Fix leak of stale duals in compute_dual_solution_from_basis

[mlubin]

The resize/scatter steps assume that y is initially empty or size zero. (If y is already size m, then resize does not overwrite y with zeros.) That assumption doesn't hold:

cuopt/cpp/src/dual_simplex/crossover.cpp

Lines 1397 to 1403 in d10c658

	if (superbasic_list.size() > 0) {
	std::vector<f_t> save_x = solution.x;
	i_t primal_push_status = primal_push(
	lp, settings, start_time, solution, ft, basic_list, nonbasic_list, superbasic_list, vstatus);
	if (primal_push_status < 0) { return return_to_status(primal_push_status); }
	compute_dual_solution_from_basis(lp, ft, basic_list, nonbasic_list, solution.y, solution.z);
	print_crossover_info(lp, settings, vstatus, solution, "Primal push complete");

At this point in the code, solution.y has already been populated in dual_push.

In practice this appears in the following logging:

Crossover start. Primal residual (abs): 1.35e-06 Primal infeasibility (abs): 0.00e+00
Crossover start. Dual   residual (abs): 1.81e-06 Dual   infeasibility (abs): 0.00e+00
Dual push complete. Primal residual (abs): 1.35e-06 Primal infeasibility (abs): 0.00e+00
Dual push complete. Dual   residual (abs): 2.53e-08 Dual   infeasibility (abs): 0.00e+00
Primal push complete. Primal residual (abs): 1.56e-11 Primal infeasibility (abs): 2.04e-05
Primal push complete. Dual   residual (abs): 1.46e+06 Dual   infeasibility (abs): 6.34e+08

Note the huge dual residual after the primal push.

With the patch instead we get:

Crossover start. Primal residual (abs): 1.35e-06 Primal infeasibility (abs): 0.00e+00
Crossover start. Dual   residual (abs): 1.81e-06 Dual   infeasibility (abs): 0.00e+00
Dual push complete. Primal residual (abs): 1.35e-06 Primal infeasibility (abs): 0.00e+00
Dual push complete. Dual   residual (abs): 2.53e-08 Dual   infeasibility (abs): 0.00e+00
Primal push complete. Primal residual (abs): 1.56e-11 Primal infeasibility (abs): 2.04e-05
Primal push complete. Dual   residual (abs): 2.91e-09 Dual   infeasibility (abs): 3.99e+08
``

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[mlubin]

/ok to test 08432ce

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

• Configuration used: Repository: NVIDIA/cuopt/.coderabbit.yaml
• Review profile: CHILL
• Plan: Enterprise
• Run ID: 100e8b6e-3fa7-4056-ba0c-49d5253f8aba

📥 Commits

Reviewing files that changed from the base of the PR and between d10c658 and 08432ce.

📒 Files selected for processing: 1

• cpp/src/dual_simplex/crossover.cpp

Included review availability: This review used your included allowance. Your plan provides up to 12 included reviews per hour; 11 remain after this review.

📝 Walkthrough

The crossover function no longer resizes y before solving. It uses y_sparse.to_dense(y) to write the computed dual solution.

Changes: Dual solution conversion

Layer / File(s)	Summary
Write the dense dual solution cpp/src/dual_simplex/crossover.cpp	The function removes the initial resize of y and uses y_sparse.to_dense(y) instead of y_sparse.scatter(y).

Priority: ➖ Normal

Estimated code review effort: 2 (Simple) | ~5 minutes

Change: Bug fix

Suggested reviewers: chris-maes

Merge Risk: ⚪ Minimal · up to 08432

The conversion now replaces stale destination values with the computed dense dual vector. No actionable merge-blocking risk remains; the PR is ready for normal checks.

Pre-merge checks: 4 · 1

Failed checks: 1 warning

Check name	Status	Explanation	Resolution
Docstring Coverage		Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 1 functions across 1 files.	Write docstrings for the functions missing them to satisfy the coverage threshold.

Passed checks: 4 passed

Check name	Status	Explanation
Linked Issues check		Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check		Check skipped because no linked issues were found for this pull request.
Title check		The title clearly and concisely identifies the main change: fixing stale dual values in compute_dual_solution_from_basis.
Description check		The description directly explains the stale dual issue, its cause, and the observed residual changes from the patch.

• Fix all pre-merge checks with AI

✨ Finishing Touches: 💡 1

🛠️ Fix failing CI checks: 💡

• Commit to this branch
• Create a new PR

🧪 Generate unit tests (beta)

• Create a new PR

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

_{Comment @coderabbitai help to get the list of available commands.}

[chris-maes]

The change looks good to me.

I'm not clear on how often this occurs.

[github-actions]

CI Test Summary

✅ All 32 test job(s) passed.

[chris-maes]

/merge