Pith. sign in

REVIEW 3 major objections 4 minor 40 references

An Exact System Optimum Assignment Model for Transit Demand Management

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Exact system-optimum transit assignment shows congestion-reduction potential of 36.35%, more than double the approximate estimate of 17.39%.

desk verdict A useful linearization insight and a novel model combination, but the headline 'exact SO' result is really the optimum of a restricted no-denied-boarding model and the cost parameter table is internally inconsistent. read the letter →

arxiv 2505.22241 v1 pith:JIO7YV6C submitted 2025-05-28 math.OC

Xia Zhou , Mark Wallace , Daniel D. Harabor , Zhenliang Ma This is my paper
classification math.OC MSC 90B2090C10
keywords schedule-basedtransitassignmentproblemsystemoptimumuserequilibriumhardcapacityintegerlinearprogrammingHongKongMTRspatial-temporalcompetingdemandflowscongestionreduction
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper claims that the system-optimum schedule-based transit assignment problem can be solved exactly while retaining hard train capacity, multi-line networks, and simultaneous departure-time and route choices. The enabling assumption is that in the optimum every waiting passenger boards the next train, which removes the need for simulation and turns the model into an integer linear program solvable to global optimality. On a five-line, twelve-origin-destination, 52,717-passenger subset of the Hong Kong MTR, the exact solution is $447,780, which is 36.35% below the user-equilibrium cost; an approximate system-optimum method gave only 17.39%. A sympathetic reader should care because this is the difference between policies aimed at a real upper bound on congestion relief and policies based on an underestimate that can also point at the wrong passenger groups to shift.

What carries the argument

The load-bearing mechanism is the 'no denied boarding' simplification: inequality (8), which normally requires the number of boarding passengers to be no more than those waiting and is resolved by simulation, is replaced by equality (25), which states that every passenger waiting for a train boards it. With the denied-boarding variables set to zero, all remaining constraints are linear in the space-time flow variables (passenger counts grouped by origin-destination pair, route, and train), and the objective in Eq. (24) is a weighted linear sum of in-vehicle, waiting, and early/late costs. The result is an integer linear program that can be handed to a standard integer-linear solver; this is why the paper can claim global optimality rather than the local optimum produced by simulation-based approximate methods.

What would settle it

Optimize the same Hong Kong MTR instance with the denied-boarding variables restored (replace equality (25) by inequality (8)); if the resulting system cost is below $447,780, the no-denied-boarding constraint is binding and the paper's 'exact SO' is not the global optimum of the original problem.

Watch

Extended reading notes

Core claim

The paper's central claim, on its own terms, is that the system optimum for the schedule-based transit assignment problem can be computed exactly in realistic settings: hard train capacity, multiple interacting lines, and passenger flows that compete over both departure times and routes. It achieves this by assuming no denied boarding in the system optimum, so the boarding inequality becomes an equality, the denied-boarding variables drop out, and the model becomes an integer linear program with a linear cost objective. Solving that model for the Hong Kong MTR case study gives a minimum system cost of $447,780, which is 36.35% lower than the user-equilibrium solution, whereas an approximate SO solution suggested only a 17.39% reduction. The exact solution also identifies different origin-destination pairs as the best targets for shifting passengers, for example OD '3-13' with $22.05 of system cost reduction per passenger shifted.

Load-bearing premise

The system optimum is assumed to have no denied boarding: every passenger waiting for a train can board it, so the optimum is found under that extra constraint; if the true optimum would leave some passengers on the platform, the computed solution is the optimum of a restricted problem rather than the exact SO.

