First-Order Convergence of Monotone Schemes for Hamilton--Jacobi Equations on the Wasserstein Space on Graphs
Pith reviewed 2026-05-22 04:22 UTC · model grok-4.3
The pith
Semi-discrete monotone schemes for Hamilton-Jacobi equations on graph Wasserstein spaces converge at first order.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
We prove first-order convergence of semi-discrete monotone finite difference schemes for Hamilton--Jacobi equations on the Wasserstein space over a finite graph. A central challenge is the boundary degeneracy of the Wasserstein simplex, which prevents the direct use of the standard L1 adjoint method and limits doubling-of-variables arguments to the suboptimal rate O(h^{1/2}). We address this issue by introducing a weighted L1 framework with a boundary-vanishing weight and by analyzing the corresponding weighted adjoint equation for the linearized operator of the scheme, featuring a new geometric drift term. Our proof relies on uniform bounds for the weighted adjoint variable and the mesh-}
What carries the argument
weighted L1 adjoint framework with boundary-vanishing weight and geometric drift term in the linearized operator
If this is right
- The schemes achieve first-order accuracy despite simplex boundary degeneracy.
- Uniform bounds on the weighted adjoint variable and mesh-parameter derivative control the error analysis.
- The approach works for two classes of monotone Hamiltonians via the bootstrap on gradient and semi-concavity bounds.
- The new geometric drift term appears in the weighted adjoint equation for the linearized scheme.
Where Pith is reading between the lines
- The weighted framework might extend to fully discrete schemes or to Wasserstein spaces over infinite graphs.
- The geometric drift term could connect to discrete curvature or geodesic structure on the underlying graph.
- Similar adjoint techniques might improve convergence proofs for related mean-field control problems on discrete structures.
Load-bearing premise
The bootstrap argument that produces uniform bounds on the discrete gradient and semi-concavity of the numerical solution holds for the two classes of monotone Hamiltonians considered.
What would settle it
Numerical computation of the scheme error on a small graph with a known exact solution that yields a convergence rate strictly less than first order would disprove the result.
read the original abstract
We prove first-order convergence of semi-discrete monotone finite difference schemes for Hamilton--Jacobi equations on the Wasserstein space over a finite graph. A central challenge is the boundary degeneracy of the Wasserstein simplex, which prevents the direct use of the standard $L^1$ adjoint method and limits doubling-of-variables arguments to the suboptimal rate $\mathcal O(h^{\frac 12})$ \cite{CDM25}. We address this issue by introducing a weighted $L^1$ framework with a boundary-vanishing weight and by analyzing the corresponding weighted adjoint equation for the linearized operator of the scheme, featuring a new geometric drift term. Our proof relies on uniform bounds for the weighted adjoint variable and the mesh-parameter derivative of the numerical solution. These estimates are derived from discrete gradient and semi-concavity bounds, obtained through a bootstrap argument for two classes of monotone Hamiltonians.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper claims to prove first-order convergence of semi-discrete monotone finite difference schemes for Hamilton--Jacobi equations on the Wasserstein space over a finite graph. It introduces a weighted L1 adjoint framework with a boundary-vanishing weight to overcome the degeneracy of the Wasserstein simplex, which otherwise limits doubling arguments to O(h^{1/2}). The proof derives uniform bounds on the weighted adjoint and mesh-parameter derivative from discrete gradient and semi-concavity estimates obtained via a bootstrap argument for two classes of monotone Hamiltonians.
Significance. If the bootstrap closure holds and yields h-independent bounds, the result would constitute a genuine improvement over the known suboptimal rate, extending monotone-scheme theory to a geometrically degenerate setting. The weighted adjoint construction with its new geometric drift term is a technically interesting device that could apply more broadly to other boundary-degenerate problems on simplices or graphs.
major comments (3)
- [abstract (proof strategy paragraph)] Proof strategy paragraph (abstract): the bootstrap argument that produces uniform bounds on the discrete gradient and semi-concavity is stated to work for the two classes of monotone Hamiltonians, yet no induction hypotheses, explicit constants, or boundary-layer estimates are supplied to confirm that the geometric drift term and vanishing weight factor remain controlled near the simplex boundary; without these, the weighted L1 adjoint estimate cannot be closed and the claimed first-order rate reverts to the known O(h^{1/2}) doubling bound.
- [weighted adjoint equation section] Section on the weighted adjoint equation: the new geometric drift term arising from the linearized operator is introduced to handle the Wasserstein geometry, but its precise form and the verification that it does not destroy the L1 contraction or the boundary-vanishing property of the weight are not shown in sufficient detail to guarantee the uniform bound on the adjoint variable.
