Fix 1.5 compatibility (#116)

* fix 1.5 compatibility

* update compatibility

* update project.toml

* rm 1.5 ci

Project.toml

@@ -1,7 +1,7 @@
 name = "OMEinsum"
 uuid = "ebe7aa44-baf0-506c-a96f-8464559b3922"
 authors = ["Andreas Peter <[email protected]>"]
-version = "0.4.7"
+version = "0.4.8"
 
 [deps]
 AbstractTrees = "1520ce14-60c1-5f80-bbc7-55ef81b5835c"
@@ -11,21 +11,19 @@ ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
 Combinatorics = "861a8166-3701-5b0c-9a16-15d98fcdc6aa"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 MacroTools = "1914dd2f-81c6-5fcd-8719-6d5c9610ff09"
-PkgBenchmark = "32113eaa-f34f-5b0d-bd6c-c81e245fc73d"
 Requires = "ae029012-a4dd-5104-9daa-d747884805df"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 TupleTools = "9d95972d-f1c8-5527-a6e0-b4b365fa01f6"
 
 [compat]
 AbstractTrees = "0.3"
 BatchedRoutines = "0.2"
-CUDA = "2, 3.1"
-ChainRulesCore = "1.0"
+CUDA = "3.4"
+ChainRulesCore = "1.3"
 Combinatorics = "1.0"
 MacroTools = "0.5"
-PkgBenchmark = "0.2"
 Requires = "0.5, 1"
-TupleTools = "1.2"
+TupleTools = "1.2, 1.3"
 julia = "1"
 
 [extras]

src/contractionorder/greedy.jl

@@ -94,7 +94,7 @@ function _tree_greedy(incidence_list::IncidenceList{VT,ET}, log2_edge_sizes; met
 end
 
 function contract_pair!(incidence_list, vi, vj, log2_edge_sizes)
-    log2dim(legs) = sum(l->log2_edge_sizes[l], legs, init=0)
+    log2dim(legs) = isempty(legs) ? 0 : sum(l->log2_edge_sizes[l], legs)  # for 1.5, you need this patch because `init` kw is not allowed.
     @assert vj > vi
     # compute time complexity and output tensor
     legsets = analyze_contraction(incidence_list, vi, vj)
@@ -186,13 +186,13 @@ function analyze_contraction(incidence_list::IncidenceList{VT,ET}, vi::VT, vj::V
 end
 
 function greedy_loss(::MinSpaceOut, incidence_list, log2_edge_sizes, vi, vj)
-    log2dim(legs) = sum(l->log2_edge_sizes[l], legs, init=0)
+    log2dim(legs) = isempty(legs) ? 0 : sum(l->log2_edge_sizes[l], legs)  # for 1.5, you need this patch because `init` kw is not allowed.
     legs = analyze_contraction(incidence_list, vi, vj)
     log2dim(legs.l01)+log2dim(legs.l02)+log2dim(legs.l012)
 end
 
 function greedy_loss(::MinSpaceDiff, incidence_list, log2_edge_sizes, vi, vj)
-    log2dim(legs) = sum(l->log2_edge_sizes[l], legs, init=0)
+    log2dim(legs) = isempty(legs) ? 0 : sum(l->log2_edge_sizes[l], legs)  # for 1.5, you need this patch because `init` kw is not allowed.
     legs = analyze_contraction(incidence_list, vi, vj)
     D1,D2,D12,D01,D02,D012 = log2dim.(getfield.(Ref(legs), 1:6))
     exp2(D01+D02+D012) - exp2(D1+D01+D12) - exp2(D2+D02+D12)  # out - in

src/cueinsum.jl

@@ -7,7 +7,7 @@ asscalar(x::DenseCuArray) = Array(x)[]
 Base.Array(x::Base.ReshapedArray{T,0,<:CuArray}) where T = Array(x.parent)
 
 function get_output_array(xs::NTuple{N, DenseCuArray{<:Any,M} where M}, size; has_repeated_indices=true) where N
-    out = CUDA.zeros(promote_type(map(eltype,xs)...), size)
+    CUDA.zeros(promote_type(map(eltype,xs)...), size)
 end
 
 CUDA.cudaconvert(A::EinArray{T}) where T = EinArray{T}(cudaconvert.(A.xs), A.x_indexers, A.y_indexer, A.size, A.ICIS, A.OCIS)
@@ -31,13 +31,19 @@ function loop_einsum!(code::EinCode{ixs, iy},
     A = einarray(code, xs, size_dict)
     if NO == length(iy)
         y = reshape(y, fill(1, ndims(A)-NO)...,size(y)...)
-        dropdims(Base.mapreducedim!(x->x, +, y, A), dims=(1:ndims(A)-NO...,))
+        raw = Base.mapreducedim!(x->x, +, y, A)
+        if ndims(A)-NO > 0  # fix 1.7 compatibility
+            raw = dropdims(raw, dims=(1:ndims(A)-NO...,))
+        end
+        return raw
     else
         y_ = CUDA.zeros(T, size(A)[end-NO+1:end]...)
         y_ = reshape(y_, fill(1, ndims(A)-NO)...,size(y_)...)
         raw = Base.mapreducedim!(x->x, +, y_, A)
-        out = dropdims(raw, dims=(1:ndims(A)-NO...,))
-        expanddims!(EinCode{((iy_...,),), iy}(), out, y)
+        if ndims(A)-NO > 0  # fix 1.7 compatibility
+            raw = dropdims(raw, dims=(1:ndims(A)-NO...,))
+        end
+        return expanddims!(EinCode{((iy_...,),), iy}(), raw, y)
     end
 end
 

src/interfaces.jl

@@ -182,12 +182,14 @@ function dynamic_einsum(ixs, xs, iy, size_dict)
     if length(ixs) == 1
         einsum(rule, ixs[1], iy, xs[1], size_dict)
     else
-        einsum(rule, ixs, iy, xs, size_dict)
+        einsum(rule, ixs, iy, (xs...,), size_dict)
     end
 end
 
+dynamic_einsum(::EinCode{ixs, iy}, xs; kwargs...) where {ixs, iy} = dynamic_einsum(ixs, xs, iy; kwargs...)
+
 # the fallback
 function einsum(::DefaultRule, ixs, iy, xs, size_dict)
     @debug "DefaultRule loop_einsum" ixs => iy size.(xs)
-    loop_einsum(EinCode{ixs, iy}(), xs, size_dict)
-end
+    loop_einsum(EinCode{ixs, iy}(), (xs...,), size_dict)
+end

src/utils.jl

@@ -8,7 +8,7 @@ If a `parent` is supplied, it will try to match the parent array type.
 """
 asarray(x) = fill(x, ())
 asarray(x::AbstractArray) = x
-asarray(x, arr::Array) = fill(x, ())
+asarray(x, arr::AbstractArray) = fill(x, ())
 asarray(x::AbstractArray, y::Array) = x
 asscalar(x) = x
 asscalar(x::AbstractArray) = x[]