Skip to content

[feature](function) add uniq_theta theta-sketch distinct-count aggregate - #66241

Open
lzydmxy wants to merge 1 commit into
apache:masterfrom
lzydmxy:feat-uniq-theta-upstream
Open

[feature](function) add uniq_theta theta-sketch distinct-count aggregate#66241
lzydmxy wants to merge 1 commit into
apache:masterfrom
lzydmxy:feat-uniq-theta-upstream

Conversation

@lzydmxy

@lzydmxy lzydmxy commented Jul 29, 2026

Copy link
Copy Markdown

Add uniq_theta, a theta-sketch based approximate distinct-count aggregate returning BIGINT, ported from ClickHouse uniqTheta and backed by the Apache DataSketches library (contrib/datasketches-cpp submodule added in #63143). The aggregate reuses the generic agg_state combinator framework so uniq_theta_state / uniq_theta_merge / uniq_theta_union are auto-derived with no extra wiring. Includes BE gtest, FE unit test, and regression suites.

What problem does this PR solve?

Issue Number: close #66240

Related PR: #63143

Problem Summary:
Doris only has HyperLogLog-based approx_count_distinct / ndv, which supports
union but not clean intersection or difference on the serialized state.

This PR adds uniq_theta, an approximate distinct-count aggregate backed by
the Apache DataSketches Theta Sketch (ported from ClickHouse uniqTheta). Theta
sketches add capabilities HLL lacks: set operations (union / intersect /
difference) on the serialized sketch, precompute-once/query-many via mergeable
agg_state, and cross-engine-compatible sketch bytes. Relative error is ~3.125%
(95% confidence, default 4096 nominal entries), exact for small cardinalities.

The contrib/datasketches-cpp submodule and its build wiring were already added
by #63143, so this PR only adds the aggregate itself.

Release note

Add aggregate function uniq_theta, a Theta-sketch (Apache DataSketches) based
approximate distinct-count function, together with its uniq_theta_state /
uniq_theta_merge / uniq_theta_union agg_state combinators.

Check List (For Author)

  • Test

    • Regression test
    • Unit Test
    • Manual test (add detailed scripts or steps below)
    • No need to test or manual test. Explain why:
      • This is a refactor/code format and no logic has been changed.
      • Previous test can cover this change.
      • No code files have been changed.
      • Other reason
  • Behavior changed:

    • No.
    • Yes.
  • Does this need documentation?

Check List (For Reviewer who merge this PR)

  • Confirm the release note
  • Confirm test cases
  • Confirm document
  • Add branch pick label

Add uniq_theta, a theta-sketch based approximate distinct-count aggregate
returning BIGINT, ported from ClickHouse uniqTheta and backed by the Apache
DataSketches library (contrib/datasketches-cpp submodule added in apache#63143). The
aggregate reuses the generic agg_state combinator framework so uniq_theta_state
/ uniq_theta_merge / uniq_theta_union are auto-derived with no extra wiring.
Includes BE gtest, FE unit test, and regression suites.

Signed-off-by: lizhuoyu5 <lizhuoyu5@jd.com>
@hello-stephen

Copy link
Copy Markdown
Contributor

Thank you for your contribution to Apache Doris.
Don't know what should be done next? See How to process your PR.

Please clearly describe your PR:

  1. What problem was fixed (it's best to include specific error reporting information). How it was fixed.
  2. Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
  3. What features were added. Why was this function added?
  4. Which code was refactored and why was this part of the code refactored?
  5. Which functions were optimized and what is the difference before and after the optimization?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Feature] Add uniq_theta: a Theta-sketch based approximate distinct-count aggregate function

2 participants