Ensure pullback of exp works for immutable arrays (#381)

* Ensure exp cotangent is mutable * Increment version number * Use convert and inplaceable trait
JuliaDiff · Mar 1, 2021 · 99c58a9 · 99c58a9 · sethaxen · Mar 1, 2021
1 parent 439f482
commit 99c58a9
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 2 deletions.
diff --git a/Project.toml b/Project.toml
@@ -1,6 +1,6 @@
 name = "ChainRules"
 uuid = "082447d4-558c-5d27-93f4-14fc19e9eca2"
-version = "0.7.52"
+version = "0.7.53"
 
 [deps]
 ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"

diff --git a/src/rulesets/LinearAlgebra/matfun.jl b/src/rulesets/LinearAlgebra/matfun.jl
@@ -129,7 +129,10 @@ function rrule(::typeof(exp), A0::StridedMatrix{<:BlasFloat})
         A = copy(A0)
         X, intermediates = _matfun!(exp, A)
         function exp_pullback(ΔX)
-            ∂A = _matfun_frechet_adjoint!(exp, ΔX, A, X, intermediates)
+            # Ensures ∂X is mutable. The outer `adjoint` is unwrapped without copy by
+            # the default _matfun_frechet_adjoint!
+            ∂X = ChainRulesCore.is_inplaceable_destination(ΔX) ? ΔX : convert(Matrix, ΔX')'
+            ∂A = _matfun_frechet_adjoint!(exp, ∂X, A, X, intermediates)
             return NO_FIELDS, ∂A
         end
         return X, exp_pullback