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
No worries I'm just plyaing for now. For the record, it's easy to work around by extending MergingDigest.centroids and returning an empty list when size() returns 0.
By the way some initial testing seems to suggest MergingDigest is slower than AVLTreeDigest (at least on a uniform distribution) while I was expecting this impl to be faster?
I haven't touched speed questions yet. There are knobs to turn still. I would think it could be faster, but hard to say. The serious goal is to provide a basis for translation to C. That means we can't have any dynamic allocation that assumes GC.
When calling
centroids
on an empty MergingDigest, a list containing a single element with NaN as a centroid is returned instead of an empty list:The text was updated successfully, but these errors were encountered: