pith. sign in

arxiv: 2605.21843 · v2 · pith:5JSJHXNQnew · submitted 2026-05-21 · 🧮 math.OC

Spectral analysis of the logit mapping and implications for stochastic user equilibrium algorithms

Debojjal Bagchi , Stephen D. Boyles This is my paper

Pith reviewed 2026-05-25 05:55 UTC · model grok-4.3

classification 🧮 math.OC
keywords stochastic user equilibriumlogit mappingJacobian analysismethod of successive averagesNewton methodtraffic assignmentspectral properties
0
0 comments X

The pith

The Jacobian of the logit mapping decomposes into an annihilator and a matrix with non-positive eigenvalues under monotone separable costs.

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

The paper shows that the Jacobian of the logit mapping for stochastic user equilibrium splits into two matrices. One annihilates differences among feasible path flow vectors. The other has only non-positive real eigenvalues when link costs are monotone non-decreasing and separable. These properties prove that constant-step MSA converges globally with asymptotic linear rate 1-s, support an adaptive step-size rule that keeps global convergence while adding the linear rate, and enable a Newton method whose structure makes large-network solves tractable without dense Hessians.

Core claim

The Jacobian of the logit mapping decomposes into two matrices, one annihilating differences of feasible path flow vectors and the other with all non-positive real eigenvalues when link costs are monotone non-decreasing and separable. This decomposition enables an adaptive constant step-size MSA with global convergence and asymptotic linear rate 1-s, plus a Newton method that exploits the Jacobian structure for tractable computation on large networks.

What carries the argument

Decomposition of the Jacobian of the logit mapping into an annihilator matrix and a factor whose eigenvalues are all non-positive reals.

If this is right

  • MSA with fixed step size s converges globally and achieves asymptotic linear rate 1-s.
  • An adaptive rule for the step size preserves global convergence while recovering the linear rate for large iterations.
  • The Newton method on the root-finding form of SUE becomes practical on large networks because the Jacobian structure avoids dense matrices and manifold projections.
  • Both algorithms run faster than prior methods on large networks or at high demand levels.

Where Pith is reading between the lines

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

  • The same decomposition might apply to other random-utility choice models if their Jacobians admit an analogous split.
  • Numerical verification of the eigenvalue sign on real networks would give a practical way to check the monotonicity assumption before running the algorithms.
  • The linear-rate result could guide step-size selection in other fixed-point iterations that arise in traffic equilibrium.

Load-bearing premise

Link costs are monotone non-decreasing and separable.

What would settle it

Finding a monotone separable cost function for which the second Jacobian factor has a positive eigenvalue would disprove the spectral property.

Figures

Figures reproduced from arXiv: 2605.21843 by Debojjal Bagchi, Stephen D. Boyles.

Figure 1
Figure 1. Figure 1: Convergence of RGAP (log scale) versus iteration for MSA with harmonic step-sizes and [PITH_FULL_IMAGE:figures/full_fig_p025_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Convergence rate of MSA for the Anaheim network across varying [PITH_FULL_IMAGE:figures/full_fig_p025_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Convergence of RGAP (log scale) versus iteration for [PITH_FULL_IMAGE:figures/full_fig_p030_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Runtimes (in seconds) to reach a RGAP level (log scale) [PITH_FULL_IMAGE:figures/full_fig_p032_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Braess network with one OD pair (O, D), three paths, and link cost functions shown on each link. Link l connecting nodes A and B is denoted as lAB. A.1 Computations of the Jacobian The Jacobian of path costs with respect to path flows, as defined in Equation (8), is J = C ′ (h) =   1 0 1 0 1 1 1 1 2   . We will use J := C ′ (h) as a shorthand throughout this example, consistent with Section 4.2. At ite… view at source ↗
read the original abstract

