arxiv: 2604.09031 · v1 · submitted 2026-04-10 · 🧮 math.OC

Recognition: unknown

Adaptive Subproblem Selection in Benders Decomposition for Survivable Network Design Problems

Tim Donkiewicz

Authors on Pith no claims yet

Pith reviewed 2026-05-10 17:21 UTC · model grok-4.3

classification 🧮 math.OC

keywords Benders decompositionsubproblem selectionsurvivable network designlogistic regressionscenario-based optimizationcut generationstochastic programming

0 comments

The pith

Adaptive scoring lets Benders decomposition skip most subproblems while still reaching good solutions faster.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that in scenario-based problems solved by Benders decomposition, the majority of subproblems contribute no new cuts at a given iteration and can safely be omitted. It develops scoring rules that combine past subproblem behavior with current instance features, then fits logistic regression models incrementally to rank subproblems by their chance of being useful. Selection is controlled by simple limits on the number of cuts, the fraction of subproblems solved, or score cutoffs, so that the master problem still receives enough information to converge. On survivable network design instances the approach yields faster runtimes and tighter primal-dual integrals than solving every subproblem every round, while random selection actually performs worse than the baseline. The work therefore demonstrates that informed, adaptive omission of subproblems is a practical way to scale Benders decomposition when the number of scenarios grows large.

Core claim

By training logistic regression models online on historical performance metrics and problem-specific features, then applying multi-criteria scores together with cut, proportion, or threshold stopping rules, Benders decomposition can solve only a subset of the recourse subproblems each iteration and still produce statistically significant gains in runtime and primal-dual integrals on survivable network design problems.

What carries the argument

The online logistic regression scoring model that predicts the probability a given subproblem will generate a useful Benders cut, combined with explicit stopping criteria that trade exploration against exploitation.

If this is right

An oracle that knows in advance which subproblems will produce cuts reduces total solve time by 34.4 percent.
Random selection of subproblems increases overall computation time compared with solving every subproblem.
The best adaptive scoring rule produces statistically significant reductions in runtime and improvements in primal-dual integrals.
More than half of the subproblems solved under full enumeration contribute no cuts and occur outside cut-free rounds.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same online scoring idea could be applied to other iterative decomposition schemes in which subproblem value varies across iterations.
Adding structural features derived from the underlying network might further improve the regression predictions without requiring more data.
The explicit stopping criteria offer a tunable knob that could be combined with existing Benders acceleration techniques such as cut aggregation.

Load-bearing premise

That logistic regression models trained on the fly from past iterations can keep predicting which subproblems will yield cuts without overlooking ones that are essential for solution quality.

What would settle it

If the adaptive method is run on a fresh collection of instances and produces longer total runtimes or worse primal-dual gaps than full enumeration on the same instances, the claim of reliable improvement would be refuted.

Figures

Figures reproduced from arXiv: 2604.09031 by Tim Donkiewicz.

**Figure 1.** Figure 1: Basic Branch-and-Benders-cut framework. and some scenarios can dominate others due to network topology. While our goal is not to advance the state-of-the-art for SND solving specifically, this problem class provides a suitable environment to demonstrate subproblem selection techniques. To our knowledge, adaptive subproblem selection based on computational pattern learning has not been successfully demonstr… view at source ↗

**Figure 2.** Figure 2: Adaptive subproblem selection pipeline in Benders decomposition. Scoring and subproblem selection [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Figure 3: Time breakdown across n = 108 instances. Master time (black), cut-generating subproblems (blue), and feasibility-proving subproblems in cut-free iterations (grey) represent necessary computation; unrequired subproblems (orange) represent wasted computation. total solve time in standard Benders is spent on unrequired solving of subproblems. Since solving all subproblems that yield cuts does not always lead … view at source ↗

**Figure 4.** Figure 4: Performance profile on all n = 91 instances solved by at least one method. The performance factor τ on the x-axis is the ratio of a method’s runtime to the best method’s runtime on each instance; ρ(τ ) on the y-axis is the fraction of instances where the method is within factor τ of the best. Random selection performs worse than standard Benders, regression-based methods achieve modest gains. Runtime Perfo… view at source ↗

**Figure 5.** Figure 5: Distribution of minimum parameter values required to capture all cuts per iteration during ML training, [PITH_FULL_IMAGE:figures/full_fig_p014_5.png] view at source ↗