- [Hamiltonian classes definition] Definition of the two Hamiltonian classes: the manuscript distinguishes two classes of monotone Hamiltonians for which the bootstrap is claimed to close, but the precise structural assumptions (e.g., convexity, growth, or monotonicity conditions) that enable absorption of the degeneracy are not stated explicitly enough to allow independent verification of the uniform bounds.
minor comments (2)
- [notation section] Notation for the mesh-parameter derivative should be introduced with a clear definition before its use in the adjoint estimates.
- [introduction] The reference to the suboptimal O(h^{1/2}) rate from doubling arguments (CDM25) is appropriate but would benefit from a one-sentence recap of why the standard L1 adjoint fails on the degenerate simplex.
Simulated Author's Rebuttal
We thank the referee for the careful and constructive review. The comments identify areas where additional explicit details will improve clarity and verifiability. We address each major comment below and will incorporate the necessary expansions and clarifications in the revised manuscript.
read point-by-point responses
-
Referee: [abstract (proof strategy paragraph)] Proof strategy paragraph (abstract): the bootstrap argument that produces uniform bounds on the discrete gradient and semi-concavity is stated to work for the two classes of monotone Hamiltonians, yet no induction hypotheses, explicit constants, or boundary-layer estimates are supplied to confirm that the geometric drift term and vanishing weight factor remain controlled near the simplex boundary; without these, the weighted L1 adjoint estimate cannot be closed and the claimed first-order rate reverts to the known O(h^{1/2}) doubling bound.
Authors: The induction hypotheses, explicit constants, and boundary-layer control are developed in detail in Section 4 of the manuscript. We assume O(1) bounds on the discrete gradient and O(1/h) semi-concavity, with constants depending only on the Hamiltonian class and independent of h. The vanishing weight (proportional to distance to the boundary) combined with the inward geometric drift ensures the estimates remain controlled near the boundary. We will revise the abstract to include a concise reference to these hypotheses and the closure mechanism for improved readability. revision: yes
-
Referee: [weighted adjoint equation section] Section on the weighted adjoint equation: the new geometric drift term arising from the linearized operator is introduced to handle the Wasserstein geometry, but its precise form and the verification that it does not destroy the L1 contraction or the boundary-vanishing property of the weight are not shown in sufficient detail to guarantee the uniform bound on the adjoint variable.
Authors: We will add an explicit lemma deriving the precise form of the geometric drift term from the linearized scheme and proving, via a discrete integration-by-parts identity on the graph, that the term preserves the L1 contraction while remaining bounded by the mesh size. This ensures it does not counteract the boundary-vanishing property of the weight, yielding the uniform bound on the weighted adjoint. The revised section will include this verification. revision: yes
-
Referee: [Hamiltonian classes definition] Definition of the two Hamiltonian classes: the manuscript distinguishes two classes of monotone Hamiltonians for which the bootstrap is claimed to close, but the precise structural assumptions (e.g., convexity, growth, or monotonicity conditions) that enable absorption of the degeneracy are not stated explicitly enough to allow independent verification of the uniform bounds.
Authors: We agree that the structural assumptions require more explicit statement. In the revision we will expand the definitions in Section 2.3 to list the precise conditions: Class I requires convexity in the gradient variable with quadratic growth, while Class II requires a monotonicity condition with linear growth. These are the properties that permit absorption of the degeneracy terms during the bootstrap. A short remark will explain how the assumptions close the estimates. revision: yes
Circularity Check
Direct proof via weighted adjoint and bootstrap; no reduction to self-inputs or self-citation chains
full rationale
The derivation introduces a weighted L1 adjoint with a new geometric drift term derived from the scheme itself, then obtains the required uniform bounds on the weighted adjoint and mesh-parameter derivative from discrete gradient and semi-concavity estimates produced by an internal bootstrap argument for the two monotone Hamiltonian classes. This chain does not reduce the target first-order rate to a fitted constant, a renamed input, or a load-bearing self-citation whose validity is presupposed. The cited CDM25 result is used only to motivate the suboptimal baseline, not to close the new argument. The paper therefore remains self-contained against external benchmarks.
Axiom & Free-Parameter Ledger
axioms (1)
- domain assumption The Hamiltonians belong to one of two monotone classes for which discrete gradient and semi-concavity bounds can be bootstrapped.
invented entities (1)
-
Weighted L1 adjoint framework with boundary-vanishing weight
no independent evidence
Lean theorems connected to this paper
-
IndisputableMonolith/Cost/FunctionalEquation.leanwashburn_uniqueness_aczel unclear?
unclearRelation between the paper passage and the cited Recognition theorem.
bootstrap argument for two classes of monotone Hamiltonians... uniform bounds for the weighted adjoint variable and the mesh-parameter derivative
-
IndisputableMonolith/Foundation/RealityFromDistinction.leanreality_from_one_distinction unclear?
unclearRelation between the paper passage and the cited Recognition theorem.
