[GitHub] [lucene] jmazanec15 commented on pull request #12050: Reuse HNSW graph for intialization during merge

2023-02-13 Thread via GitHub
jmazanec15 commented on PR #12050: URL: https://github.com/apache/lucene/pull/12050#issuecomment-1428397749 @jpountz yes that's correct. The random number assignment is no longer going to be the same when merging multiple graphs together, because the segment whose graph is being used to ini

[GitHub] [lucene] jmazanec15 commented on pull request #12050: Reuse HNSW graph for intialization during merge

2023-02-07 Thread via GitHub
jmazanec15 commented on PR #12050: URL: https://github.com/apache/lucene/pull/12050#issuecomment-1421527701 Thanks @benwtrent! -- 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

[GitHub] [lucene] jmazanec15 commented on pull request #12050: Reuse HNSW graph for intialization during merge

2023-01-30 Thread via GitHub
jmazanec15 commented on PR #12050: URL: https://github.com/apache/lucene/pull/12050#issuecomment-1409285148 @benwtrent thanks! I do not mind a coauthor. Was working on the rebase and just finished it. -- This is an automated message from the Apache Git Service. To respond to the message,

[GitHub] [lucene] jmazanec15 commented on pull request #12050: Reuse HNSW graph for intialization during merge

2023-01-19 Thread GitBox
jmazanec15 commented on PR #12050: URL: https://github.com/apache/lucene/pull/12050#issuecomment-1397643952 Per [this discussion](https://github.com/apache/lucene/pull/12050#discussion_r1061034056), I refactored OnHeapHnswGraph to use a TreeMap to represent the graph structure for levels g

[GitHub] [lucene] jmazanec15 commented on pull request #12050: Reuse HNSW graph for intialization during merge

2023-01-11 Thread GitBox
jmazanec15 commented on PR #12050: URL: https://github.com/apache/lucene/pull/12050#issuecomment-1379247267 @msokolov The main reason I did not do this was to avoid having to modify the ordering of the vectors from the MergedVectorValues. I believe that the ordinals in the graph map to the