**Figure 6.** Figure 6: Cut yield evolution across iterations for full enumeration, showing the fraction of examined subproblems [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗

**Figure 7.** Figure 7: Distribution of learned feature weights across instances. Feature reference in Table 1. Box plots show [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗

**Figure 8.** Figure 8: Distribution of hit quality (cuts found / subproblems examined) across methods. Higher values indicate [PITH_FULL_IMAGE:figures/full_fig_p016_8.png] view at source ↗

**Figure 9.** Figure 9: Performance profile based on adjusted subproblem time, which deducts the necessary subproblem time [PITH_FULL_IMAGE:figures/full_fig_p016_9.png] view at source ↗

read the original abstract

Scenario-based optimization problems can be solved via Benders decomposition, which separates first-stage (master problem) decisions from second-stage (subproblem) recourse actions and iteratively refines the master problem with Benders cuts. In conventional Benders decomposition, all subproblems are solved at each iteration. For problems with many scenarios, solving only a selected subset can reduce computation. We quantify the potential in selecting only those subproblems that yield cuts, and develop subproblem scoring and selection strategies. The proposed multi-criteria scoring methods combine historical subproblem performance metrics with problem-specific features, trained online via logistic regression to adapt to the changing likelihood of subproblem usefulness. Multiple stopping criteria balance exploration and exploitation: cut limits, proportional solve limits, and score thresholds. We evaluate our approach on a variant of the survivable network design problem, which serves as a testbed due to its natural decomposition into many subproblems of varying importance. Computational experiments on 135 test instances demonstrate the potential and practical performance of subproblem selection. Analysis reveals that 52.1% of all subproblems solved are unnecessary (they contribute no cuts and occur outside cut-free rounds). An oracle with perfect foresight reduces total solve times by 34.4%. Random selection performs significantly worse than full enumeration, showing that naive strategies can degrade performance. Our best-scoring and selection method achieves statistically significant improvements in both runtime and primal-dual integrals. These results provide empirical evidence that informed subproblem selection can improve Benders decomposition in this setting, while highlighting challenges in developing reliable prediction models.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This paper shows that online logistic regression on subproblem history and features can cut unnecessary solves in Benders decomposition for survivable network design, delivering statistically significant runtime and integral gains over full enumeration on their 135 instances.

read the letter

The core result is that informed selection beats solving every subproblem every iteration. They score subproblems with a mix of historical cut yield, instance features, and online logistic regression, then apply cut-count, proportional, or score-based stopping rules. On the survivable network design testbed this removes 52% of the subproblems that would have been solved anyway, an oracle shows 34% time savings are possible, and their best variant improves both wall-clock time and primal-dual integrals with statistical significance. Random selection does worse than full enumeration, which is a useful negative control. The experiments are run on a decent number of instances and the comparisons are straightforward. That is the useful part: a concrete, reproducible way to prune work inside an existing decomposition loop without changing the cuts or the master problem. The soft spots are modest but real. The gains rest on the regression not systematically missing high-value cuts on unseen instances, and the paper does not yet show how sensitive the online training is to the order of scenarios or to larger instance sizes. The testbed is narrow, so it is not clear how much of the improvement travels to other multi-scenario problems. No theoretical convergence guarantee is claimed or needed for the contribution, but the empirical claims would benefit from more detail on variance across random seeds and on whether solution quality ever degrades. This is the kind of paper that belongs in a computational optimization journal or conference. Readers working on Benders or column generation for stochastic programs will find the selection heuristics and the oracle baseline directly usable. It is not a foundational advance, but the evidence is strong enough that a serious referee should see it. I would send it out for review.

Referee Report

2 major / 1 minor

Summary. The paper proposes adaptive subproblem selection strategies within Benders decomposition for a variant of the survivable network design problem. It develops multi-criteria scoring methods that combine historical performance metrics with instance features, trained online via logistic regression, and employs stopping rules based on cut limits, proportional solves, and score thresholds. On 135 test instances, the work reports that 52.1% of solved subproblems are unnecessary, an oracle with perfect foresight yields a 34.4% runtime reduction, random selection degrades performance relative to full enumeration, and the best proposed method delivers statistically significant gains in both runtime and primal-dual integrals.