Editorial extensions

If this is right

  • If the exact SO claim is correct, the Hong Kong MTR case study's congestion-reduction potential is 36.35%, more than double the 17.39% reported by approximate SO, so policies built on approximate optima understate what is achievable.
  • The exact SO solution identifies which origin-destination pairs deserve the most attention: in the case study, OD '3-13' contributes the most to system cost and gives the largest per-passenger improvement ($22.05).
  • Approximately 80% of passengers who need to change their departure time can be shifted within a 30-minute window, and about 94% within a 60-minute window, suggesting individualized incentives are practical.
  • The improvement ratio (1 - SO/UE) grows from 36% at current demand to 53% at 165% demand, while capacity level matters less for the SO than for the UE, because the SO has no denied-boarding waits.
  • The exact SO model remains solvable when extra routes and higher demand are added, with runtimes around 41-64 seconds on the test network.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The no-denied-boarding constraint means the computed $447,780 is the optimum of a restricted problem; if the unconstrained system optimum would ever leave passengers on the platform, the true SO cost would be no higher, so the 36.35% figure may be conservative, but the computed plan would then not be directly implementable.
  • Reintroducing the denied-boarding variables as integers would yield a harder but still exact model for the unconstrained SO; the paper does not attempt this, leaving the size of the gap from the constraint as an open question.
  • As a routine check before incentive design, one could run the same UE-versus-exact-SO comparison on any network with timetable, capacity, and smart-card demand data; the model requires no simulation and no behavioral parameters beyond the cost weights.
  • A testable prediction of the paper is that if passengers followed the SO schedule, platform queues would disappear; an experiment or agent-based simulation with the SO plan could check whether the predicted load redistribution actually removes denied boarding.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper proposes an exact system optimum (SO) model for the schedule-based transit assignment problem (STAP) with hard train capacity, multi-line networks, and simultaneous departure-time and route choice. The model is formulated as an integer linear program under a no-denied-boarding assumption and solved with Gurobi via MiniZinc. In a case study on part of the Hong Kong MTR network (5 lines, 12 OD pairs, 52,717 passengers), the authors report an exact-SO cost of $447,780, which is 36.35% lower than the user equilibrium (UE) cost and more than double the 17.39% reduction estimated by their approximate SO (Approx. SO) method. The paper also presents sensitivity analyses for capacity and demand levels and policy insights on passenger shifting and stop-skipping.

Significance. If the reported numbers were reliable, the claim that exact SO reveals a much larger congestion-reduction potential than approximate methods would be of practical value for transit demand management. The paper also demonstrates a fast integer-linear formulation and provides a detailed formal appendix and algorithm description. However, the current formulation solves a restricted, no-denied-boarding problem rather than the original STAP, and the cost parameter table is inconsistent with the model equations, so the central quantitative claims are not reproducible as stated. The contribution may be salvageable if reframed as a policy-constrained SO benchmark, but the present version overstates what it computes.

major comments (3)
  1. [Section IV-B, Eq. (25)] The simplification f^k_ne(s,l_t)(t,r) = g^k_s(t,r) and the setting db_k_s(t)=0 are extra constraints, not consequences of system optimality. The original model in Eqs. (1)-(20) includes denied boarding as a feasible state with its own cost in Eq. (17), so the feasible set of the SO problem is strictly restricted. When cumulative demand on a leg exceeds cumulative train capacity over the modeling horizon, the no-denied-boarding model is infeasible even though the original problem has a finite optimum. Therefore the reported $447,780 and the 36.35% improvement in Section VI-B are the optimum of a constrained problem, and they are an upper bound on the true SO cost of the original STAP, not the exact SO potential. The abstract and conclusion consequently overstate the comparison with Approx. SO.
  2. [Section VI-A, Table V vs. Eqs. (13), (15), (17)] Table V labels α as the waiting cost per unit time ($18/h) and β as the in-vehicle cost per unit time ($6/h), but Eqs. (13) and (15) define α as the in-vehicle cost and β as the waiting cost, and Eq. (17) uses β for denied-boarding waiting cost. The text says waiting time is weighted as three times in-vehicle time, which matches the table's values but not the equation labels. This internal inconsistency means the reported system costs, the percentages 17.39% and 36.35%, and the policy comparisons are not reproducible from the stated parameterization and must be corrected.
  3. [Section V-B / Section VI-B, model comparison] The comparison between Approx. SO and exact SO is asymmetric: the Approx. SO model (Section V-B) is evaluated with the simulation that allows denied boarding, whereas the exact SO model prohibits it. Thus the difference between 17.39% and 36.35% reflects a difference in problem definition as well as in solution quality. If the no-denied-boarding variant is the intended contribution, the paper should clearly frame it as a policy-constrained SO benchmark and not as the exact SO of the original STAP; alternatively, the exact SO should be solved with the denied-boarding variables included.
minor comments (4)
  1. [Section I, Introduction] There is a duplicated word in the sentence "This UE approach is useful for for approximating a system state" (second column, end of first paragraph).
  2. [Section III-C, Eq. (8)] The sentence "However the actual number that board the train - and the number who fail to board it - is predicted by a simulation rather than a mathematical constraint" is misleading because Appendix A provides formal constraints with a simulation variable; please clarify that simulation supplies the value of the boarding variable in the UE/Approx. SO models.
  3. [Appendix A, Eq. (A.3)] The condition (i=1 ∧ i=tct_l+1) in Eq. (A.3) is never satisfied, so the equation incorrectly states that the denied-boarding variable is never forced to zero for the first train or the last train-plus-one; it should use a logical OR (i=1 ∨ i=tct_l+1), consistent with Eqs. (5) and (6).
  4. [Table I] The last two rows of Table I are difficult to read because the checkmarks are not aligned with the column headers "UE", "Approx. SO", and "Exact SO"; please reformat the table for clarity.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the 36.35% figure is the computed optimum of an explicitly restricted MILP, not a refit of the input data; the main concerns are modeling restrictions and an internal label swap, not circular derivation.

