simple fix for #140 (#142)

* simple fix for #140

* test for issue #140

* print test name for issue 140

* a fix of #140 on Julia 1.0; going back to the old stream of random numbers

Author: mateuszbaran

src/derivatives/broadcast.jl

@@ -276,10 +276,10 @@ for (M, f, arity) in DiffRules.diffrules()
         end
         for A in ARRAY_TYPES
             @eval begin
-                @inline materialize(bc::RDBroadcasted{typeof($M.$f), <:Tuple{$A,TrackedArray}}) = _materialize(bc.f, bc.args)
-                @inline materialize(bc::RDBroadcasted{typeof($M.$f), <:Tuple{TrackedArray, $A}}) = _materialize(bc.f, bc.args)
-                @inline materialize(bc::RDBroadcasted{typeof($M.$f), <:Tuple{$A, TrackedReal}}) = _materialize(bc.f, bc.args)
-                @inline materialize(bc::RDBroadcasted{typeof($M.$f), <:Tuple{TrackedReal,$A}}) = _materialize(bc.f, bc.args)
+                @inline materialize(bc::RDBroadcasted{typeof($M.$f), <:Tuple{$A{<:Number},TrackedArray}}) = _materialize(bc.f, bc.args)
+                @inline materialize(bc::RDBroadcasted{typeof($M.$f), <:Tuple{TrackedArray, $A{<:Number}}}) = _materialize(bc.f, bc.args)
+                @inline materialize(bc::RDBroadcasted{typeof($M.$f), <:Tuple{$A{<:Number}, TrackedReal}}) = _materialize(bc.f, bc.args)
+                @inline materialize(bc::RDBroadcasted{typeof($M.$f), <:Tuple{TrackedReal,$A{<:Number}}}) = _materialize(bc.f, bc.args)
             end
         end
         for R in REAL_TYPES

src/derivatives/elementwise.jl

@@ -337,10 +337,10 @@ for (F, broadcast_f) in ((typeof(+), :broadcast_plus),
     end
     for A in ARRAY_TYPES
         @eval begin
-            @inline Base.broadcast(::$F, x::TrackedArray{X,D}, y::$A) where {X,D} = $(broadcast_f)(x, y, D)
-            @inline Base.broadcast(::$F, x::$A, y::TrackedArray{Y,D}) where {Y,D} = $(broadcast_f)(x, y, D)
-            @inline Base.broadcast(::$F, x::TrackedReal{X,D}, y::$A) where {X,D} = $(broadcast_f)(x, y, D)
-            @inline Base.broadcast(::$F, x::$A, y::TrackedReal{Y,D}) where {Y,D} = $(broadcast_f)(x, y, D)
+            @inline Base.broadcast(::$F, x::TrackedArray{X,D}, y::$A{<:Real}) where {X,D} = $(broadcast_f)(x, y, D)
+            @inline Base.broadcast(::$F, x::$A{<:Real}, y::TrackedArray{Y,D}) where {Y,D} = $(broadcast_f)(x, y, D)
+            @inline Base.broadcast(::$F, x::TrackedReal{X,D}, y::$A{<:Real}) where {X,D} = $(broadcast_f)(x, y, D)
+            @inline Base.broadcast(::$F, x::$A{<:Real}, y::TrackedReal{Y,D}) where {Y,D} = $(broadcast_f)(x, y, D)
         end
     end
     for R in REAL_TYPES

test/api/GradientTests.jl

@@ -172,6 +172,11 @@ function test_ternary_gradient(f, a, b, c)
     end
 end
 
+# issue https://github.com/JuliaDiff/ReverseDiff.jl/issues/140
+nested_array_mul_140(x) = sum(sum(x[1] * [[x[2], x[3]]]))
+test_println("Issue #140", nested_array_mul_140)
+test_unary_gradient(nested_array_mul_140, [1.0, 2.0, 1.0, -2.4, 4.0])
+
 for f in DiffTests.MATRIX_TO_NUMBER_FUNCS
     test_println("MATRIX_TO_NUMBER_FUNCS", f)
     test_unary_gradient(f, rand(5, 5))