Michael Abbott

Can you give more context? I don't think it makes much sense to track gradients with respect to undefined quantities.

Have not read this closely, but if I understand right on first look, it doesn't significantly change the logic of basic case? Am curious whether you see such changes allowing...

Re breaking, I marked #39 as being v0.4, although I'm not entirely sure we must. Maybe 0.4 should contain both changes? No great rush for #39. And perhaps it's worth...

Unfortunately Tullio gets this wrong too -- it infers the type as being that of the product without the sum. But perhaps it can learn to do better. ```julia julia>...

Not sure, I know nothing about this package. I doubt that this can vectorize, given that it wraps a string. ```julia julia> a[1] |> isbits false julia> a[1] |> dump...

In case someone is watching this: index shifts of the form `@einsum A[i] := X[i+:offset]` seem to be broken on Julia 1.0. This feature never made it into the readme...

I think it's to do with the global scope issue (see for example [here](https://discourse.julialang.org/t/new-scope-solution/16707)). That's why it works fine inside a function, but not here: ``` julia> M = collect(reshape(1:4,2,2))...

I agree that's the standard meaning of Einstein notation. But it would be far from simple to make this package do that, and if you wanted it to be efficient,...

Before: ``` julia> @time using Zygote; @time gradient(x -> sum(abs2, x .+ x')^3, [1,2,3]); 1.376268 seconds (3.42 M allocations: 247.586 MiB, 1.79% gc time, 4.58% compilation time: 38% of which...

> It would be great if someone worked on a "TupleCast" package for this sort of thing. There is [StructArrays](https://github.com/JuliaArrays/StructArrays.jl), with which @piever showed me this: ```julia using StructArrays function...