Michael Abbott

Some linear algebra, via `reinterpret`

2

This is a start on #46.

Add links to alternatives to the readme

4

As discussed here https://discourse.julialang.org/t/how-do-i-know-if-a-package-is-good/82133, it might be nice if this package linked to alternative ways to read CSV files. Really all packages should do this, but this one is what...

Functions like gather/scatter give scalar indexing errors if used on CuArrays without remembering to load NNlibCUDA. Since there is now a very lightweight GPUArraysCore, I think NNlib should depend on...

Add gradients for `conv_bias_act`, and a similar `dense_bias_act`

4

This aims to add gradient definitions for the existing `conv_bias_act`. That is, however, very much WIP, and I don't recommend anyone try to read it just yet. It also adds...

Don't overload `map(f)`

1

Fixes #179

Piracy of `map(f)`

1

This package overwrites Base's behaviour for one-argument `map`, for example: ``` julia> map(println) # calls println() julia> using FillArrays julia> map(println) ERROR: MethodError: reducing over an empty collection is not...

`real(::Fill)` gives an array of arrays?

6

This is surprising: ```julia julia> real(Fill(1+im)) isa Array{ real(fill(1+im)) isa Array{Int,0} true (jl_AeiBcQ) pkg> st Status `/private/var/folders/yq/4p2zwd614y59gszh7y9ypyhh0000gn/T/jl_AeiBcQ/Project.toml` [1a297f60] FillArrays v0.11.7 ``` I guess it comes from this difference, via `broadcast_preserving_zero_d`:...

Apply ChainRulesCore.jl's projection operators

22

ChainRules now has a projection mechanism to preserve, among other things, the structure of structured arrays. This should probably apply to FillArrays. So this PR writes a few methods. However,...

Collision of LazyArrays and FillArrays broadcasting

1

This came up here: https://discourse.julialang.org/t/preventing-collect-on-broadcasting-expressions/57996 ```julia julia> using LazyArrays, FillArrays julia> BroadcastArray(+, fill(1,2), 3) 2-element BroadcastVector{Int64, typeof(+), Tuple{Vector{Int64}, Int64}}: 4 4 julia> broadcast(+, Fill(1,2), 3) 2-element Fill{Int64, 1, Tuple{Base.OneTo{Int64}}} =...

Lazy broadcasting bug?

4

I just ran into the following error: ``` julia> R = rand(3); s = 1; julia> using LazyArrays: lazy julia> @. lazy(2 * s * R) 3-element LazyArrays.BroadcastArray{Float64,1,Base.Broadcast.Broadcasted{Base.Broadcast.DefaultArrayStyle{1},Tuple{Base.OneTo{Int64}},typeof(*),Tuple{Int64,Int64,Array{Float64,1}}}}: 0.9510075392608135 1.3657778460965129...