full rationale

The derivation chain is self-contained. The exact SO model is formulated as an integer-linear program (Section V-C) using the objective in Eq. (24) and constraints Eqs. (1), (4)-(7), (9)-(20), (25)-(27). The cost coefficients (Table V) are taken from external sources [36] and [37]; no parameter is fitted to reproduce the UE, Approx. SO, or exact SO system costs, and the claimed 36.35% reduction is an output of the solver, not an input. Self-citations ([27], [28], [29], [30]) support simulation and algorithm components but are not load-bearing for the central claim: the exact SO optimum is produced by Gurobi on a MiniZinc model, an independent computational check. The one passage that could be mistaken for circularity is Section IV-B, where the paper asserts "in a system optimal solution, therefore, it is required that all passengers waiting for a train will be able to board it" and replaces Eq. (8) by Eq. (25). This is a modeling restriction, not a circular derivation: it constrains the feasible region of the original STAP, so the reported optimum is the exact optimum of the restricted no-denied-boarding model rather than of the full STAP with denied boarding. That is a validity or scope limitation, not a reduction of the result to its inputs. The comparison with Approx. SO is also asymmetric because Approx. SO permits denied boarding while the exact model forbids it, so the gap partly reflects different feasible regions; again, this is a modeling asymmetry rather than a circularity. I additionally note Table V swaps the labels of α and β relative to Table II and Eqs. (13)/(15); that is an internal-consistency error affecting cost interpretation, but it is not circularity.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The central numerical result depends mainly on the no-denied-boarding axiom, which converts the STAP into a MILP. Cost coefficients and network data are taken from prior literature or real-world sources rather than fitted. No new entities are postulated.

assumptions (3)
  • domain assumption In-vehicle travel times are constant, schedules are fixed, and train capacity is fixed.
    Section III-A states these assumptions for the transit network model. They are standard for schedule-based transit assignment and are not validated against real operations in this paper.
  • ad hoc to paper No passenger is denied boarding in the system optimum (db_k_s(t) = 0 for all k, t, s).
    Section IV-B asserts 'In a system optimal solution, therefore, it is required that all passengers waiting for a train will be able to board it.' This is not proven; it is imposed to make the model linear and solvable by a standard ILP solver. If it does not hold, the 'exact SO' is not the true optimum.
  • standard math The integer-linear model can be solved to proven optimality with Gurobi.
    Section V-C relies on MiniZinc 2.8.5 and Gurobi 9.5.1 to solve the model. This assumes the solver returns a globally optimal solution for the given instance.

how reviews work

0 comments
Cite this review

Pith. "Pith review of An Exact System Optimum Assignment Model for Transit Demand Management." pith.science (2026). https://pith.science/paper/JIO7YV6C

@misc{pith2026250522241,
  author       = {Pith},
  title        = {Pith review of: An Exact System Optimum Assignment Model for Transit Demand Management},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JIO7YV6C}},
  note         = {Machine review of arXiv:2505.22241}
}
read the original abstract