Significance. If the empirical claims hold under closer scrutiny of controls and variance, the results supply concrete evidence that informed, adaptive selection of subproblems can accelerate Benders decomposition in scenario-rich problems without compromising solution quality. The quantification of unnecessary subproblems and the oracle benchmark are particularly useful for establishing the potential upside, while the online logistic regression approach illustrates a practical way to balance exploration and exploitation in decomposition algorithms.

major comments (2)

[Computational experiments] Computational experiments section: the claim of statistically significant improvements in runtime and primal-dual integrals is load-bearing for the central empirical contribution, yet the manuscript provides no details on the statistical test(s) employed, exact p-values, effect sizes, or how variance and potential selection biases across the 135 instances were controlled, leaving the robustness of the significance assertions unverifiable.
[Methodology for scoring and selection] Methodology for scoring and selection: the weakest assumption—that online logistic regression on historical metrics and features can reliably predict subproblem usefulness without missing important cuts—is not supported by any reported analysis of false-negative rates, cut-quality degradation, or cases where selected subsets produced inferior master-problem bounds relative to full enumeration.

minor comments (1)

[Abstract] The abstract would benefit from explicitly naming the survivable network design variant and the precise set of instance features used in the logistic regression.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. The comments identify important gaps in the presentation of our empirical results and validation of the selection methodology. We address each point below and will revise the manuscript accordingly to strengthen the claims.

read point-by-point responses

Referee: [Computational experiments] Computational experiments section: the claim of statistically significant improvements in runtime and primal-dual integrals is load-bearing for the central empirical contribution, yet the manuscript provides no details on the statistical test(s) employed, exact p-values, effect sizes, or how variance and potential selection biases across the 135 instances were controlled, leaving the robustness of the significance assertions unverifiable.

Authors: We agree that the statistical details are missing and must be added for verifiability. In the revised manuscript we will expand the computational experiments section with: the exact test used (Wilcoxon signed-rank test on paired differences for runtime and primal-dual integral across the 135 instances), all reported p-values, effect sizes (rank-biserial correlation), and controls for variance (median and IQR reporting plus instance-level pairing to mitigate selection bias). These additions will be placed in a dedicated subsection on statistical analysis. revision: yes
Referee: [Methodology for scoring and selection] Methodology for scoring and selection: the weakest assumption—that online logistic regression on historical metrics and features can reliably predict subproblem usefulness without missing important cuts—is not supported by any reported analysis of false-negative rates, cut-quality degradation, or cases where selected subsets produced inferior master-problem bounds relative to full enumeration.

Authors: We acknowledge that the current manuscript does not contain an explicit analysis of false-negative rates or bound degradation. While the reported improvements in both runtime and primal-dual integrals relative to full enumeration provide indirect evidence that solution quality was not compromised, a direct examination is warranted. We will add to the revised paper: (i) false-negative rates of the logistic regression predictor computed via rolling cross-validation on the training history, and (ii) a comparison of final master-problem bounds (and primal-dual gaps) between the adaptive selection and full-enumeration runs on all 135 instances to quantify any degradation. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper describes an algorithmic framework for adaptive subproblem selection in Benders decomposition, using online logistic regression on historical metrics and instance features to score subproblems, combined with stopping rules. All central claims rest on empirical evaluation across 135 test instances, with explicit comparisons to full enumeration, random selection, and an oracle upper bound; runtime and primal-dual integral improvements are reported as statistically significant without any reduction of the method or results to self-referential definitions, fitted parameters renamed as predictions, or load-bearing self-citations. The logistic regression serves as a standard predictive model whose outputs are validated externally against independent benchmarks rather than by construction, and no derivation chain or uniqueness theorem is invoked that collapses to the paper's own inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The claim rests on standard Benders decomposition validity and empirical performance of trained predictors; no new entities or fixed free parameters are introduced beyond online-trained logistic coefficients.

axioms (1)

standard math Benders cuts generated from subproblems are valid and tighten the master problem relaxation
Core assumption of Benders decomposition invoked throughout the method description.

pith-pipeline@v0.9.0 · 5574 in / 1195 out tokens · 34675 ms · 2026-05-10T17:21:34.283622+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

33 extracted references · 30 canonical work pages

[1]

PhD thesis, Technische Universität Berlin, 2007

Tobias Achterberg.Constraint integer programming. PhD thesis, Technische Universität Berlin, 2007. doi:10.14279/depositonce-1634

work page doi:10.14279/depositonce-1634 2007
[2]

