upgrade CUDA (#153)

Project.toml

@@ -1,7 +1,7 @@
 name = "OMEinsum"
 uuid = "ebe7aa44-baf0-506c-a96f-8464559b3922"
 authors = ["Andreas Peter <[email protected]>"]
-version = "0.7.2"
+version = "0.7.3"
 
 [deps]
 AbstractTrees = "1520ce14-60c1-5f80-bbc7-55ef81b5835c"
@@ -19,7 +19,7 @@ TupleTools = "9d95972d-f1c8-5527-a6e0-b4b365fa01f6"
 [compat]
 AbstractTrees = "0.3, 0.4"
 BatchedRoutines = "0.2"
-CUDA = "3.10"
+CUDA = "4"
 ChainRulesCore = "1"
 Combinatorics = "1.0"
 MacroTools = "0.5"

src/cueinsum.jl

@@ -5,7 +5,7 @@ _unwrap(x::LinearAlgebra.Adjoint{T,<:CuArray{T}}) where T = CuArray(x)
 _unwrap(x::LinearAlgebra.Transpose{T,<:CuArray{T}}) where T = CuArray(x)
 _unwrap(x::CuArray) = x
 
-asarray(x, arr::CuArray) where T = CuArray(fill(x, ()))
+asarray(x, arr::CuArray) = CuArray(fill(x, ()))
 asarray(x::AbstractArray, y::CuArray) = x
 asscalar(x::DenseCuArray) = Array(x)[]
 
@@ -64,7 +64,7 @@ end
     Expr(:tuple, ids...)
 end
 
-function expanddims!(::Val{ixs}, ::Val{iy}, x, y) where {LT,ixs,iy}
+function expanddims!(::Val{ixs}, ::Val{iy}, x, y) where {ixs,iy}
     nthreads = 256
     nblocks = cld(prod(size(x)), nthreads)
     CIS = CartesianIndices(x)