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
So A f¨ B implies pairwise (A[1 1] f B[1 1]) (B[1 2] f B[2 2]).....
In ivy A f@ B does (A f B[1]) A f B[2]. @f@ does a full 'cross product' of all elements.
Most of the built in ops will operate elementwise anyway, but for those that don't (like ,), or for user defined ops, I've had to resort to reshaping as a flat vector, pairing them up, applying a unary function that takes the pairs, and then reshaping the result.
Is an elementwise version of @ achievable?
The text was updated successfully, but these errors were encountered:
Ivy's
@
cannot (I think) currently do element-wise each.e.g.
In TryAPL:
So
A f¨ B
implies pairwise(A[1 1] f B[1 1]) (B[1 2] f B[2 2])....
.In ivy
A f@ B
does(A f B[1]) A f B[2]
.@f@
does a full 'cross product' of all elements.Most of the built in ops will operate elementwise anyway, but for those that don't (like
,
), or for user defined ops, I've had to resort to reshaping as a flat vector, pairing them up, applying a unary function that takes the pairs, and then reshaping the result.Is an elementwise version of
@
achievable?The text was updated successfully, but these errors were encountered: