Skip to content

Optimize TopGroups.merge() maxScore computation for relevance sorting #15879

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
gaobinlong wants to merge 2 commits into
base: main
Choose a base branch
from
+13 −4
Open

Optimize TopGroups.merge() maxScore computation for relevance sorting #15879

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
90c974b
Optimize TopGroups.merge() maxScore computation for relevance sorting
gaobinlong Mar 26, 2026
44d5906
Merge main
gaobinlong Apr 8, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions lucene/CHANGES.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -291,6 +291,8 @@ Optimizations

* GITHUB#15779: Improve BytesRefHash.add performance by optimize rehash operation (tyronecai)

* GITHUB#15879: Optimize TopGroups.merge() maxScore computation for relevance sorting (Binlong Gao)

* GITHUB#15886: Cache frozen FieldType to skip redundant schema validation. (Tim Brooks)

Bug Fixes
Expand Down
15 changes: 11 additions & 4 deletions lucene/grouping/src/java/org/apache/lucene/search/grouping/TopGroups.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -153,7 +153,8 @@ public static <T> TopGroups<T> merge(
final GroupDocs<T>[] mergedGroupDocs = new GroupDocs[numGroups];

final TopDocs[] shardTopDocs;
if (docSort.equals(Sort.RELEVANCE)) {
final boolean sortByRelevance = docSort.equals(Sort.RELEVANCE);
if (sortByRelevance) {
shardTopDocs = new TopDocs[shardGroups.length];
} else {
shardTopDocs = new TopFieldDocs[shardGroups.length];
Expand Down Expand Up @@ -187,7 +188,7 @@ public static <T> TopGroups<T> merge(
}
*/

if (docSort.equals(Sort.RELEVANCE)) {
if (sortByRelevance) {
shardTopDocs[shardIDX] =
new TopDocs(shardGroupDocs.totalHits(), shardGroupDocs.scoreDocs());
} else {
Expand All @@ -200,15 +201,21 @@ public static <T> TopGroups<T> merge(
shardTopDocs[shardIDX].scoreDocs[i].shardIndex = shardIDX;
}

maxScore = nonNANmax(maxScore, shardGroupDocs.maxScore());
if (!sortByRelevance) {
maxScore = nonNANmax(maxScore, shardGroupDocs.maxScore());
}
assert shardGroupDocs.totalHits().relation() == Relation.EQUAL_TO;
totalHits += shardGroupDocs.totalHits().value();
scoreSum += shardGroupDocs.score();
}

final TopDocs mergedTopDocs;
if (docSort.equals(Sort.RELEVANCE)) {
if (sortByRelevance) {
mergedTopDocs = TopDocs.merge(docOffset + docTopN, shardTopDocs);
// When sorting by relevance, the highest-scoring doc is first, so we can
// derive maxScore directly instead of accumulating across shards.
maxScore =
mergedTopDocs.scoreDocs.length == 0 ? Float.NaN : mergedTopDocs.scoreDocs[0].score;
} else {
mergedTopDocs = TopDocs.merge(docSort, docOffset + docTopN, (TopFieldDocs[]) shardTopDocs);
}
Expand Down
Loading