Skip to content

Add bounded fair-local assignment strategy - #4

Merged
appleweiping merged 1 commit into
mainfrom
feat/fair-local-assignment-v014
Sep 20, 2026
Merged

appleweiping merged 1 commit into
mainfrom
feat/fair-local-assignment-v014

Conversation

@appleweiping

Copy link
Copy Markdown
Owner

Adds an opt-in fair-local strategy that starts from a certified maximum-cardinality flow assignment and makes deterministic, bounded leximin improvements through admissible replacements, transfers, and two-document swaps. It preserves all existing hard constraints and explicitly rejects incompatible load penalties and oversized instances. This is a local heuristic, not OpenReview FairFlow or a global maximin guarantee; docs and a three-cycle counterexample make that boundary explicit. Tests: Windows Python 3.13 912 passed, 1 symlink skip; Ubuntu Python 3.12 913 passed; pure branch coverage 2204/2416 = 91.23%; Ruff, strict mypy, Bandit, uv audit, sdist/wheel build, and isolated wheel CLI passed. Signed-off-by: appleweiping vipinapple986@gmail.com

Signed-off-by: appleweiping <vipinapple986@gmail.com>
@appleweiping
appleweiping merged commit 42b8f58 into main Sep 20, 2026
7 checks passed
@appleweiping
appleweiping deleted the feat/fair-local-assignment-v014 branch September 20, 2026 05:56
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.

1 participant