benwtrent commented on issue #13403: URL: https://github.com/apache/lucene/issues/13403#issuecomment-2128355186
> If the number of dimensions are reduced, you don't even need to quantize them? Does PCA work in non-Euclidean spaces? Does it work on out-of-domain queries? If PCA ends up being cheaper and better than PQ, I say let's do it. The reason I reach for PQ is that it's fairly standard right now. Maybe it is for a reason? As far as doing PQ at merge, there needs to be experimentation here. Having to redo code books on merge and re-quantizing will increase merge time significantly, no matter how many threads we throw at it. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
