JuliaDiff / JuliaDiff/ChainRules.jl

cholesky decomposition

Open
#414 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Julia
Stars
475
Forks
98
PR merge metrics
No merged PRs in 30d

Description

I'm getting incorrect results when working with the rrule for cholesky where A <: LinearAlgebra.HermOrSym

Passing the input matrix through Matrix fixes the issue. The mul! fix relates to this issue.

using Zygote, ChainRules,LinearAlgebra
# Example matrix
A = [2. -1. 0.0; -1. 2. -1.; 0. -1. 2. ]

import LinearAlgebra.mul!
LinearAlgebra.mul!(C, ::ChainRulesCore.ZeroTangent, ::Any, ::Any, b) = C *=b

# produces zeros
Zygote.jacobian(a -> cholesky(Hermitian(a)).L , A)[1]

# both of the following produce expected result
Zygote.jacobian(a -> cholesky(Matrix(Hermitian(a))).L , A)[1]
Zygote.jacobian(a -> cholesky(a).L , A)[1]

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 with the rrule for cholesky where A <: LinearAlgebra.HermOrSym in src/rulesets/LinearAlgebra/factorization.jl around line 452, then reproduce the Zygote.jacobian examples from the issue. Compare the Hermitian path with the Matrix and plain cholesky paths, including the linked ChainRulesCore mul! issue. Done means the Hermitian cholesky Jacobian no longer produces zeros.

Written by the indexing model from the issue text.

Assessment

Tech stack
julia
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.