arxiv: 2604.02661 · v1 · submitted 2026-04-03 · 🧮 math.OC

Recognition: no theorem link

Quantum Optimisation for Transport Vulnerability Identification

Junxiang Xu , Chence Niu , Divya Jayakumar Nair , Vinayak Dixit

Authors on Pith no claims yet

Pith reviewed 2026-05-13 19:48 UTC · model grok-4.3

classification 🧮 math.OC

keywords transport network vulnerabilityquantum optimizationQUBO formulationlink failure analysisD-Wave annealingMINLP reformulationurban resiliencenonlinear interactions

0 comments

The pith

Reformulating transport vulnerability analysis as a QUBO enables quantum optimization to solve for simultaneous link failures on networks with thousands of links in minutes.

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

The paper shows how to convert the problem of finding the most damaging sets of simultaneous link failures in a transport network into a form that quantum computers can solve directly. Traditional approaches either become impossible to compute when considering many links at once or simplify the effects by adding them up linearly, missing real interactions. By turning the bi-level optimization model into a quadratic unconstrained binary optimization problem, the method explores all possible disruption combinations in parallel on quantum hardware. Tests on city-scale networks demonstrate that large instances finish in minutes rather than the hours required by classical algorithms, while maintaining accuracy on nonlinear effects. This opens the door to practical vulnerability assessments that better reflect how real networks behave under multiple disruptions.

Core claim

The bi-level mixed-integer nonlinear program for identifying critical link sets in transport networks can be exactly recast as a quadratic unconstrained binary optimization model whose quadratic terms directly encode the nonlinear interaction costs of simultaneous failures. When solved on D-Wave quantum annealers for networks ranging from 914 to 6018 links, the approach yields optimal or near-optimal disruption scenarios within 2.8 to 31.2 minutes, representing a one-to-two order of magnitude speedup over classical metaheuristic solvers while preserving the nonlinear effects.

What carries the argument

The QUBO reformulation of the bi-level MINLP vulnerability model, which maps link failure decisions to binary variables and nonlinear interaction effects to quadratic penalty terms.

If this is right

Multiple simultaneous link failures can now be evaluated without combinatorial explosion limiting the analysis to small numbers of links.
Nonlinear interaction effects between failures are captured directly rather than approximated by linear sums.
Large real-world networks become solvable on current quantum hardware in practical run times.
Hybrid quantum-classical workflows can validate results on small instances before scaling to full networks.

Where Pith is reading between the lines

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

If quantum hardware continues to scale, the same framework could handle national or continental transport systems.
The QUBO structure may transfer to vulnerability analysis in power grids or communication networks with similar failure interactions.
Future extensions could incorporate uncertainty in demand or repair times by adding stochastic terms to the quadratic objective.

Load-bearing premise

The transformation from the original bi-level MINLP to QUBO preserves the exact nonlinear interaction effects without distortion from penalty coefficients or hardware embedding.

What would settle it

Compare the disruption scenarios and total travel-time impacts returned by the quantum solver against those from an exhaustive classical enumeration on a small network or a high-precision nonlinear solver on a medium network; mismatch beyond numerical tolerance would falsify the equivalence.

Figures

Figures reproduced from arXiv: 2604.02661 by Chence Niu, Divya Jayakumar Nair, Junxiang Xu, Vinayak Dixit.

