Scalar Indexing Error: multiplying matrix by scalar

Specific Domains GPU
2

I think the problem here is that your scalar k=0.25 is Float64, while the arrays are Float32. This promotes the array (which you probably don’t want, for performance reasons), and that leads to mixed element types in the gradient, which is what’s giving an error:

julia> using CUDA, LinearAlgebra

julia> CUDA.allowscalar(false)

julia> dot(CuArray([1f0, 2f0]), CuArray([3f0, 4f0]))
11.0f0

julia> dot(CuArray([1f0, 2f0]), CuArray([3.0, 4.0]))
ERROR: Scalar indexing is disallowed.
...
Stacktrace:
...
 [5] dot(x::CuArray{Float32, 1, CUDA.Mem.DeviceBuffer}, y::CuArray{Float64, 1, CUDA.Mem.DeviceBuffer})

julia> using Zygote

julia> gradient(x -> sum(x * 3f0), CuArray([1f0, 2f0]))[1]
2-element CuArray{Float32, 1, CUDA.Mem.DeviceBuffer}:
 3.0
 3.0

julia> gradient(x -> sum(x * 3.0), CuArray([1f0, 2f0]))[1]
ERROR: Scalar indexing is disallowed.
...
  [5] dot(x::CuArray{Float64, 1, CUDA.Mem.DeviceBuffer}, y::CuArray{Float32, 1, CUDA.Mem.DeviceBuffer})