Fix #117: [Model] GraphPartitioning

[GiggleLiu]

Summary

Add the GraphPartitioning (Minimum Bisection) problem model. Given an undirected graph with an even number of vertices, find a balanced partition into two equal halves minimizing the number of crossing edges.

Fixes #117

[codecov]

Codecov Report

❌ Patch coverage is 99.24242% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 96.49%. Comparing base (52648df) to head (9f777fa).
⚠️ Report is 6 commits behind head on main.

Files with missing lines	Patch %	Lines
src/models/graph/graph_partitioning.rs	97.56%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #570      +/-   ##
==========================================
+ Coverage   96.37%   96.49%   +0.11%     
==========================================
  Files         200      206       +6     
  Lines       27871    28301     +430     
==========================================
+ Hits        26861    27309     +448     
+ Misses       1010      992      -18     

☔ 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.

[GiggleLiu]

Implementation Summary

Changes

• src/models/graph/graph_partitioning.rs — New model: GraphPartitioning (Minimum Bisection) with single type param G, balanced bisection feasibility check, minimize cut edge count
• src/unit_tests/models/graph/graph_partitioning.rs — 11 unit tests covering evaluation, solver, odd/even/unbalanced cases, serialization
• src/models/graph/mod.rs, src/models/mod.rs, src/lib.rs — Module registration and re-export
• problemreductions-cli/src/dispatch.rs — CLI load/serialize dispatch
• problemreductions-cli/src/problem_name.rs — Alias resolution
• problemreductions-cli/src/commands/create.rs — CLI creation + random generation (with even-vertex warning)
• docs/paper/reductions.typ — Problem definition with formal definition, background, and CeTZ visualization

Deviations from Plan

• None

Open Questions

• None

[Copilot]

Pull request overview

Adds a new NP-hard graph model, GraphPartitioning (Minimum Bisection), to the problemreductions model catalog, with supporting CLI wiring, tests, and paper documentation.

Changes:

• Introduce GraphPartitioning<SimpleGraph> model with schema registration, variants, and evaluation logic.
• Add CLI support (alias resolution, JSON dispatch serialization/deserialization, and pred create support including random instance generation).
• Add unit tests and paper docs entry for Graph Partitioning.

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
src/models/graph/graph_partitioning.rs	New GraphPartitioning model implementation, schema registration, variants, and tests module hook.
src/models/graph/mod.rs	Registers/exposes the new graph model in the graph models module.
src/models/mod.rs	Re-exports GraphPartitioning from the top-level models module.
src/lib.rs	Exposes GraphPartitioning via the crate prelude.
src/unit_tests/models/graph/graph_partitioning.rs	Adds unit tests covering evaluation, direction, serialization, and brute-force solving.
problemreductions-cli/src/problem_name.rs	Adds CLI alias mapping for graphpartitioning.
problemreductions-cli/src/dispatch.rs	Adds load/serialize dispatch cases for GraphPartitioning<SimpleGraph>.
problemreductions-cli/src/commands/create.rs	Adds pred create GraphPartitioning support (explicit graph + random generation).
docs/paper/reductions.typ	Adds paper definition + figure entry for Graph Partitioning.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

count_ones is computed as config.iter().sum(), which treats any value >1 as multiple vertices in partition 1. This allows invalid assignments (e.g., a 2 value) to pass the balance check and produce incorrect cut sizes. Consider validating config is binary (all values <2) and/or counting ones via filter(|&&x| x == 1).count() after a domain check.

Suggested change

	// Check balanced: exactly n/2 vertices in partition 1
	let count_ones: usize = config.iter().sum();
	// Optional: ensure configuration length matches number of vertices
	if config.len() != n {
	return SolutionSize::Invalid;
	}
	// Check balanced: exactly n/2 vertices in partition 1, and ensure binary labels
	let mut count_ones: usize = 0;
	for &x in config.iter() {
	if x > 1 {
	// Invalid label (must be 0 or 1)
	return SolutionSize::Invalid;
	}
	if x == 1 {
	count_ones += 1;
	}
	}

[Copilot]

This repo appears to treat docs/src/reductions/reduction_graph.json as an auto-generated artifact (see examples/export_graph.rs) and it currently does not include GraphPartitioning. After adding a new model, please regenerate and commit the updated JSON so docs/tooling that consume it stay in sync.

[Copilot]

evaluate indexes config[u]/config[v] while iterating edges, but it never checks config.len() == num_vertices(). For shorter configs this will panic at runtime instead of returning SolutionSize::Invalid. Add a length check (and consider using get() like other graph models) before indexing.