**Figure 1.** Figure 1: Schematic topology of the Nguyen-Dupuis network. 6.2 Details of research results To construct the QUBO objective function, this study first calculates the marginal disruption impact s c for each link and the pairwise interaction coefficients  st for all link pairs. These values quantify how single and simultaneous disruptions affect the total system travel time (TSTT) under medium demand conditions. As sh… view at source ↗

**Figure 2.** Figure 2: Convergence of SQA energy [PITH_FULL_IMAGE:figures/full_fig_p015_2.png] view at source ↗

**Figure 4.** Figure 4: Energy distributions of disruption combinations under different values of k . 6.2.3 TSTT Results and Analysis [PITH_FULL_IMAGE:figures/full_fig_p017_4.png] view at source ↗

**Figure 5.** Figure 5: shows the growth surface of TSTT under multiple road link disruption scenarios, focusing on the top five combinations. The results reveal a sharp and nonlinear escalation in TSTT as the number of disrupted road links increases, which indicates that network vulnerability is shaped by complex interaction effects rather than the simple accumulation of individual failures. A small set of road links, particular… view at source ↗

**Figure 6.** Figure 6: Sensitivity analysis results of remaining capacity ratio e (2) Sensitivity analysis on the penalty coefficient  The sensitivity analysis of the penalty coefficient  is presented in [PITH_FULL_IMAGE:figures/full_fig_p018_6.png] view at source ↗

**Figure 7.** Figure 7: Sensitivity analysis on the penalty coefficient. While the previous sensitivity analysis considered  at broad magnitudes, it is also necessary to examine intermediate values to ensure that the robustness is not an artefact of coarse scaling. To this end, [PITH_FULL_IMAGE:figures/full_fig_p019_7.png] view at source ↗

**Figure 10.** Figure 10: Energy curve for multiple disruptions in the SF network using D-Wave. (2) Energy values and runtime comparison [PITH_FULL_IMAGE:figures/full_fig_p023_10.png] view at source ↗

**Figure 11.** Figure 11: Energy curve for multiple disruptions in the Anaheim network using D-Wave [PITH_FULL_IMAGE:figures/full_fig_p024_11.png] view at source ↗

**Figure 12.** Figure 12: Runtime comparison of algorithms on the SF network [PITH_FULL_IMAGE:figures/full_fig_p025_12.png] view at source ↗

read the original abstract

Transport network vulnerability analysis plays a crucial role in safeguarding urban resilience. Traditional vulnerability identification approaches have provided valuable insights, yet they face two major limitations. First, the number of disruption scenarios increases combinatorially with the number of disrupted links considered simultaneously, making classical approaches computationally prohibitive. Second, most studies approximate the impacts of multiple simultaneous link failures through linear aggregation, which fails to capture the nonlinear interaction effects observed in real networks. To address these gaps, we reformulate the bi-level Mixed-Integer Nonlinear Programming (MINLP) model into a quantum-compatible Quadratic Unconstrained Binary Optimisation (QUBO) structure, enabling parallel exploration of complex disruption scenarios while incorporating nonlinear interaction effects. We develop a hybrid optimisation framework that integrates the quantum optimisation algorithm with the Frank-Wolfe method to validate the model's effectiveness on the small-scale network. Then, we further verify the framework through the D-Wave hardware across benchmark networks of different scales, including Sioux Falls, Anaheim, Chicago Sketch, and Berlin Full, to examine scalability and feasibility. The results show that this framework achieves strong solvability and stability. In particular, optimisation for large and larger networks is completed within minutes (Approximately 2.8 minutes for the 914-link, 9.8 minutes for the 2950-link, and 31.2 minutes for the 6018-link on D-Wave), demonstrating a computational efficiency improvement by one to two orders of magnitude compared with classical metaheuristic algorithms. These findings highlight the feasibility and potential of applying quantum computing to network vulnerability identification and open a new avenue for resilience-oriented planning.

