feat: Add Chebyshev distance (L_∞ norm) support by cbueth · Pull Request #290 · sdd/kiddo

cbueth · 2026-02-12T23:44:07Z

This is a PR #286, reopened for the fitting branch. I rebased my suggested changes, dropping all other commits. Locally all test still pass. I have added a few cases to increase code coverage. Find the PR description over at #286. I will just add onto. it here.

For the default implementation of accumulate I understood best to use num_traits::SaturatingAdd, but this trait is only implemented for integer types, not for floating-point types (f32, f64). So now I changed it to use std::ops::Add as the bound, which works for both integers and floats. This means the default rd + delta works correctly for floating-point types (produces inf on overflow), but fixed-point types need to override accumulate with rd.saturating_add(delta) to prevent overflow. I ignorantly assume that this covers a larger userbase and think this solution is the best we can do without specialization (is currently unstable https://std-dev-guide.rust-lang.org/policy/specialization.html / rust-lang/rust#31844). To my understanding this is still a breaking change for users with fixed-point types who implement custom DistanceMetrics. The only feasible alternative I see would be with a macro to generate implementations that can differentiate the two. Please correct me if I got this wrong.

Secondly, I want to clarify a statement from my previous #286 (comment).

Finally, I was not able to find a test where < vs <= makes a difference, but programmatically I agree it could strictly be the right choice. As of now I changed some back to <. You can see the two places it was added (D::IS_MAX_BASED removed in this commit c934656. NON_STRICT_PRUNING would be a better name). I'd suggest you treat this as a separate issue.

I was only not able to find a relevant test where the <= removed in c934656 did (dist_belongs_in_heap) make a difference, so I changed them to <. But still, there are other occurences that are still included and that did make a difference in the test with gaussian data. So if you think there are some cases where having the <= for dist_belongs_in_heap, these can be re-added. Right now there are many parametrizes test cases with nearest_n, but not for within.

If there are suggestions for changes, don't heasitate to say so.

Closes #286

sdd · 2026-02-16T08:02:47Z

Hi again Carlson! Just seen this, I will check them out over the next day or two. Thanks!

codecov · 2026-02-16T08:03:56Z

Codecov Report

❌ Patch coverage is 95.52980% with 27 lines in your changes missing coverage. Please review.
✅ Project coverage is 95.04%. Comparing base (f87b965) to head (c4e8db9).
⚠️ Report is 17 commits behind head on v5.x.x.

Files with missing lines	Patch %	Lines
src/traits.rs	29.41%	12 Missing ⚠️
src/float/distance.rs	98.21%	3 Missing and 5 partials ⚠️
src/fixed/distance.rs	93.63%	5 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           v5.x.x     #290      +/-   ##
==========================================
+ Coverage   94.89%   95.04%   +0.14%     
==========================================
  Files          54       54              
  Lines        5685     6252     +567     
  Branches     5685     6252     +567     
==========================================
+ Hits         5395     5942     +547     
- Misses        273      286      +13     
- Partials       17       24       +7

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

cbueth · 2026-02-16T08:17:57Z

Sounds good, thanks :)
Let me know of last changes you want me to integrate, e.g. coverage, formatting.

sdd · 2026-02-16T20:12:19Z

Thanks. As it is, the commit lint step fails because some of your commit message subjects are not confirming to Conventional Commits standard. You can either fix them individually with interactive rebase, or squash them into one commit with a catch-all subject like feat: Add Chebyshev distance metric

cbueth · 2026-02-26T10:56:52Z

Thanks. As it is, the commit lint step fails because some of your commit message subjects are not confirming to Conventional Commits standard. You can either fix them individually with interactive rebase, or squash them into one commit with a catch-all subject like feat: Add Chebyshev distance metric

Yes, I will do that asap for all three PRs.

cbueth · 2026-02-26T14:44:55Z

Hello @sdd, I have changed the commit messages to conform with the conventional commits standard, rebasing this PR, #293 and #294. If there is any more feedback, let me know. Otherwise, I'd be happy to these features in a coming release, so I can use this as a dependency going beta with infomeasure-rs. Again, thanks for your constructive feedback and welcoming my feature requests.

cbueth · 2026-02-27T08:32:46Z

Thanks for running the CI. commitlint failes because I capitalized the message. I think this should only be done in PR titles, not commits.
In stable without features there seems to be one test fixed::query::best_n_within::tests::can_query_best_items_within_radius_large_scale with a panick due to overflow. I will try to reproduce locally.

sdd

Thanks for getting this into shape Carlson, it's a very welcome addition and very thoroughly tested 😄

I just need to work out why the CI didn't seem to run on this. Once we've worked out why and got it to run on this again, I'm happy to merge and we'll get it released.

sdd · 2026-02-27T08:22:44Z

src/fixed/distance.rs

+    #[case([ZERO, ZERO], [ZERO, ZERO], ZERO)]
+    #[case([ZERO, ZERO], [ONE, ZERO], ONE)]
+    #[case([ZERO, ZERO], [ZERO, ONE], ONE)]
+    #[case([ZERO, ZERO], [ONE, ONE], FxdU16::lit("2"))]


We could use TWO here instead of FxdU16::lit("2") but I'm not going to block merging for such a trivial change :-)

Good catch, maybe I missed one click in find+replace there.

Done in f33b6e9.

