JuliaDiff / JuliaDiff/Diffractor.jl

Wrong derivative of exp(cos(x))

Open
#294 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Julia
Stars
453
Forks
33
PR merge metrics
No merged PRs in 30d

Description

It seems to me that there is a problem with cos derivatives in composite functions:

julia> D = (f, x) -> AbstractDifferentiation.derivative(DiffractorForwardBackend(), f, x)[1];

julia> D(x -> exp(cos(x)), 1.0)
-1.952024889228184

julia> exp(cos(1.0)) * (-sin(1.0))
-1.4444065708474794

julia> exp(sin(1.0)) * (-sin(1.0))
-1.952024889228184

Surprisingly, everything is OK with sin derivatives:

julia> D(x -> exp(sin(x)), 1.0)
1.253380767493447

julia> exp(sin(1.0)) * cos(1.0)
1.253380767493447

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by running the Julia REPL reproduction in the issue with DiffractorForwardBackend and compare the derivative of exp(cos(x)) with the stated analytic value. Trace how cos is handled inside the composite derivative path, then verify that the result matches -1.4444065708474794 while the existing exp(sin(x)) behavior remains correct.

Written by the indexing model from the issue text.

Assessment

Tech stack
julia
Domain
devtools
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Clearly specified
Newbie friendliness
52/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.