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

The paper maps a transport vulnerability MINLP to QUBO for D-Wave and reports concrete scaling times, but skips direct validation against exact classical solutions on small instances.

read the letter

The main contribution is turning the bi-level MINLP for multi-link disruption scenarios into a QUBO that runs on quantum hardware. They add a hybrid Frank-Wolfe step for small cases and then test on standard benchmarks including Sioux Falls, Anaheim, Chicago Sketch, and Berlin Full. The reported D-Wave times—around 2.8 minutes for the 914-link network, 9.8 for 2950 links, and 31.2 for 6018 links—beat the classical metaheuristics they compare against by one to two orders of magnitude. That scaling data is the clearest new piece and shows the approach can handle the combinatorial growth without falling back to linear approximations of failure interactions. The work is honest about using real networks and gives explicit hardware runtimes rather than just theoretical claims. The soft spot is the missing cross-check. No section compares the QUBO outputs (link sets or objective values) to an exact classical MINLP solver on any instance small enough for both to reach optimality. Penalty terms and embedding overhead can distort results, so the claim that nonlinear effects are preserved accurately rests on an untested assumption. Without that side-by-side, the speed advantage is hard to interpret as reliable. This is for transportation researchers already working on network resilience who want to test quantum tools. A reader who knows the classical models will get value from the scaling numbers, but anyone planning to reuse the method will need the validation details filled in. I would send it to peer review. The application is new enough and the empirical results concrete enough to justify referee time, even with the validation gap.

Referee Report

2 major / 2 minor

Summary. The paper claims to reformulate a bi-level MINLP model for transport network vulnerability identification—capturing nonlinear interaction effects of simultaneous link failures—into an equivalent QUBO form. It develops a hybrid quantum-classical framework (integrating quantum optimization with the Frank-Wolfe method) and demonstrates it on D-Wave hardware for benchmark networks (Sioux Falls, Anaheim, Chicago Sketch, Berlin Full), reporting runtimes of minutes (e.g., 2.8 min for 914 links, 9.8 min for 2950 links, 31.2 min for 6018 links) and 1–2 orders of magnitude speedup over classical metaheuristics.

Significance. If the QUBO reformulation is shown to be faithful to the original MINLP, the work would provide a concrete demonstration of quantum annealing applied to a combinatorially hard network resilience problem, enabling analysis of larger disruption scenarios than classical methods allow. The explicit hardware runtimes and hybrid validation on real benchmarks constitute a practical contribution to quantum optimization in transportation.

major comments (2)

[QUBO reformulation and hybrid framework sections] The central claim that the bi-level MINLP is reformulated into an equivalent QUBO without material error from penalty terms or embedding is not supported by any side-by-side comparison of optimal link sets or objective values against an exact classical MINLP solver on small instances (e.g., Sioux Falls) that both methods can solve to optimality. This verification is required to confirm that nonlinear interaction effects are preserved.
[Numerical experiments and results] Section reporting D-Wave results (including the 2.8 min / 9.8 min / 31.2 min runtimes) provides no quantitative error analysis (e.g., objective-value gap or Hamming distance between quantum and classical solutions) or details on penalty-coefficient selection and minor-embedding overhead, leaving the accuracy of the reported speedups unverified.

minor comments (2)

[Model formulation] Notation for the upper- and lower-level variables and the penalty terms in the QUBO should be introduced with explicit definitions and cross-references to the original MINLP constraints.
[Abstract and results] The abstract and results text should clarify which networks correspond to the cited link counts (914, 2950, 6018) and whether the reported times include embedding or only annealing.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which identify key areas where additional validation will strengthen the manuscript. We address each major comment below and will incorporate the suggested revisions to improve the rigor of the QUBO equivalence claims and experimental reporting.

read point-by-point responses

Referee: [QUBO reformulation and hybrid framework sections] The central claim that the bi-level MINLP is reformulated into an equivalent QUBO without material error from penalty terms or embedding is not supported by any side-by-side comparison of optimal link sets or objective values against an exact classical MINLP solver on small instances (e.g., Sioux Falls) that both methods can solve to optimality. This verification is required to confirm that nonlinear interaction effects are preserved.