…lidean distance metrics

- deprecate `rd_update` with `D::accumulate` for consistent sum-based and max-based metrics - conditional logic for SIMD (L1/L2) and general L∞ - differentiate distance accumulation behaviour

- integration `nearest_n` tests (Chebyshev, Manhattan, SquaredEuclidean).

- improve `DistanceMetric` doc - add Gaussian scenario to tests

… trait - improve test coverage

- `A::dist` - `saturating_mul`

cbueth · 2026-02-27T09:05:02Z

While I was not able to reproduce the overflow, it happend with FixedU16 and points to the saturating arithmetic. The 4D test probably exceeded the range of the type. So I replaced the dist and dist1 for all three fixed metrics. Because I did not create a test case that triggers this locally for me (Maybe increasing TREE_SIZE or NUM_QUERIES in the test at src/fixed/query/best_n_within.rs:148 could trigger this overflow.), I'd suggest we let the CI use the existing tests.

Manual absolute difference if a > b { a - b } else { b - a }) -> the dist() of fixed::traits::Fixed.
For SquaredEuclidean now using saturating_mul()

For both I did not know before these exist for Fixed.

I just need to work out why the CI didn't seem to run on this. Once we've worked out why and got it to run on this again, I'm happy to merge and we'll get it released.

That's great to hear, thanks!

sdd · 2026-03-02T07:39:29Z

Hi Carlson - I've merged this in now. Looking at your other PR now (#291), there's a small merge conflict now that this one has been merged, as well as a failure on the formatting check. If you can get that one back into shape, I'll merge that one in as well so we can get both into a 5.3.0 release. Thanks!

cbueth · 2026-03-02T13:06:41Z

Thanks for the review and merge! 👏

cbueth mentioned this pull request Feb 26, 2026

Optional Boundary Inclusiveness for Within Queries #293

Closed

cbueth force-pushed the feature/chebychev branch from 7adffe8 to 80c8d89 Compare February 26, 2026 12:50

cbueth mentioned this pull request Feb 26, 2026

feat: Optional Boundary Inclusiveness for Within Queries #294

Open

test: add coverage for Manhattan and Squared Euclidean distance metrics

55143d2

sdd approved these changes Feb 27, 2026

View reviewed changes

cbueth added 15 commits February 27, 2026 09:35

feat: add Chebyshev distance metric and test coverage

a7471bf

test: add integration tests for Chebyshev, Manhattan, and Squared Euc…

a5cf629

…lidean distance metrics

refactor: inclusive radius matching and leaf note remainder loops

898da96

fix: fix over-pruning for L_inf

728f3ee

feat: add D::accumulate and D::IS_MAX_BASED

78cf373

- deprecate `rd_update` with `D::accumulate` for consistent sum-based and max-based metrics - conditional logic for SIMD (L1/L2) and general L∞ - differentiate distance accumulation behaviour

feat: add fixed Chebyshev distance metric

ec7fcb3

- integration `nearest_n` tests (Chebyshev, Manhattan, SquaredEuclidean).

refactor: in-loop accumulation for max-based metrics

ccecab0

refactor: unify distance accumulation logic with D::accumulate

f9feccd

refactor: remove D::IS_MAX_BASED, unify heap logic, doc & test

53759e2

- improve `DistanceMetric` doc - add Gaussian scenario to tests

chore: change test comment & lint

8239e39

refactor: make metric property tests reusable

142a74e

chore: add default implementation of accumulate to DistanceMetric…

faa6dab

… trait - improve test coverage

chore: saturating add for fixed metrics

2b4dc67

chore: remove outdated docstring test information

0de47af

chore: replace FxdU16::lit("2") with TWO

f33b6e9

cbueth force-pushed the feature/chebychev branch from 7479563 to f33b6e9 Compare February 27, 2026 08:42

fix: use saturating arithmetic for fixed-point distances

c4e8db9

- `A::dist` - `saturating_mul`

cbueth force-pushed the feature/chebychev branch from 6605b39 to c4e8db9 Compare February 27, 2026 08:56

sdd merged commit 568b496 into sdd:v5.x.x Mar 1, 2026
10 checks passed

This was referenced Mar 2, 2026

feat(WIP): Add Chebyshev distance ($L_∞$ norm) support #286

Closed

feat: Generalised Minkowski Metric (L_p norm) #287

Closed

cbueth deleted the feature/chebychev branch March 2, 2026 13:06

Conversation

cbueth commented Feb 12, 2026

Uh oh!

sdd commented Feb 16, 2026

Uh oh!

codecov bot commented Feb 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

cbueth commented Feb 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sdd commented Feb 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cbueth commented Feb 26, 2026

Uh oh!

cbueth commented Feb 26, 2026

Uh oh!

cbueth commented Feb 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sdd left a comment

Choose a reason for hiding this comment

Uh oh!

sdd Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

cbueth Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

cbueth Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

cbueth commented Feb 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

sdd commented Mar 2, 2026

Uh oh!

cbueth commented Mar 2, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

codecov bot commented Feb 16, 2026 •

edited

Loading

cbueth commented Feb 16, 2026 •

edited

Loading

sdd commented Feb 16, 2026 •

edited

Loading

cbueth commented Feb 27, 2026 •

edited

Loading

cbueth commented Feb 27, 2026 •

edited

Loading