You are viewing a plain text version of this content. The canonical link for it is here.
Posted to notifications@asterixdb.apache.org by AsterixDB Code Review <do...@asterix-gerrit.ics.uci.edu> on 2026/10/05 17:55:22 UTC

[XL] Change in asterixdb[master]: [NO ISSUE][RT] CLUSTER BY: draw per partition in oversampling

From Shahrzad Shirazi <sh...@ucr.edu>:

Shahrzad Shirazi has uploaded this change for review. ( https://asterix-gerrit.ics.uci.edu/c/asterixdb/+/21996?usp=email )


Change subject: [NO ISSUE][RT] CLUSTER BY: draw per partition in oversampling
......................................................................

[NO ISSUE][RT] CLUSTER BY: draw per partition in oversampling

- user model changes: no
- storage format changes: no
- interface changes: no

Details:
The oversampling draw's random number came from the vector, the
seed, the round and the copy index, which restarts on every
partition. Copy n of a duplicated vector got the same number on
every partition, so a point held c times across p partitions had
about c/p independent trials instead of c. With 40 copies over 4
partitions a point sat out all five rounds in about one query in
two hundred, RECLUSTER kept one candidate, and
kmeans-duplicate-init failed intermittently.

The partition now enters the hash, so each row is its own trial.

- Seeded results change where the drawn candidates differ; the
  clusters are equivalent. Expected results for kmeans-seed and
  kmeans-budget-sensitivity regenerated, in results and
  results_cloud; the budget arms still agree byte for byte.
- join-in-from moves from seed 10 to 11: with the new draws,
  seed 10 hits a bad k-means++ start in the cloud layout (two
  centres in one blob), which plain k-means++ does for about 2%
  of seeds on that data.

Ext-ref: MB-73589
Generated-by: Claude Fable 5.1 <no...@anthropic.com>
Change-Id: Iee972a23b0331b5fce8e909fd49637ea3c15d8bf
---
M asterixdb/asterix-app/src/test/resources/runtimets/queries_sqlpp/cluster-by/join-in-from/join-in-from.3.query.sqlpp
M asterixdb/asterix-app/src/test/resources/runtimets/results/cluster-by/kmeans-budget-sensitivity/kmeans-budget-sensitivity.3.adm
M asterixdb/asterix-app/src/test/resources/runtimets/results/cluster-by/kmeans-budget-sensitivity/kmeans-budget-sensitivity.4.adm
M asterixdb/asterix-app/src/test/resources/runtimets/results/cluster-by/kmeans-budget-sensitivity/kmeans-budget-sensitivity.5.adm
M asterixdb/asterix-app/src/test/resources/runtimets/results/cluster-by/kmeans-seed/kmeans-seed.3.adm
M asterixdb/asterix-app/src/test/resources/runtimets/results/cluster-by/kmeans-seed/kmeans-seed.4.adm
M asterixdb/asterix-app/src/test/resources/runtimets/results/cluster-by/kmeans-seed/kmeans-seed.5.adm
M asterixdb/asterix-app/src/test/resources/runtimets/results_cloud/cluster-by/kmeans-budget-sensitivity/kmeans-budget-sensitivity.3.adm
M asterixdb/asterix-app/src/test/resources/runtimets/results_cloud/cluster-by/kmeans-budget-sensitivity/kmeans-budget-sensitivity.4.adm
M asterixdb/asterix-app/src/test/resources/runtimets/results_cloud/cluster-by/kmeans-budget-sensitivity/kmeans-budget-sensitivity.5.adm
M asterixdb/asterix-app/src/test/resources/runtimets/results_cloud/cluster-by/kmeans-seed/kmeans-seed.3.adm
M asterixdb/asterix-app/src/test/resources/runtimets/results_cloud/cluster-by/kmeans-seed/kmeans-seed.4.adm
M asterixdb/asterix-app/src/test/resources/runtimets/results_cloud/cluster-by/kmeans-seed/kmeans-seed.5.adm
M asterixdb/asterix-runtime/src/main/java/org/apache/asterix/runtime/operators/kmeans/KMeansLoopIO.java
M asterixdb/asterix-runtime/src/main/java/org/apache/asterix/runtime/operators/kmeans/KMeansSampleOperatorDescriptor.java
M asterixdb/asterix-runtime/src/test/java/org/apache/asterix/runtime/operators/kmeans/KMeansFingerprintTest.java
16 files changed, 800 insertions(+), 795 deletions(-)



  git pull ssh://asterix-gerrit.ics.uci.edu:29418/asterixdb refs/changes/96/21996/1


-- 
To view, visit https://asterix-gerrit.ics.uci.edu/c/asterixdb/+/21996?usp=email
To unsubscribe, or for help writing mail filters, visit https://asterix-gerrit.ics.uci.edu/settings?usp=email

Gerrit-MessageType: newchange
Gerrit-Project: asterixdb
Gerrit-Branch: master
Gerrit-Change-Id: Iee972a23b0331b5fce8e909fd49637ea3c15d8bf
Gerrit-Change-Number: 21996
Gerrit-PatchSet: 1
Gerrit-Owner: Shahrzad Shirazi <sh...@ucr.edu>