Authors: We agree that a direct side-by-side comparison with an exact classical MINLP solver on small instances such as Sioux Falls is necessary to rigorously confirm equivalence and preservation of nonlinear effects. The manuscript validates the hybrid quantum-Frank-Wolfe framework on small networks but does not include exact MINLP solver comparisons. In the revised version, we will add this analysis for Sioux Falls, reporting optimal link sets, objective values, and any discrepancies between the original MINLP (solved to optimality) and the QUBO formulation. revision: yes
Referee: [Numerical experiments and results] Section reporting D-Wave results (including the 2.8 min / 9.8 min / 31.2 min runtimes) provides no quantitative error analysis (e.g., objective-value gap or Hamming distance between quantum and classical solutions) or details on penalty-coefficient selection and minor-embedding overhead, leaving the accuracy of the reported speedups unverified.

Authors: We acknowledge that quantitative error metrics and implementation details are required to verify the accuracy of the D-Wave results and speedups. The current manuscript reports runtimes and overall comparisons but omits error analysis and specifics on penalties and embedding. We will revise the numerical experiments section to include objective-value gaps, Hamming distances where relevant, penalty-coefficient selection methodology, and minor-embedding overhead details for the benchmark networks. revision: yes

Circularity Check

0 steps flagged

No significant circularity; reformulation and external benchmarks are independent

full rationale

The paper presents a mathematical reformulation of a bi-level MINLP model into QUBO form, followed by hybrid validation on small networks and D-Wave runs on standard benchmark networks (Sioux Falls, Anaheim, etc.). Runtimes are compared directly to external classical metaheuristic algorithms, with no fitted parameters, self-definitional loops, or load-bearing self-citations that reduce the claimed speedup or equivalence to a tautology by construction. The derivation chain remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the mathematical equivalence of the MINLP-to-QUBO reformulation and on the assumption that D-Wave hardware can reliably solve the resulting instances at the reported scales.

axioms (1)

domain assumption The bi-level MINLP for vulnerability identification admits an exact or sufficiently accurate QUBO encoding that preserves nonlinear interaction effects.
Invoked when the abstract states the reformulation enables parallel exploration while incorporating nonlinear effects.

pith-pipeline@v0.9.0 · 5591 in / 1301 out tokens · 43978 ms · 2026-05-13T19:48:11.719851+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

5 extracted references · 5 canonical work pages · 1 internal anchor

[1]

Bilevel optimization with sustainability perspective: A survey on applications

Bilevel optimization with sustainability perspective: a survey on applications. arXiv preprint arXiv:2406.07184. Chen, A., Yang, C., Kongsomsaksakul, S. & Lee, M

work page internal anchor Pith review Pith/arXiv arXiv
[2]

arXiv preprint arXiv:1811.11538

A tutorial on formulating and using QUBO models. arXiv preprint arXiv:1811.11538. Glover, F., Kochenberger, G., Hennig, R. and Du, Y.,

work page arXiv
[3]

Journal of Advanced Transportation, 2021, 5513311

Analysing the vulnerability of public transport networks. Journal of Advanced Transportation, 2021, 5513311. Nitheesh, K. & Bhavathrathan, B

work page 2021
[4]

Transportmetrica A: Transport Science, 1-31

Disaster -resilient transport network design considering risk -averse evacuation traffic demand. Transportmetrica A: Transport Science, 1-31. Niu, C., Irannezhad, E., Myers, C. & Dixit, V. 2025a. Quantum Computing in Transport Science: A Review. arXiv preprint arXiv:2503.21302. Niu, C., Rastogi, P., Soman, J., Tamuli, K. & Dixit, V. 2025b. Applying Quantu...

work page arXiv
[5]

Research Outputs BackgroundQuestionContent In this research In this research In this research Figure A

Larger networks Extend to Sioux Falls and other bigger benchmark networks, assess scalability. Research Outputs BackgroundQuestionContent In this research In this research In this research Figure A. Overall research technical approach framework diagram. Page 34 of 46 Appendix B: Steps and pseudocode for executing a hybrid optimisation algorithm Hybrid qua...

work page arXiv 1906