Fix #215: [Model] EnsembleComputation

[GiggleLiu]

Summary

• Add an implementation plan for EnsembleComputation

Fixes #215

[GiggleLiu]

Implementation Summary

Changes

• Added the EnsembleComputation satisfaction model with validated construction, registry/variant metadata, prelude exports, and canonical example wiring.
• Added pred create EnsembleComputation support, CLI help text, and CLI coverage for creation, validation, and --example.
• Added model-focused tests plus a paper problem definition with background, worked example, and citation.

Deviations from Plan

• Kept the paper example on the issue's 4-element instance, but used a smaller canonical --example instance so the example-db optimality check can solve it without an ILP reduction.

Open Questions

• None.

[GiggleLiu]

Agentic Review Report

Note: This PR currently has merge conflicts with main (mergeStateStatus=DIRTY, mergeable=CONFLICTING). These must be resolved before merging.

Structural Check

Structural Review: model EnsembleComputation

Structural Completeness

#	Check	Status
1	Model file exists	PASS
2	inventory::submit! present	PASS
3	#[derive(...Serialize, Deserialize)] on struct	PASS
4	Problem trait impl	PASS
5	SatisfactionProblem impl	PASS
6	#[cfg(test)] + #[path = ...] test link	PASS
7	Test file exists	PASS
8	Test file has >= 3 test functions	PASS
9	Registered in misc/mod.rs	PASS
10	Re-exported in models/mod.rs	PASS
11	declare_variants! entry exists	PASS
12	CLI problem-name resolution	PASS — registry-backed resolve_alias() requires no per-model hardcoded entry
13	CLI create support	PASS
14	Canonical model example registered	PASS
15	Paper display-name entry	PASS
16	Paper problem-def block	PASS
17	Blacklisted auto-generated files absent	PASS

Build Status

• make test: PASS
• make clippy: PASS
• make paper: PASS

Semantic Review

• evaluate() correctness: OK — enforces config length, rejects future references via decode_operand, checks disjointness before union, and preserves the issue's j <= J semantics by accepting as soon as a valid prefix has produced every required subset.
• dims() correctness: OK — vec![universe_size + budget; 2 * budget] matches the issue's flat operand encoding, with evaluate() filtering out future references.
• Size getter consistency: OK — universe_size(), num_subsets(), and budget() line up with the schema and complexity metadata.
• Weight handling: OK — N/A for this unweighted satisfaction problem.

Issue Compliance

#	Check	Status
1	Problem name matches issue	OK
2	Mathematical definition matches	OK
3	Problem type (sat) matches	OK
4	Type parameters match	OK — no generic parameters
5	Configuration space matches	OK
6	Feasibility check matches	OK
7	Objective function matches	OK
8	Complexity matches	OK

Summary

• 17/17 structural checks passed
• 8/8 issue compliance checks passed
• Deterministic context note: the pre-generated implementation packet flagged files outside the current model whitelist (problemreductions-cli/..., src/lib.rs, src/models/mod.rs, src/models/misc/mod.rs). Those are real whitelist deviations and should be sanity-checked in final review, although they match current extension points for adding a user-facing model.

---

Quality Check

Quality Review

Design Principles

• DRY: OK — operand decoding, disjointness checking, and union construction are factored into small helpers instead of duplicated inline logic.
• KISS: OK — the model uses the simplest workable flat encoding from the issue and avoids extra abstraction.
• HC/LC: OK — problem logic, CLI construction, tests, and paper wiring stay in the expected modules.

HCI (if CLI/MCP changed)

• Error messages: OK — missing flags and invalid budgets are actionable, and the failing solve path at least suggests --solver brute-force.
• Discoverability: ISSUE — pred create --example EnsembleComputation succeeds, but the next natural step pred solve <instance> fails under defaults because the model has no ILP/reduction path yet.
• Consistency: OK — create/help wiring follows existing CLI patterns.
• Least surprise: ISSUE — exposing a canonical example through pred create --example suggests a direct solve flow, but the default solver path dead-ends for this model.
• Feedback: OK — the failing solve path gives a usable fallback hint.

Test Quality

• Naive test detection: ISSUE
  • problemreductions-cli/tests/cli_tests.rs:7415 only checks pred create --example EnsembleComputation; there is no CLI coverage for solving that example, so the broken default pred solve path was not exercised.

Issues

Critical (Must Fix)

• None.

Important (Should Fix)

• The canonical EnsembleComputation example is not solvable via the default CLI path. Reproduction from the reviewed branch: pred create --example EnsembleComputation > /tmp/ensemble.json succeeds, but pred solve /tmp/ensemble.json fails with No reduction path from EnsembleComputation to ILP or ILP solver found no solution. Try --solver brute-force. Rerunning with --solver brute-force succeeds. This is a real user-path mismatch between --example and the default solve flow.
• Add CLI regression coverage for the solve path in addition to the existing create/example coverage at problemreductions-cli/tests/cli_tests.rs:7415.

Minor (Nice to Have)

• None.

Summary

• The implementation itself looks coherent and the build/tests are clean, but the user-facing CLI path is incomplete: the new canonical example requires an explicit --solver brute-force flag that users only discover after a failed default solve.

---

Agentic Feature Tests

Feature: EnsembleComputation

Use case: A downstream CLI user discovers the new model from the catalog, inspects it, creates the canonical example, and tries to solve it.

• Discoverability: Partial — pred list shows EnsembleComputation, pred show EnsembleComputation renders the schema/complexity correctly, and pred create --example EnsembleComputation emits a valid example instance.
• Setup: PASS — the existing CLI binary built and ran successfully in the review environment.
• Functionality: Partial — the default solve path failed; the brute-force fallback worked.
• Expected vs Actual: Partial — expected the canonical example to be directly solvable from the default CLI flow, but in practice the user must add --solver brute-force after the default solve fails.
• Blocked steps: None.
• Friction points: The PR adds a canonical example and create coverage, but not a successful default solve path or test coverage for the required brute-force invocation.
• Doc suggestions: Either make the expected solver choice explicit in user-facing docs/help for this model, or improve the default solve behavior for standalone problems with no reduction path.

Confirmed issues

• Important — Reproduced in the reviewed branch:
  1. target/debug/pred create --example EnsembleComputation > /tmp/ensemble.json
  2. target/debug/pred solve /tmp/ensemble.json → fails with the no-ILP/reduction-path error and a --solver brute-force hint
  3. target/debug/pred solve /tmp/ensemble.json --solver brute-force → succeeds and returns a satisfying assignment

---

Generated by review-pipeline

[codecov]

Codecov Report

❌ Patch coverage is 97.57282% with 5 lines in your changes missing coverage. Please review.
✅ Project coverage is 97.62%. Comparing base (77577dd) to head (027323b).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/models/misc/ensemble_computation.rs	96.45%	5 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff            @@
##             main     #721    +/-   ##
========================================
  Coverage   97.61%   97.62%            
========================================
  Files         401      403     +2     
  Lines       49906    50112   +206     
========================================
+ Hits        48717    48923   +206     
  Misses       1189     1189            

☔ 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.