You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: docs/src/opting_out_of_rules.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -27,7 +27,7 @@ end
27
27
That is a fairly reasonable `rrule` for the vast majority of cases.
28
28
29
29
You might have a custom array type for which you could write a faster rule.
30
-
For example, the pullback for summing a [`SkewSymmetric` (anti-symmetric)](https://en.wikipedia.org/wiki/Skew-symmetric_matrix) matrix can be optimized to basically be `Diagonal(fill(ȳ, size(x,1)))`.
30
+
For example, the pullback for summing a [`SkewSymmetric` (anti-symmetric)](https://en.wikipedia.org/wiki/Skew-symmetric_matrix) matrix can be optimized all the way to `ZeroTangent()` as all off-diagonals cancel and the diagaonal is structurally zeros.
31
31
To do that, you can indeed write another more specific [`rrule`](@ref).
32
32
But another case is where the AD system itself would generate a more optimized case.
0 commit comments