We analyze the Jacobian of the logit mapping for stochastic user equilibrium (SUE) and use it to develop two improved algorithms for path-based SUE. We show that the Jacobian decomposes into two matrices: one that annihilates differences of feasible path flow vectors, and another whose eigenvalues are all non-positive reals, provided link costs are monotone non-decreasing and separable. Using these properties, we first show that the method of successive averages (MSA) with a small constant step-size $s$ converges linearly at a rate $1-s$, with the largest admissible step-size depending on the eigenvalues of the Jacobian of the logit mapping. Building on this result, we develop an adaptive constant step-size rule that retains the global convergence of MSA while achieving asymptotic linear convergence. Our second algorithm is a Newton-based method using a reformulation of SUE as a root-finding problem. Unlike gradient-projection approaches that operate on the Hessian of the SUE objective function (a dense matrix), our method exploits the structure of the Jacobian of the logit mapping, making computations tractable and removing the need for manifold optimization. Numerical experiments show superlinear convergence on most tested networks, with our methods outperforming existing approaches on large networks or when demand is high. To our knowledge, this article is the first to report runtimes for logit-based SUE on networks as large as Chicago Regional and Philadelphia, providing a benchmark for future algorithmic development.

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.

Referee Report

3 major / 2 minor

Summary. The paper claims that the Jacobian of the logit mapping in path-based stochastic user equilibrium (SUE) decomposes into two matrices—one annihilating differences of feasible path flow vectors and the other with all non-positive real eigenvalues when link costs are monotone non-decreasing and separable. This decomposition is used to prove global convergence with asymptotic linear rate 1-s for an adaptive constant-step-size method of successive averages (MSA), and to derive a tractable Newton method for the root-finding reformulation of SUE that avoids dense Hessians and manifold optimization. Numerical experiments are reported to show superlinear convergence and outperformance versus existing methods on large networks including Chicago Regional and Philadelphia.

Significance. If the decomposition and spectral properties hold under the stated assumptions, the work supplies a theoretical basis for constant-step MSA variants with explicit linear rates and a structure-exploiting Newton scheme that scales to large instances. The reported first runtimes on Chicago Regional and Philadelphia networks constitute a concrete benchmark contribution for logit-based SUE algorithms.

major comments (3)
  1. [Abstract] Abstract and the section deriving the Jacobian decomposition: the eigenvalue property (all non-positive real eigenvalues for the second factor) is invoked to obtain the contraction rate 1-s for constant-step MSA and local stability of Newton, yet the manuscript supplies no detailed derivation or proof of this spectral claim, leaving the central convergence arguments unverified.
  2. [Abstract] The convergence analysis for the adaptive MSA (and the Newton local analysis) rests on the separability assumption for the non-positive eigenvalue guarantee. No counter-example, alternative bound, or numerical robustness check is provided for non-separable costs (e.g., turning penalties or multi-class interactions), which are standard and could introduce positive real eigenvalues that invalidate the rate 1-s claim.
  3. [Numerical experiments] Numerical experiments section: the claims of superlinear convergence on most tested networks and outperformance on large instances or high demand lack accompanying error analysis, tabulated metrics, or data tables, making it impossible to assess the practical magnitude of improvement or reproducibility.
minor comments (2)
  1. Notation for the step-size parameter s and the eigenvalue bound should be introduced consistently when first used in the MSA convergence statement.
  2. [Abstract] The abstract states that the Newton method removes the need for manifold optimization; a brief sentence clarifying how the structure exploitation achieves this would improve readability.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the careful review and constructive feedback on our manuscript. We address each major comment below and indicate planned revisions to strengthen the paper.

read point-by-point responses

  1. Referee: [Abstract] Abstract and the section deriving the Jacobian decomposition: the eigenvalue property (all non-positive real eigenvalues for the second factor) is invoked to obtain the contraction rate 1-s for constant-step MSA and local stability of Newton, yet the manuscript supplies no detailed derivation or proof of this spectral claim, leaving the central convergence arguments unverified.

    Authors: We acknowledge that the derivation of the spectral property in the Jacobian decomposition section is presented concisely. In the revised manuscript we will expand this section with a complete, self-contained proof of the eigenvalue claim under the monotone non-decreasing and separable link-cost assumptions. This will explicitly verify the contraction mapping used for the rate 1-s and the local stability of the Newton iteration. revision: yes

  2. Referee: [Abstract] The convergence analysis for the adaptive MSA (and the Newton local analysis) rests on the separability assumption for the non-positive eigenvalue guarantee. No counter-example, alternative bound, or numerical robustness check is provided for non-separable costs (e.g., turning penalties or multi-class interactions), which are standard and could introduce positive real eigenvalues that invalidate the rate 1-s claim.