Mathematical Programming Computation , volume =

Tobias Achterberg. SCIP: solving constraint integer programs.Mathematical Programming Compu- tation, 1(1):1–41, 2009.doi:10.1007/s12532-008-0001-1

work page doi:10.1007/s12532-008-0001-1 2009
[3]

Embedding cuts in a branch&cut framework: a computational study with{0, 1 2 }-cuts.INFORMS Journal on Computing, 19(2):229– 238, 2007.doi:10.1287/ijoc.1050.0162

Giovanni Andreello, Alberto Caprara, and Matteo Fischetti. Embedding cuts in a branch&cut framework: a computational study with{0, 1 2 }-cuts.INFORMS Journal on Computing, 19(2):229– 238, 2007.doi:10.1287/ijoc.1050.0162

work page doi:10.1287/ijoc.1050.0162 2007
[4]

Metric Inequalities and the Network Loading Problem

Pasquale Avella, Sara Mattia, and Antonio Sassano. Metric Inequalities and the Network Loading Problem. InDiscrete Optimization: The State of the Art, pages 53–84. Elsevier, 2004. doi: 10.1016/j.disopt.2006.10.002

work page doi:10.1016/j.disopt.2006.10.002 2004
[5]

Partitioning procedures for solving mixed-variables programming problems

Jacques F Benders. Partitioning procedures for solving mixed-variables programming problems. Numerische mathematik, 4(1):238–252, 1962.doi:10.1007/bf01386316

work page doi:10.1007/bf01386316 1962
[6]

Bezanson, A

Jeff Bezanson, Alan Edelman, Stefan Karpinski, and Viral B Shah. Julia: A fresh approach to numerical computing.SIAM review, 59(1):65–98, 2017.doi:10.1137/141000671

work page doi:10.1137/141000671 2017
[7]

Xavier Blanchot, François Clautiaux, Boris Detienne, Aurélien Froger, and Manuel Ruiz. The Benders by batch algorithm: Design and stabilization of an enhanced algorithm to solve multicut Benders reformulation of two-stage stochastic programs.European Journal of Operational Research, 309(1):202–216, 2023.doi:10.1016/j.ejor.2023.01.004

work page doi:10.1016/j.ejor.2023.01.004 2023
[8]

S Borozan, S Giannelos, P Falugi, A Moreira, G Strbac, and T Zhang. Machine learning-enhanced Benders decomposition approach for the multi-stage stochastic transmission expansion planning problem.Electric Power Systems Research, 226:109956, 2024.doi:10.1016/j.epsr.2024.110985

work page doi:10.1016/j.epsr.2024.110985 2024
[9]

Benders decomposition for the hop-constrained survivable network design problem.INFORMS Journal on Computing, 25(1):13–26, 2013.doi:10.1287/ijoc.1110.0472

Quentin Botton, Bernard Fortz, Luis Gouveia, and Michael Poss. Benders decomposition for the hop-constrained survivable network design problem.INFORMS Journal on Computing, 25(1):13–26, 2013.doi:10.1287/ijoc.1110.0472

work page doi:10.1287/ijoc.1110.0472 2013
[10]

Exact two-step Benders decomposition for the time window assignment traveling salesperson problem.Transportation Science, 59(2):210–228, 2025.doi:10.1287/trsc.2024.0750

Şifa Çelik, Layla Martin, Albert H Schrotenboer, and Tom Van Woensel. Exact two-step Benders decomposition for the time window assignment traveling salesperson problem.Transportation Science, 59(2):210–228, 2025.doi:10.1287/trsc.2024.0750. 11

work page doi:10.1287/trsc.2024.0750 2025
[11]

An implicit optimization approach for survivable network design

Richard LY Chen, Amy Cohn, and Ali Pinar. An implicit optimization approach for survivable network design. In2011 IEEE Network Science Workshop, pages 159–163. IEEE, 2011. doi: 10.1016/j.ins.2011.09.004

work page doi:10.1016/j.ins.2011.09.004 2011
[12]

Partial Benders decomposition: general methodology and application to stochastic network design.Transportation Science, 55(2):414–435, 2021.doi:10.1287/trsc.2020.1022

