CombinedFieldQuery computes its max score as score(+Infinity, 1), which is NaN under AxiomaticF2LOG/F2EXP, so top-k pruning silently drops the best hits
Finding LFV-253Area: search / CombinedFieldQuery (BM25F) max-score bound under dynamic pruning (CombinedFieldScorer.getMaxScore as used by MaxScoreBulkScorer)Priority: mediumImpact: silent wrong resultsNovelty: new
AI-assisted finding, confirmed by automated checks, not yet triaged by a human Lucene committer. This site files nothing upstream.
What it shows
You score a query with CombinedFieldQuery under the core AxiomaticF2LOG (or AxiomaticF2EXP) similarity, which the class javadoc counts as compatible because it keeps the default computeNorm, put it in a BooleanQuery next to another scoring clause such as a TermQuery, and expect IndexSearcher#search(q, 10) to return the same top hits as an exhaustive search. CombinedFieldScorer reports score(Float.POSITIVE_INFINITY, 1) as its max score, although SimScorer#score requires a finite freq and documents score(Float.MAX_VALUE, 1) as the bound; for the F2 similarities that is Inf/Inf = NaN, and MaxScoreBulkScorer, for which NaN is never below the minimum competitive score, then makes the other clause required. Documents that match only the CombinedFieldQuery are skipped: in a 3000-document index the default search returns doc 0 (score 2.1975672) as the top hit instead of doc 2000 (6.383921), and the upstream oracle CheckHits.checkTopScores fails.
The test shows that CombinedFieldQuery under AxiomaticF2LOG reports a NaN max score (score(+Infinity, 1)), so top-k pruning in a SHOULD CombinedFieldQuery + SHOULD TermQuery disjunction skips the best hit.
Expected vs actual
Expected the pruned search (TopScoreDocCollectorManager(1, 1)) to return the exhaustive top hit, doc 256; actual: doc 0 (top hit with pruning expected:<256> but was:<0>), with -ea and -da.
Files
Run it
- Tested on Lucene main e1faeba08bde (2026-11-02): still reproduces.
- Tested on Lucene main 645d4fe14a20 (2026-10-28): reproduced.
- Recorded outcome of run 1 on Lucene main 645d4fe14a20: 1 test(s) run, 1 failed.
- How to read the outcome: a failing test is the finding reproducing (the failure message states expected against actual).
- Seeds: a rerun with the same seed replays the same random choices. Assertions off (-da -dsa -Dtests.asserts=false) is the production-like setting.
# Lucene checkout at the tested commit, built with: ./gradlew --no-daemon assemble (JDK 25)
LUCENE=/path/to/lucene; V=12.0.0-SNAPSHOT
CP=$LUCENE/lucene/core/build/libs/lucene-core-$V.jar
CP=$CP:$LUCENE/lucene/codecs/build/libs/lucene-codecs-$V.jar
CP=$CP:$LUCENE/lucene/test-framework/build/libs/lucene-test-framework-$V.jar
CP=$CP:randomizedtesting-runner-2.9.2.jar:junit-4.13.2.jar:hamcrest-3.0.jar # test-framework dependencies (Maven Central)
CP=$CP:junit-platform-console-standalone-6.1.3.jar # JUnit Platform console launcher (Maven Central)
javac -d classes -cp "$CP" $(find src -name '*.java')
# run 1: assertions on, seed DEADBEEF
java -ea -esa --add-modules jdk.incubator.vector --enable-native-access=ALL-UNNAMED -Dtests.seed=DEADBEEF \
-cp "classes:$CP" org.junit.platform.console.ConsoleLauncher \
execute --select-class org.apache.lucene.search.TestCombinedFieldQueryNaNMaxScoreGradle alternative: copy the test into the matching module's src/test/ tree of the checkout and run ./gradlew -p lucene/<module> test --tests <class> -Ptests.seed=<seed> (add -Ptests.asserts=false for a run without assertions).
Feedback
Open an issue in lucene-ai-bug-hunt. Please do not open issues or pull requests in apache/lucene about this finding unless a Lucene committer agrees it should be filed.
Snapshot of 2026-11-04. AI-assisted; not yet triaged by a human Lucene committer. Feedback: open an issue in lucene-ai-bug-hunt.