    Authors: The referee correctly notes that the non-positive eigenvalue guarantee requires separability of link costs. This assumption is standard for the logit SUE model analyzed in the paper. In the revision we will add an explicit discussion subsection on the role of separability, clarifying that the rate 1-s is guaranteed only under this condition and that non-separable costs (such as turning penalties) may require different analysis. We will not add counter-examples or robustness checks, as these lie outside the stated scope of the separable case; instead we will reference relevant literature on non-separable SUE. revision: partial

  3. Referee: [Numerical experiments] Numerical experiments section: the claims of superlinear convergence on most tested networks and outperformance on large instances or high demand lack accompanying error analysis, tabulated metrics, or data tables, making it impossible to assess the practical magnitude of improvement or reproducibility.

    Authors: We agree that the numerical results section would be strengthened by additional quantitative detail. In the revised manuscript we will insert tables reporting iteration counts, CPU times, final gap values, and observed convergence rates for each algorithm and network. We will also include a short error-analysis paragraph discussing the measured superlinear behavior and factors influencing performance on the Chicago Regional and Philadelphia instances. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The derivation rests on an algebraic decomposition of the Jacobian of the logit mapping into an annihilator matrix and a second factor whose eigenvalues are shown non-positive under the explicit domain assumptions of monotone non-decreasing separable link costs. Convergence statements for constant-step MSA and the Newton method are obtained directly from this spectral property and the step-size rule; neither the eigenvalue sign nor the linear rate 1-s is obtained by fitting parameters to data or by renaming a self-citation. No load-bearing step reduces by construction to its own inputs, and the paper supplies no self-citation chain that substitutes for an independent proof. The result is therefore self-contained once the separability/monotonicity assumptions are granted.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claims rest on the standard domain assumption of monotone separable link costs plus the existing SUE formulation; no free parameters or new entities are introduced in the abstract.

axioms (1)
  • domain assumption Link costs are monotone non-decreasing and separable.
    Required to conclude that one matrix factor of the Jacobian has only non-positive real eigenvalues.

pith-pipeline@v0.9.0 · 5786 in / 1182 out tokens · 25997 ms · 2026-05-25T05:55:53.605384+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

118 extracted references · 118 canonical work pages

  1. [1]

    Boyles and Nicholas E

    Stephen D. Boyles and Nicholas E. Lownes and Avinash Unnikrishnan. Transportation Network Analysis. 2025

    work page 2025

  2. [2]

    Transportation Science , volume=

    The convergence of equilibrium algorithms with predetermined step sizes , author=. Transportation Science , volume=. 1982 , publisher=

    work page 1982

  3. [3]

    Falvo , title =

    Ron Larson and David C. Falvo , title =. 2009 , chapter =

    work page 2009

  4. [4]

    Bagchi, Debojjal , title =

    work page

  5. [5]

    The annals of mathematical statistics , pages=

    A stochastic approximation method , author=. The annals of mathematical statistics , pages=. 1951 , publisher=

    work page 1951

  6. [6]

    Transportation network analysis , author=. Vol. I: Static and Dynamic Traffic Assignment , year=

    work page

  7. [7]

    2011 , url =

    Lecture Notes , title =. 2011 , url =

    work page 2011

  8. [8]

    Transportmetrica A: Transport Science , volume=

    Convergence behavior for traffic assignment characterization metrics , author=. Transportmetrica A: Transport Science , volume=. 2021 , publisher=

    work page 2021

  9. [9]

    Taylor’s Theorem in One and Several Variables , year = "2011", AUTHOR =

    work page 2011

  10. [10]

    Convergence of traffic assignments: how much is enough?

    David Boyce and Biljana Ralevic-Dekic and Hillel Bar-Gera. Convergence of traffic assignments: how much is enough?. Journal of Transportation Engineering. 2004

    work page 2004

  11. [11]

    Solving a non-convex combined travel forecasting model by the method of successive averages with constant step sizes

    Hillel Bar-Gera and David Boyce. Solving a non-convex combined travel forecasting model by the method of successive averages with constant step sizes. Transportation Research Part B. 2006

    work page 2006

  12. [12]

    https://math.stackexchange.com/q/4896336 , URL =

    How far is the spectral norm from the Frobenius norm? , AUTHOR =. https://math.stackexchange.com/q/4896336 , URL =

    work page arXiv

  13. [13]

    https://math.stackexchange.com/q/1545118 , URL =

    A rank-one matrix is the product of two vectors , AUTHOR =. https://math.stackexchange.com/q/1545118 , URL =

    work page arXiv

  14. [14]

