Michael Abbott

Inconsistencies in `setindex!`, and thus `vcat`

5

It would be nice if these all worked: ```julia julia> reduce(vcat, [OffsetArray([i; 10i;;], 3,4) for i in 1:2]) 4×1 Matrix{Int64}: 1 10 2 20 julia> ans == vcat([OffsetArray([i; 10i;;], 3,4)...

From https://github.com/FluxML/Zygote.jl/issues/1428, these two should make the same `ProjectTo{NoTangent}()`: ```julia julia> using ChainRulesCore, SparseArrays julia> ProjectTo(diagm([true, false])) ProjectTo{NoTangent}() julia> ProjectTo(spdiagm([true, false])) ProjectTo{SparseMatrixCSC}(element = ProjectTo{NoTangent}(), axes = (Base.OneTo(2), Base.OneTo(2)), rowval =...

Cannot generate `frule` seed via `one(x)`

1

Calling `frule_via_ad(cfg, (NoTangent(), one(x)), f, x)` to work out the derivative works for numbers but not in general. So this path: https://github.com/JuliaDiff/ChainRules.jl/blob/5855c10bdbe691fc07822752f5b5865b9cea44d3/src/rulesets/Base/broadcast.jl#L104-L110 fails if broadcasting over an array of `Ref`,...

Recurse into some arrays of arrays?

1

I wonder whether `adapt` should treat an Array of Arrays as a container, like a Tuple, rather than storage to be converted. Convert the innermost Array, not the outermost: ```julia...

One-argument `fmapstructure(x)`

9

Maybe this should be called `fstructure(x)` since there's no `map`. The point is to strip all the types for saving etc. Apparently others believe this method should exist: https://discourse.julialang.org/t/how-to-load-bson-file-of-the-model-build-with-flux-0-12-10-to-use-with-flux-0-13-flux-diagonal-deprecated-problem/91588/2 ###...

`@functor` is confused about properties vs fields

5

It was noticed in https://github.com/FluxML/Flux.jl/issues/2107 that Functors.jl + ProtoStruct.jl doesn't work, as Functors uses `fieldnames` + `getproperty`, which is overloaded by ProtoStruct.jl. https://github.com/FluxML/Functors.jl/blob/v0.3.0/src/functor.jl#L11-L16 This is a bug, Functors should use...

`significand`, `frexp` not defined

1

```julia julia> significand(Float64x2(123.4)) ERROR: MethodError: no method matching significand(::MultiFloat{Float64, 2}) Closest candidates are: significand(::T) where T

This is a draft of a way to start addressing #1334, for comment. It implements what I called level 2 here: https://github.com/EnzymeAD/Enzyme.jl/issues/1334#issuecomment-2016573247 On arrays like these, no change. Natural and...

Maximum Nesting Depth + Julia

4

I came across this page https://github.com/codereport/array-language-comparisons/blob/main/comparisons/leetcode/P1614_Max_Paren_Depth.md For Julia, I'm not too sure what "no row-wise minus reduction" means, but it sounds like you are looking for `maximum(cumsum(...`, something like this:...

Had this locally long ago, but forgot why I didn't commit it. Closes #72, closes https://github.com/mcabbott/TensorCast.jl/issues/64