastmath operators. 2. This is done by looping over fastmath [operators](https://github.com/JuliaLang/julia/blob/master/base/fastmath.jl#L31) and then retrieving defined differentiation expression for each from [DiffRules](https://github.com/JuliaDiff/DiffRules.jl/blob/master/src/rules.jl). 3. Using differentiation expression create adjoint function simillar to https://github.com/FluxML/Zygote.jl/blob/master/src/lib/number.jl#L7 I have added bunch of test cases covering unary and binary operators treating normal operators as a expected value. Co-authored-by: Satya <satyap.kommaraju@gmail.com>
Build succeeded |
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.
Fixes issue #90
I have added bunch of test cases covering unary and binary operators treating normal operators as a expected value.