Mass transit systems are experiencing increasing congestion in many cities. The schedule-based transit assignment problem (STAP) involves a joint choice model for departure times and routes, defining a space-time path in which passengers decide when to depart and which route to take. User equilibrium (UE) models for the STAP indicates the current congestion cost, while a system optimum (SO) models can provide insights for congestion relief directions. However, current STAP methods rely on approximate SO (Approx. SO) models, which underestimate the potential for congestion reduction in the system. The few studies in STAP that compute exact SO solutions ignore realistic constraints such as hard capacity, multi-line networks, or spatial-temporal competing demand flows. The paper proposes an exact SO method for the STAP that overcomes these limitations. We apply our approach to a case study involving part of the Hong Kong Mass Transit Railway network, which includes 5 lines, 12 interacting origin-destination pairs and 52,717 passengers. Computing an Approx. SO solution for this system indicates a modest potential for congestion reduction measures, with a cost reduction of 17.39% from the UE solution. Our exact SO solution is 36.35% lower than the UE solution, which is more than double the potential for congestion reduction. We then show how the exact SO solution can be used to identify opportunities for congestion reduction: (i) which origin-destination pairs have the most potential to reduce congestion; (ii) how many passengers can be reasonably shifted; (iii) future system potential with increasing demand and expanding network capacity.

Figures

Figures reproduced from arXiv: 2505.22241 by the authors.

Figure 1
Figure 1. Frameworks for evaluating promotion policies in the transit system [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Comparison between Approx. SO and exact SO: (a) Measured system potential for promotions (current: UE, target: SO); (b) Potential cost reduction [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Extracted Hong Kong MTR network (adapted from web image: [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Comparison of total system costs among user equilibrium, Approx. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: System performance comparison: departure time choice only versus [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 8
Figure 8. Figure 8: Link flows over time periods: (a) Link: Tai Wai to Kowloon Tong; [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 7
Figure 7. Figure 7: The number of passengers needing to be shifted in the system by [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 10
Figure 10. Figure 10: Scenarios on demand levels 100%-165%: System cost evolution [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]
Figure 9
Figure 9. Figure 9: Scenarios on capacity levels 60%-140%: System cost evolution curves [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

40 extracted references · 40 canonical work pages

  1. [1]

    C. Ma, C. Wang, and X. Xu, ”A multi-objective robust optimization model for customized bus routes,” IEEE Trans. Intell. Transp. Syst., vol. 22, no. 4, pp. 2359–2370, Sep. 2020

    work page 2020

  2. [2]

    R. Yin, Z. Liu, and N. Zheng, ”A simulation-based model for continuous network design problem using Bayesian optimization,” IEEE Trans. Intell. Transp. Syst., vol. 23, no. 11, pp. 20352–20367, May 2022

    work page 2022

  3. [3]

    Q. An, X. Fu, D. Huang, Q. Cheng, and Z. Liu, ”Analysis of adding-runs strategy for peak-hour regular bus services,” Transportation Research Part E: Logistics and Transportation Review, vol. 143, p. 102100, Nov. 2020

    work page 2020

  4. [4]

    D. Rey, V . V . Dixit, J. L. Ygnace, and S. T. Waller, ”An endogenous lottery-based incentive mechanism to promote off-peak usage in con- gested transit systems,” Transp. Policy, vol. 46, pp. 46–55, Feb. 2016

    work page 2016

  5. [5]

    X. S. Lu, X. Zhang, J. Long, and Y . Li, ”Managing rail transit peak-hour congestion with step fare schemes,” Transportmetrica A, Transp. Sci., vol. 16, no. 3, pp. 1490–1511, Jan. 2020

    work page 2020

  6. [6]

    L. L. Cheung and A. S. Shalaby, ”System optimal re-routing transit assignment heuristic: A theoretical framework and large-scale case study,” Int. J. Transp. Sci. Technol., vol. 6, no. 4, pp. 287–300, Dec. 2017

    work page 2017

  7. [7]

    J. Sun, J. Wu, F. Xiao, Y . Tian, and X. Xu, ”Managing bottleneck congestion with incentives,” Transp. Res. B, Methodol., vol. 134, pp. 143–166, Apr. 2020

    work page 2020

  8. [8]

    Zhang and S

    P. Zhang and S. Qian, ”Path-based system optimal dynamic traffic assignment: A subgradient approach,” Transp. Res. B, Methodol., vol. 134, pp. 41–63, Apr. 2020

    work page 2020

Show all 40 references
  1. [9]

    Zhu and S

    F. Zhu and S. V . Ukkusuri, ”Efficient and fair system states in dy- namic transportation networks,” Transp. Res. B, Methodol., vol. 104, pp. 272–289, Oct. 2017

    2017

  2. [10]

    Hamdouch, M

    Y . Hamdouch, M. Florian, D. W. Hearn, and S. Lawphongpanich, ”Congestion pricing for multi-modal transportation systems,” Transp. Res. B, Methodol., vol. 41, no. 3, pp. 275–291, Mar. 2007