    1991 , publisher=

    Calculus, Volume 1 , author=. 1991 , publisher=

    work page 1991

  15. [15]

    1958 , publisher=

    Mathematical analysis , author=. 1958 , publisher=

    work page 1958

  16. [16]

    Gallier , title =

    Jean H. Gallier , title =. 2023 , url =

    work page 2023

  17. [17]

    Springer texts in statistics, Springer, New York, NY, doi , volume=

    Matrix algebra , author=. Springer texts in statistics, Springer, New York, NY, doi , volume=. 2007 , publisher=

    work page 2007

  18. [18]

    2002 , publisher=

    Real mathematical analysis , author=. 2002 , publisher=

    work page 2002

  19. [19]

    2003 , publisher=

    Complex analysis , author=. 2003 , publisher=

    work page 2003

  20. [20]

    Washington, DC , volume=

    Highway capacity manual , author=. Washington, DC , volume=

    work page

  21. [21]

    Transportation Research Record , volume=

    Modified bureau of public roads link function , author=. Transportation Research Record , volume=. 2023 , publisher=

    work page 2023

  22. [22]

    1964 , publisher=

    Traffic assignment manual for application with a large, high speed computer , author=. 1964 , publisher=

    work page 1964

  23. [23]

    Holomorphic function --- Wikipedia , The Free Encyclopedia

    Wikipedia contributors. Holomorphic function --- Wikipedia , The Free Encyclopedia. 2025

    work page 2025

  24. [24]

    1965 , publisher=

    Introduction to calculus and analysis , author=. 1965 , publisher=

    work page 1965

  25. [25]

    Applied Mathematics: Body and Soul - Calculus in Several Dimensions, Volume III , publisher =

    Kenneth Eriksson and Claes Johnson and Donald Estep , title =. Applied Mathematics: Body and Soul - Calculus in Several Dimensions, Volume III , publisher =. 2003 , chapter =. doi:10.1007/978-3-662-05800-8 , url =

    work page doi:10.1007/978-3-662-05800-8 2003

  26. [26]

    2011 , month =

    Hans Lindblad , title =. 2011 , month =

    work page 2011

  27. [27]

    Transportation Networks for Research , year =

    work page

  28. [28]

    The American Economic Review , volume=

    Pricing in urban and suburban transport , author=. The American Economic Review , volume=. 1963 , publisher=

    work page 1963

  29. [29]

    The American economic review , volume=

    Congestion theory and transport investment , author=. The American economic review , volume=. 1969 , publisher=

    work page 1969

  30. [30]

    Transportation Research Record , volume=

    Queue spillback and demand uncertainty in dynamic network loading , author=. Transportation Research Record , volume=. 2019 , publisher=

    work page 2019

  31. [31]

    2023 , publisher=

    Mathematical Analysis , author=. 2023 , publisher=

    work page 2023

  32. [32]

    The Quarterly Journal of Economics , volume=

    Some fallacies in the interpretation of social cost , author=. The Quarterly Journal of Economics , volume=. 1924 , publisher=

    work page 1924

  33. [33]

    Preprint , volume=

    Higher-order derivatives and taylor’s formula in several variables , author=. Preprint , volume=. 2005 , note =

    work page 2005

  34. [34]

    Transportmetrica B: Transport Dynamics , volume=

    Properties of dynamic user equilibrium solution: existence, uniqueness, stability, and robust solution methodology , author=. Transportmetrica B: Transport Dynamics , volume=. 2013 , publisher=

    work page 2013

  35. [35]

    Transportation Science , volume=

    On stochastic models of traffic assignment , author=. Transportation Science , volume=. 1977 , publisher=

    work page 1977

  36. [36]

    Transportation Research Part B: Methodological , volume=

    Some developments in equilibrium traffic assignment , author=. Transportation Research Part B: Methodological , volume=. 1980 , publisher=

    work page 1980

  37. [37]

    2004 , publisher=

    Prashker, Joseph N and Bekhor, Shlomo , journal=. 2004 , publisher=

    work page 2004

  38. [38]