Teodor Gabriel Crainic, Mike Hewitt, Francesca Maggioni, and Walter Rei. Partial Benders decomposition: general methodology and application to stochastic network design.Transportation Science, 55(2):414–435, 2021.doi:10.1287/trsc.2020.1022

work page doi:10.1287/trsc.2020.1022 2021
[13]

Partial decomposition strategies for two- stage stochastic integer programs.CIRRELT (Université de Montréal), 88, 2014.doi:10.1007/ 978-3-662-04331-8_30

Teodor Gabriel Crainic, Mike Hewitt, and Walter Rei. Partial decomposition strategies for two- stage stochastic integer programs.CIRRELT (Université de Montréal), 88, 2014.doi:10.1007/ 978-3-662-04331-8_30

2014
[14]

Machine learning for cutting planes in integer programming: A survey.arXiv preprint arXiv:2302.09166, 2023.doi:10.24963/ijcai.2023/739

Antoine Deza and Elias B Khalil. Machine learning for cutting planes in integer programming: A survey.arXiv preprint arXiv:2302.09166, 2023.doi:10.24963/ijcai.2023/739

work page doi:10.24963/ijcai.2023/739 2023
[15]

JuMP: A modeling language for mathematical optimization.SIAM Review, 59(2):295–320, 2017.doi:10.1137/15m1020575

Iain Dunning, Joey Huchette, and Miles Lubin. JuMP: A modeling language for mathematical optimization.SIAM Review, 59(2):295–320, 2017.doi:10.1137/15m1020575

work page doi:10.1137/15m1020575 2017
[16]

An improved Benders decomposition applied to a multi-layer network design problem.Operations Research Letters, 37(5):359–364, 2009.doi:10.1016/j.orl

Bernard Fortz and Michael Poss. An improved Benders decomposition applied to a multi-layer network design problem.Operations Research Letters, 37(5):359–364, 2009.doi:10.1016/j.orl. 2009.05.007

work page doi:10.1016/j.orl 2009
[17]

Gurobi Optimizer Reference Manual, 2024

Gurobi Optimization, LLC. Gurobi Optimizer Reference Manual, 2024. URL:https://www.gurobi. com

2024
[18]

Deepest Cuts for Benders Decomposition.Operations Research, 73(5):2591–2609, 2024.doi:10.1287/opre.2021.0503

Mojtaba Hosseini and John Turner. Deepest Cuts for Benders Decomposition.Operations Research, 73(5):2591–2609, 2024.doi:10.1287/opre.2021.0503

work page doi:10.1287/opre.2021.0503 2024
[19]

Idzikowski, S

F. Idzikowski, S. Orlowski, C. Raack, H. Woesner, and A. Wolisz. Dynamic routing at different layers in IP-over-WDM networks – Maximizing energy savings.Optical Switching and Networking, Special Issue on Green Communications, 2011.doi:10.1016/j.osn.2011.03.007

work page doi:10.1016/j.osn.2011.03.007 2011
[20]

Koster, Manuel Kutschka, and Christian Raack

Arie M.C.A. Koster, Manuel Kutschka, and Christian Raack. Towards Robust Network Design using Integer Linear Programming Techniques. InProceedings of the NGI 2010, Paris, France, Paris, France, June 2010. Next Generation Internet.doi:10.1109/ngi.2010.5534462

work page doi:10.1109/ngi.2010.5534462 2010
[21]

Learning to Configure Separators in Branch-and-Cut

Sirui Li, Wenbin Ouyang, Max B Paulus, and Cathy Wu. Learning to Configure Separators in Branch-and-Cut. InAdvances in Neural Information Processing Systems, volume 37, 2023. doi:10.52202/075280-2622

work page doi:10.52202/075280-2622 2023
[22]

Pricing filtering in Dantzig-Wolfe decomposition.Operations Research Letters, 58:107207, 2025.doi:10.1016/j.orl.2024.107207

Abdellah Bulaich Mehamdi, Mathieu Lacroix, and Sébastien Martin. Pricing filtering in Dantzig-Wolfe decomposition.Operations Research Letters, 58:107207, 2025.doi:10.1016/j.orl.2024.107207

work page doi:10.1016/j.orl.2024.107207 2025
[23]

The surprising performance of random partial Benders decomposition

Jean Pauphilet and Yupeng Wu. The surprising performance of random partial Benders decomposition. Optimization Online, 2025. 32181