    2007

  3. [11]

    A. H. Chow, ”Dynamic system optimal traffic assignment–a state- dependent control theoretic approach,” Transportmetrica, vol. 5, no. 2, pp. 85–106, May 2009. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, OCTOBER 2024 13

    2009

  4. [12]

    J. T. van Essen and G. H. Correia, ”Exact formulation and comparison between the user optimum and system optimum solution for routing privately owned automated vehicles,” IEEE Trans. Intell. Transp. Syst., vol. 20, no. 12, pp. 4567–4578, Jan. 2019

    2019

  5. [13]

    A. H. Chow, ”Properties of system optimal traffic assignment with de- parture time choice and its solution method,” Transp. Res. B, Methodol., vol. 43, no. 3, pp. 325–344, Mar. 2009

    2009

  6. [14]

    Salonen and T

    M. Salonen and T. Toivonen, ”Modelling travel time in urban networks: Comparable measures for private car and public transport,” J. Transport Geogr., vol. 31, pp. 143–153, Jul. 2013

    2013

  7. [15]

    Hamdouch, H

    Y . Hamdouch, H. W. Ho, A. Sumalee, and G. Wang, ”Schedule-based transit assignment model with vehicle capacity and seat availability,” Transp. Res. B, Methodol., vol. 45, no. 10, pp. 1805–1830, Dec. 2011

    2011

  8. [16]

    Gentile, M

    G. Gentile, M. Florian, Y . Hamdouch, O. Cats, A. Nuzzolo, and M. Friedrich, ”The theory of transit assignment: basic modelling frame- works,” in Modelling Public Transport Passenger Flows in the Era of Intelligent Transport Systems: COST Action TU1004 (TransITS), Cham, Switze...

    2016

  9. [17]

    J. Xie, S. Zhan, S. C. Wong, and S. M. Lo, ”A schedule-based model for passenger-oriented train planning with operating cost and capacity constraints,” IEEE Trans. Intell. Transp. Syst., vol. 23, no. 11, pp. 21314–21328, May 2022

    2022

  10. [18]

    B. Mo, H. N. Koutsopoulos, Z. J. Shen, and J. Zhao, ”Robust path recommendations during public transit disruptions under demand uncer- tainty,” Transp. Res. B, Methodol., vol. 169, pp. 82–107, Mar. 2023

    2023

  11. [19]

    Yang and Q

    H. Yang and Q. Meng, ”Departure time, route choice and congestion toll in a queuing network with elastic demand,” Transp. Res. B, Methodol., vol. 32, no. 4, pp. 247–260, May 1998

    1998

  12. [20]

    Y . Tang, Y . Jiang, H. Yang, and O. A. Nielsen, ”Modeling and optimizing a fare incentive strategy to manage queuing and crowding in mass transit systems,” Transp. Res. B, Methodol., vol. 138, pp. 247–267, Aug. 2020

    2020

  13. [21]

    Q. Tian, P. Liu, G. P. Ong, and H. J. Huang, ”Morning commuting pattern and crowding pricing in a many-to-one public transit system with heterogeneous commuters,” Transp. Res. E, Logist. Transp. Rev., vol. 145, p. 102182, Jan. 2021

    2021

  14. [22]

    T. Y . Ma and J. P. Lebacque, ”Dynamic system optimal routing in multimodal transit network,” Transp. Res. Rec., vol. 2351, no. 1, pp. 76–84, Jan. 2013

    2013

  15. [23]

    Z. Y . Sun, Y . Li, W. C. Qu, and T. Muhammad, ”Unified framework for optimal routing choice under guidance information,” Complexity, vol. 2018, p. 4801489, Aug. 2018

    2018

  16. [24]

    Araldo, S

    A. Araldo, S. Gao, R. Seshadri, C. L. Azevedo, H. Ghafourian, Y . Sui, S. Ayaz, D. Sukhin, and M. Ben-Akiva, ”System-level optimization of multi- modal transportation networks for energy efficiency using personalized incentives: formulation, implementation, and performance,” T...

