JuliaArrays / JuliaArrays/BlockArrays.jl

Broadcasting regression with adjoints in 0.16.22

Open
#237 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Julia
Stars
254
Forks
39
PR merge metrics
No merged PRs in 30d

Description

m = mortar([rand(i,j) for i=1:3, j=2:3])
m .- (1:5)'

In 0.16.22-23:

ERROR: BoundsError: attempt to access 1-element Vector{Int64} at index [2]
Stacktrace:
  [1] getindex
    @ .\array.jl:861 [inlined]
  [2] iterate
    @ C:\Users\sternlab\.julia\packages\BlockArrays\0XNk5\src\blockbroadcast.jl:128 [inlined]
  [3] _zip_iterate_some
    @ .\iterators.jl:358 [inlined]
  [4] _zip_iterate_some (repeats 2 times)
    @ .\iterators.jl:360 [inlined]
  [5] _zip_iterate_all
    @ .\iterators.jl:350 [inlined]
  [6] iterate
    @ .\iterators.jl:341 [inlined]
  [7] macro expansion
    @ C:\Users\sternlab\.julia\packages\BlockArrays\0XNk5\src\blockbroadcast.jl:172 [inlined]
  [8] _generic_blockbroadcast_copyto!(dest::BlockMatrix{Float64, Matrix{Matrix{Float64}}, Tuple{BlockedUnitRange{Vector{Int64}}, BlockedUnitRange{Vector{Int64}}}}, bc::Base.Broadcast.Broadcasted{BlockArrays.BlockStyle{2}, Tuple{BlockedUnitRange{Vector{Int64}}, BlockedUnitRange{Vector{Int64}}}, typeof(-), Tuple{BlockMatrix{Float64, Matrix{Matrix{Float64}}, Tuple{BlockedUnitRange{Vector{Int64}}, BlockedUnitRange{Vector{Int64}}}}, LinearAlgebra.Adjoint{Int64, UnitRange{Int64}}}})
    @ BlockArrays C:\Users\sternlab\.julia\packages\BlockArrays\0XNk5\src\blockbroadcast.jl:148
  [9] copyto!(dest::BlockMatrix{Float64, Matrix{Matrix{Float64}}, Tuple{BlockedUnitRange{Vector{Int64}}, BlockedUnitRange{Vector{Int64}}}}, bc::Base.Broadcast.Broadcasted{BlockArrays.BlockStyle{2}, Tuple{BlockedUnitRange{Vector{Int64}}, BlockedUnitRange{Vector{Int64}}}, typeof(-), Tuple{BlockMatrix{Float64, Matrix{Matrix{Float64}}, Tuple{BlockedUnitRange{Vector{Int64}}, BlockedUnitRange{Vector{Int64}}}}, LinearAlgebra.Adjoint{Int64, UnitRange{Int64}}}})
    @ BlockArrays C:\Users\sternlab\.julia\packages\BlockArrays\0XNk5\src\blockbroadcast.jl:190
 [10] copy
    @ .\broadcast.jl:885 [inlined]
 [11] materialize(bc::Base.Broadcast.Broadcasted{BlockArrays.BlockStyle{2}, Nothing, typeof(-), Tuple{BlockMatrix{Float64, Matrix{Matrix{Float64}}, Tuple{BlockedUnitRange{Vector{Int64}}, BlockedUnitRange{Vector{Int64}}}}, LinearAlgebra.Adjoint{Int64, UnitRange{Int64}}}})
    @ Base.Broadcast .\broadcast.jl:860
 [12] top-level scope
    @ Untitled-4:8

This works fine in 0.16.21.

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

Reproduce the broadcasting expression from the issue in Julia 0.16.22-23 and compare it with 0.16.21. Start in src/blockbroadcast.jl around the iterate and _generic_blockbroadcast_copyto! locations shown in the stack trace. Done means the adjoint expression no longer raises the BoundsError and retains the behavior seen in 0.16.21.

Written by the indexing model from the issue text.

Assessment

Tech stack
julia
Domain
data
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.