2025
[24]

The Benders decomposition algorithm: A literature review.European Journal of Operational Research, 259(3):801– 817, 2017.doi:10.1016/j.ejor.2016.12.005

Ragheb Rahmaniani, Teodor Gabriel Crainic, Michel Gendreau, and Walter Rei. The Benders decomposition algorithm: A literature review.European Journal of Operational Research, 259(3):801– 817, 2017.doi:10.1016/j.ejor.2016.12.005

work page doi:10.1016/j.ejor.2016.12.005 2017
[25]

Mathematical Programming , year =

Lara Scavuzzo, Karen Aardal, Andrea Lodi, and Neil Yorke-Smith. Machine learning augmented branch and bound for mixed integer linear programming.Mathematical Programming, 2024.doi: 10.1007/s10107-024-02130-y

work page doi:10.1007/s10107-024-02130-y 2024
[26]

Reinforcement learning for integer programming: Learning to cut

Yunhao Tang, Shipra Agrawal, and Yuri Faenza. Reinforcement learning for integer programming: Learning to cut. InInternational Conference on Machine Learning, pages 9367–9376. PMLR, 2020. doi:10.48550/arXiv.1906.04859

work page doi:10.48550/arxiv.1906.04859 2020
[27]

Cutting Plane Selection with Analytic Centers and Multiregression

Mark Turner, Timo Berthold, Mathieu Besançon, and Thorsten Koch. Cutting Plane Selection with Analytic Centers and Multiregression. InIntegration of Constraint Programming, Artificial Intelligence, and Operations Research (CPAIOR), Lecture Notes in Computer Science, pages 52–68. Springer, 2023.doi:10.1007/978-3-031-33271-5_4. 12

work page doi:10.1007/978-3-031-33271-5_4 2023
[28]

Adaptive cut selection in mixed-integer linear programming.Open Journal of Mathematical Optimization, 4:1–25, 2023

Mark Turner, Thorsten Koch, Felipe Serrano, and Michael Winkler. Adaptive cut selection in mixed-integer linear programming.Open Journal of Mathematical Optimization, 4:1–25, 2023. doi:10.5802/ojmo.25

work page doi:10.5802/ojmo.25 2023
[29]

Jie Wang, Zhihai Wang, Xijun Li, Yufei Kuang, Zhihao Shi, Fangzhou Zhu, Mingxuan Yuan, Jia Zeng, Yongdong Zhang, and Feng Wu. Learning to Cut via Hierarchical Sequence/Set Model for Efficient Mixed-Integer Programming.IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(12):9697–9713, 2024.doi:10.1109/TPAMI.2024.3432716

work page doi:10.1109/tpami.2024.3432716 2024
[30]

B. P. Welford. Note on a method for calculating corrected sums of squares and products.Techno- metrics, 4(3):419–420, 1962.doi:10.1080/00401706.1962.10490022

work page doi:10.1080/00401706.1962.10490022 1962
[31]

Biometrics Bulletin , author =

Frank Wilcoxon. Individual comparisons by ranking methods.Biometrics Bulletin, 1(6):80–83, 1945. doi:10.2307/3001968

work page doi:10.2307/3001968 1945
[32]

Optimized scenario reduction: Solving large-scale stochastic programs with quality guarantees.INFORMS Journal on Computing, 35(4):886–908, 2023.doi:10.1287/ijoc.2023.1295

Wei Zhang, Kai Wang, Alexandre Jacquillat, and Shuaian Wang. Optimized scenario reduction: Solving large-scale stochastic programs with quality guarantees.INFORMS Journal on Computing, 35(4):886–908, 2023.doi:10.1287/ijoc.2023.1295

work page doi:10.1287/ijoc.2023.1295 2023
[33]

Learning to select cutting planes in mixed-integer linear programming solving.Expert Systems with Applications, page 129924, 2025.doi:10.1016/j.eswa.2025.129924

Xuefeng Zhang, Liangyu Chen, Zhengfeng Yang, and Zhenbing Zeng. Learning to select cutting planes in mixed-integer linear programming solving.Expert Systems with Applications, page 129924, 2025.doi:10.1016/j.eswa.2025.129924. 13 Appendix A Subproblem Statistics This section provides detailed subproblem statistics for all instances and the distribution of ...

work page doi:10.1016/j.eswa.2025.129924 2025