    Equilibrium and advanced transportation modelling , pages=

    Stochastic assignment to transportation networks: models and algorithms , author=. Equilibrium and advanced transportation modelling , pages=. 1998 , publisher=

    work page 1998

  39. [39]

    Transportation Research , volume=

    A probabilistic multipath traffic assignment model which obviates path enumeration , author=. Transportation Research , volume=. 1971 , publisher=

    work page 1971

  40. [40]

    Transportation planning and technology , volume=

    An improved Dial's algorithm for logit-based traffic assignment within a directed acyclic network , author=. Transportation planning and technology , volume=. 2010 , publisher=

    work page 2010

  41. [41]

    Transportation research record , volume=

    Accelerating traffic assignment with customizable contraction hierarchies , author=. Transportation research record , volume=. 2020 , publisher=

    work page 2020

  42. [42]

    NCHRP Report , number=

    Highway traffic data for urbanized area project planning and design , author=. NCHRP Report , number=

    work page

  43. [43]

    Journal of Transportation Engineering , volume=

    Traffic assignment in practice: overview and guidelines for users , author=. Journal of Transportation Engineering , volume=. 1991 , publisher=

    work page 1991

  44. [44]

    Journal of Advanced Transportation , volume=

    Network reliability-based optimal toll design , author=. Journal of Advanced Transportation , volume=. 2008 , publisher=

    work page 2008

  45. [45]

    Transportation Research Part C: Emerging Technologies , volume=

    Traffic congestion pricing methodologies and technologies , author=. Transportation Research Part C: Emerging Technologies , volume=. 2011 , publisher=

    work page 2011

  46. [46]

    1924 , publisher=

    The Economics of Welfare , author=. 1924 , publisher=

    work page 1924

  47. [47]

    2009 , publisher=

    Liu, Henry X and He, Xiaozheng and He, Bingsheng , journal=. 2009 , publisher=

    work page 2009

  48. [48]

    Handbook of transport systems and traffic control , pages=

    Traffic congestion and congestion pricing , author=. Handbook of transport systems and traffic control , pages=. 2001 , publisher=

    work page 2001

  49. [49]

    Transportation Research Part B: Methodological , volume=

    Profit maximization by a private toll road with cars and trucks , author=. Transportation Research Part B: Methodological , volume=. 2016 , publisher=

    work page 2016

  50. [50]

    Equitable tolling: a case study on Sioux Falls , author=

    work page

  51. [51]

    Transport Reviews , volume=

    Models and algorithms for road network design: a review and some new developments , author =. Transport Reviews , volume=. 1998 , publisher=

    work page 1998

  52. [52]

    2011 , publisher=

    Estimation of origin-destination matrix from traffic counts: the state of the art , author=. 2011 , publisher=

    work page 2011

  53. [53]

    Journal of Transportation Engineering , volume=

    Scheduling of lane closures using genetic algorithms with traffic assignments and distributed simulations , author=. Journal of Transportation Engineering , volume=. 2004 , publisher=

    work page 2004

  54. [54]

    Journal of Applied Probability , volume=

    Avoiding the Braess paradox in non-cooperative networks , author=. Journal of Applied Probability , volume=. 1999 , publisher=

    work page 1999

  55. [55]

    Transportation Research Part A: Policy and Practice , volume=

    A capacity paradox in network design and how to avoid it , author=. Transportation Research Part A: Policy and Practice , volume=. 1998 , publisher=

    work page 1998

  56. [56]

    Willumsen, L. G. , title =. 1978 , notes =

    work page 1978

  57. [57]

    Transportation Research Part B: Methodological , volume=

    Heuristic algorithms for the bilevel origin-destination matrix estimation problem , author=. Transportation Research Part B: Methodological , volume=. 1995 , publisher=

    work page 1995

  58. [58]

    2005 , publisher=

    Selfish routing and the price of anarchy , author=. 2005 , publisher=

    work page 2005

  59. [59]

    Proceedings of the thiry-fourth annual ACM symposium on Theory of computing , pages=

    The price of anarchy is independent of the network topology , author=. Proceedings of the thiry-fourth annual ACM symposium on Theory of computing , pages=

    work page

  60. [60]