    2019

  17. [25]

    Yang and Y

    H. Yang and Y . Tang, ”Managing rail transit peak-hour congestion with a fare-reward scheme,” Transp. Res. B, Methodol., vol. 110, pp. 122–136, Apr. 2018

    2018

  18. [26]

    Khani, ”An online shortest path algorithm for reliable routing in schedule-based transit networks considering transfer failure probability,” Transp

    A. Khani, ”An online shortest path algorithm for reliable routing in schedule-based transit networks considering transfer failure probability,” Transp. Res. B, Methodol., vol. 126, pp. 549–564, Aug. 2019

    2019

  19. [27]

    X. Zhou, H. S. Mahmassani, and K. Zhang, ”Dynamic micro-assignment modeling approach for integrated multimodal urban corridor manage- ment,” Transp. Res. C, Emerg. Technol., vol. 16, no. 2, pp. 167–186, Apr. 2008

    2008

  20. [28]

    Ma and H

    Z. Ma and H. N. Koutsopoulos, ”Optimal design of promotion-based demand management strategies in urban rail systems,” Transp. Res. C, Emerg. Technol., vol. 109, pp. 155–173, Dec. 2019

    2019

  21. [29]

    B. Mo, Z. Ma, H. N. Koutsopoulos, and J. Zhao, ”Capacity-constrained network performance model for urban rail systems,” Transp. Res. Rec., vol. 2674, no. 5, pp. 59–69, May 2020

    2020

  22. [30]

    C. C. Lu, H. S. Mahmassani, and X. Zhou, ”Equivalent gap function- based reformulation and solution algorithm for the dynamic user equilib- rium problem,” Transp. Res. B, Methodol., vol. 43, no. 3, pp. 345–364, Mar. 2009

    2009

  23. [31]

    Verbas, H

    ¨O. Verbas, H. S. Mahmassani, and M. F. Hyland, ”Gap-based transit assignment algorithm with vehicle capacity constraints: Simulation-based implementation and large-scale application,” Transp. Res. B, Methodol., vol. 93, pp. 1–16, 2016

    2016

  24. [32]

    L. Pan, E. O. Waygood, and Z. Patterson, ”Would you wait? Bus choice behavior analysis considering various incentives,” Transp. Res. Rec., vol. 2676, no. 7, pp. 117–127, Jul. 2022

    2022

  25. [33]

    M ¨uller, M

    D. M ¨uller, M. G. M ¨uller, D. Kress, and E. Pesch, ”An algorithm selection approach for the flexible job shop scheduling problem: Choosing constraint programming solvers through machine learning,” Eur. J. Oper. Res., vol. 302, no. 3, pp. 874–891, Nov. 2022

    2022

  26. [34]

    Nethercote, P

    N. Nethercote, P. J. Stuckey, R. Becket, S. Brand, G. J. Duck, and G. Tack, ”MiniZinc: Towards a standard CP modelling language,” in Proc. Int. Conf. Principles Practice Constraint Programming, Berlin, Germany: Springer Berlin Heidelberg, 2007, pp. 529–543

    2007

  27. [35]

    Gurobi Optimization, LLC, Gurobi Optimizer Reference Manual, 2024

    2024

  28. [36]

    De Palma, M

    A. De Palma, M. Kilani, and S. Proost, ”Discomfort in mass transit and its implication for scheduling and pricing,” Transp. Res. B, Methodol., vol. 71, pp. 1–8, Jan. 2015

    2015

  29. [37]

    Balcombe, R

    R. Balcombe, R. Mackett, N. Paulley, J. Preston, J. Shires, H. Titheridge, M. Wardman, and P. White, ”The demand for public transport: a practical guide,” TRL Report, TRL593, Transport Research Laboratory, Woking- ham, UK, 2004

    2004

  30. [38]

    Y . Wang, B. De Schutter, T. J. van den Boom, B. Ning, and T. Tang, ”Efficient bilevel approach for urban rail transit operation with stop- skipping,” IEEE Trans. Intell. Transp. Syst., vol. 15, no. 6, pp. 2658–2670, Oct. 2014

    2014

  31. [39]

    Freyss, R

    M. Freyss, R. Giesen, and J. C. Mu ˜noz, ”Continuous approximation for skip-stop operation in rail transit,” Procedia Soc. Behav. Sci., vol. 80, pp. 186–210, Jun. 2013

    2013

  32. [40]

    Zhang, M

    L. Zhang, M. Zhang, W. Yang, and D. Dong, ”Golden ratio genetic algorithm based approach for modelling and analysis of the capacity expansion of urban road traffic network,” Comput. Intell. Neurosci., vol. 2015, no. 1, p. 512715, 2015. Xia Zhoureceived the B.S. degree in Engin...

    2015

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.