- Rating
- 3
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
The paper proposes RRA, a neural solver architecture for parametric constrained optimization that enforces feasibility through a geometric construction rather than relying on iterative stopping tolerances. It first maps a network proposal to a strictly feasible anchor using a Gauss–Newton interior map, then retracts along the ray toward the proposal until reaching the first boundary contact. Closed-form exit calculations are provided for affine, convex-quadratic, box, and second-order-cone constraints. Across ten benchmark families, RRA achieves 100% feasibility under the paper's strict evaluation criterion, improves optimality over FSNet, and achieves competitive accuracy with substantially lower training costs than ENFORCE.
Critical strengths
- The paper presents an interesting geometric approach to enforcing feasibility by combining a differentiable interior map with radial retraction.
- The experiments cover various benchmark problems and multiple baselines, demonstrating strong performance in terms of feasibility, optimality, and training efficiency.
Critical weaknesses
- Dependence on feasible anchors. The method requires a strictly feasible anchor for each instance, which can be computationally expensive, especially for nonconvex SOCPs, where the inference time is substantially longer than that of FSNet, as shown in Table 16.
- Limitations of the feasibility guarantee. For nonconvex SOCPs, the scanning procedure does not guarantee that the first boundary crossing is found. Although this limitation is acknowledged, a quantitative analysis of potentially missed crossings would help clarify its practical significance.
- Insufficient ablation of the interior map. The experiments show that the Gauss–Newton interior map improves performance, but they do not clearly distinguish the benefits of better anchor locations from those of differentiating through the interior map.
Questions for authors
- How often does the interior map fail and fall back to the global anchor?
- Could the authors compare full differentiation with stop-gradient while keeping the same Gauss–Newton forward map?
- How sensitive are the results to the interior margin, damping parameter, scan resolution, and evaluator feasibility thresholds?
Reply to Reviewer PzYB
We thank the reviewer for the thoughtful questions, which led to the analyses below, now in the revised paper (changes in blue).
W1. Cost of the anchor. On eight of the ten families RRA takes 0.2 to 3.0 times the inference time of FSNet and is faster on five of them (Section 4.4). On the two nonconvex SOC families the anchor search is costly, and it buys much lower gaps than FSNet and feasible outputs that ENFORCE does not return.
W2. Missed crossings on nonconvex SOC. The quantitative analysis the reviewer asked for finds no missed crossing in the reported runs, and the solver now checks the first crossing itself.
W3, Q2. Anchor location against the derivative of the map. With the same Gauss–Newton forward map, replacing only its derivative raises the gap on every family, so training through the map matters beyond where the anchor lies. The comparison the reviewer proposed is now Table 3 (Section 4.3).
On the nine families where RRA's gap is positive:
Q1. Fallback to the global anchor. At the reported settings the returned output starts from the global anchor on at most one instance per seed of any family (Appendix H.1).
Q3. Sensitivity. No setting we varied made a returned output fail the strict rule. The gap responds to the margin and hardly to the damping or the scan.
Gap (%) for three values of the margin μ:
At μ=10−8 the output of the map misses the acceptance test on 102 of 6,000 nonconvex QP instances, which then start from the global anchor.
The damping ρ of Eq. (1) changes the gap by at most 2.5% of its value from 10−12 to 10−4.
Scans with 8 to 128 cells and 5 to 50 bisection steps change the gap on the four trigonometric families by at most 0.00037 points.
On nonconvex SOC, the margin and the ridge move the gap by at most 0.45 points.
Every output passes at every common evaluator tolerance down to 10−10 (Appendix I.2). At 10−12, 86.8% and 93.4% of the nonconvex SOC outputs pass, since their sliding stages accept rows up to 10−10.
We hope these analyses help, and we would be glad to discuss any point further.