    Proceedings of the IEEE , volume=

    The price of anarchy in transportation networks: Data-driven evaluation and reduction strategies , author=. Proceedings of the IEEE , volume=. 2018 , publisher=

    work page 2018

  61. [61]

    2018 Annual American Control Conference (ACC) , pages=

    The price of anarchy for transportation networks with mixed autonomy , author=. 2018 Annual American Control Conference (ACC) , pages=. 2018 , organization=

    work page 2018

  62. [62]

    Transportation research procedia , volume=

    An optimization approach for deriving upper and lower bounds of transportation network vulnerability under simultaneous disruptions of multiple links , author=. Transportation research procedia , volume=. 2017 , publisher=

    work page 2017

  63. [63]

    Transportation Research Part B: Methodological , volume=

    Traffic assignment by paired alternative segments , author=. Transportation Research Part B: Methodological , volume=. 2010 , publisher=

    work page 2010

  64. [64]

    Transportation Research Part B: Methodological , volume=

    A path-based user-equilibrium traffic assignment algorithm that obviates path storage and enumeration , author=. Transportation Research Part B: Methodological , volume=. 2006 , publisher=

    work page 2006

  65. [65]

    Transportmetrica A: Transport Science , volume=

    Local User Cost Equilibrium: a bush-based algorithm for traffic assignment , author=. Transportmetrica A: Transport Science , volume=. 2014 , publisher=

    work page 2014

  66. [66]

    Transportation Science , volume=

    Origin-based algorithm for the traffic assignment problem , author=. Transportation Science , volume=. 2002 , publisher=

    work page 2002

  67. [67]

    Transportation Science , volume=

    A note on Bar-Gera's algorithm for the origin-based traffic assignment problem , author=. Transportation Science , volume=. 2012 , publisher=

    work page 2012

  68. [68]

    European Journal of Operational Research , volume=

    Bounding the inefficiency of logit-based stochastic user equilibrium , author=. European Journal of Operational Research , volume=. 2010 , publisher=

    work page 2010

  69. [69]

    2006 , publisher=

    A self instructing course in mode choice modeling: multinomial and nested logit models , author=. 2006 , publisher=

    work page 2006

  70. [70]

    Computers & Operations Research , volume=

    A variational inequality formulation for stochastic user equilibrium with a bounded choice set , author=. Computers & Operations Research , volume=. 2024 , publisher=

    work page 2024

  71. [71]

    Transportation research part B: methodological , volume=

    Post-disaster recovery sequencing strategy for road networks , author=. Transportation research part B: methodological , volume=. 2021 , publisher=

    work page 2021

  72. [72]

    , title =

    Crocker, Abigail J. , title =. 2024 , school =

    work page 2024

  73. [73]

    International journal of disaster risk reduction , volume=

    Long-term scheduling for road network disaster recovery , author=. International journal of disaster risk reduction , volume=. 2020 , publisher=

    work page 2020

  74. [74]

    Transportation Science , volume=

    On the convergence of the method of successive averages for calculating equilibrium in traffic networks , author=. Transportation Science , volume=. 2015 , publisher=

    work page 2015

  75. [75]

    Transportation Research Part B: Methodological , volume=

    An examination of convergence error in equilibrium traffic assignment models , author=. Transportation Research Part B: Methodological , volume=. 1988 , publisher=

    work page 1988

  76. [76]

    Mathematical Programming , volume=

    A unified approach to error bounds for structured convex optimization problems , author=. Mathematical Programming , volume=. 2017 , publisher=

    work page 2017

  77. [77]

    2003 , publisher=

    Finite-dimensional variational inequalities and complementarity problems , author=. 2003 , publisher=

    work page 2003

  78. [78]

    2026 , note =

    Linear Convergence of Method of Successive Averages in Logit-based Stochastic User Equilibrium , author =. 2026 , note =

    work page 2026

  79. [79]

    Transportation research record , volume=

    Application of cross-nested logit route choice model in stochastic user equilibrium traffic assignment , author=. Transportation research record , volume=. 2007 , publisher=

    work page 2007

  80. [80]

    Transportmetrica A: Transport Science , volume=

    A self-adaptive Armijo stepsize strategy with application to traffic assignment models and algorithms , author=. Transportmetrica A: Transport Science , volume=. 2013 , publisher=

    work page 2013

Showing first 80 references.