You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@lucene.apache.org by "Robert Muir (Jira)" <ji...@apache.org> on 2021/03/15 22:28:00 UTC
[jira] [Commented] (LUCENE-9845) Improve encoding of HNSW graph
offsets
[ https://issues.apache.org/jira/browse/LUCENE-9845?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17302068#comment-17302068 ]
Robert Muir commented on LUCENE-9845:
-------------------------------------
I was looking at this... please correct me if i am wrong, but some of the tradeoffs such as long[] and int[] are in trying to manage sparse and dense representations.
Both cases today seem inefficient in different ways.
I feel like the IndexedDISI already sitting right there in the package is probably ideal? It is already used by docvalues to solve this exact problem.
> Improve encoding of HNSW graph offsets
> --------------------------------------
>
> Key: LUCENE-9845
> URL: https://issues.apache.org/jira/browse/LUCENE-9845
> Project: Lucene - Core
> Issue Type: Wish
> Reporter: Michael Sokolov
> Priority: Major
>
> Today we use a simple {{long[]}} to encode the offset of each document's array of neighbors in the HNSW graph. Instead we should use a data structure that is optimized for encoding an increasing numeric array, like monotonic {{PackedInts}}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@lucene.apache.org
For additional commands, e-mail: issues-help@lucene.apache.org