weighted L1 framework with a boundary-vanishing weight... geometric drift term S
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
-
[1]
G. Barles and P. E. Souganidis. Convergence of approximation schemes for fully nonlinear second order equations.Asymptotic Anal., 4(3):271–283, 1991
work page 1991
-
[2]
E. Bayraktar, A. Cosso, and H. Pham. Randomized dynamic programming principle and Feynman-Kac representation for optimal control of McKean-Vlasov dynamics.Trans. Amer. Math. Soc., 370(3):2115– 2160, 2018
work page 2018
-
[3]
A. Bensoussan, J. Frehse, and S. C. P. Yam. The master equation in mean field theory.J. Math. Pures Appl. (9), 103(6):1441–1474, 2015
work page 2015
-
[4]
F. Cagnetti, D. Gomes, and H. V. Tran. Convergence of a semi-discretization scheme for the Hamilton- Jacobi equation: a new approach with the adjoint method.Appl. Numer. Math., 73:2–15, 2013
work page 2013
-
[5]
P. Cardaliaguet, S. Daudin, J. Jackson, and P. E. Souganidis. An algebraic convergence rate for the optimal control of McKean-Vlasov dynamics.SIAM J. Control Optim., 61(6):3341–3369, 2023
work page 2023
-
[6]
P. Cardaliaguet and A. Porretta. An introduction to mean field game theory. InMean Field Games, volume 2281 ofLecture Notes in Math., pages 1–158. Springer, Cham, 2020
work page 2020
-
[7]
R. Carmona and F. Delarue.Probabilistic Theory of Mean Field Games with Applications I: Mean Field FBSDEs, Control, and Games, volume 83 ofProbability Theory and Stochastic Modelling. Springer, Cham, 2018
work page 2018
-
[8]
R. Carmona and F. Delarue.Probabilistic Theory of Mean Field Games with Applications II: Mean Field Games with Common Noise and Master Equations, volume 84 ofProbability Theory and Stochastic Mod- elling. Springer, Cham, 2018
work page 2018
-
[9]
S.-N. Chow, W. Huang, Y. Li, and H. Zhou. Fokker-Planck equations for a free energy functional or Markov process on a graph.Arch. Ration. Mech. Anal., 203(3):969–1008, 2012
work page 2012
-
[10]
S.-N. Chow, W. Li, and H. Zhou. A discrete Schrödinger equation via optimal transport on graphs.J. Funct. Anal., 276(8):2440–2469, 2019
work page 2019
- [11]
-
[12]
M. G. Crandall and P.-L. Lions. Two approximations of solutions of Hamilton-Jacobi equations.Math. Comp., 43(167):1–19, 1984
work page 1984
-
[13]
J. Cui, T. Dang, and C. Mou. Finite difference schemes for Hamilton–Jacobi equation on wasserstein space on graphs.To appear in SIAM Journal on Numerical Analysis, 2026
work page 2026
-
[14]
J. Cui, S. Liu, and H. Zhou. Wasserstein Hamiltonian flow with common noise on graph.SIAM J. Appl. Math., 83(2):484–509, 2023
work page 2023
- [15]
- [16]
-
[17]
Y. Feng, Y. Xiang, and H. Zhou. Discrete mean field games on finite graphs as initial value optimization, 2026
work page 2026
- [18]
- [19]
- [20]
- [21]
-
[22]
W. Gangbo and A. Tudorascu. On differentiability in the Wasserstein space and well-posedness for Hamilton-Jacobi equations.J. Math. Pures Appl. (9), 125:119–174, 2019
work page 2019
-
[23]
J. Hofbauer and K. Sigmund.The Theory of Evolution and Dynamical Systems: Mathematical Aspects of Selection, volume 7 ofLondon Mathematical Society Student Texts. Cambridge University Press, Cam- bridge, 1988
work page 1988
- [24]
-
[25]
Lang.Real and Functional Analysis, volume 142 ofGraduate Texts in Mathematics
S. Lang.Real and Functional Analysis, volume 142 ofGraduate Texts in Mathematics. Springer-Verlag, New York, third edition, 1993
work page 1993
-
[26]
J.-M. Lasry and P.-L. Lions. Mean field games.Jpn. J. Math., 2(1):229–260, 2007
work page 2007
- [27]
-
[28]
J. Maas. Gradient flows of the entropy for finite Markov chains.J. Funct. Anal., 261(8):2250–2292, 2011
work page 2011
-
[29]
S. Mayorga and A. Święch. Finite dimensional approximations of Hamilton-Jacobi-Bellman equations for stochastic particle systems with common noise.SIAM J. Control Optim., 61(2):820–851, 2023
work page 2023
-
[30]
A. Mielke. Geodesic convexity of the relative entropy in reversible Markov chains.Calc. Var. Partial Differential Equations, 48(1-2):1–31, 2013
work page 2013
-
[31]
C.-W. Shu. High order weighted essentially nonoscillatory schemes for convection dominated problems. SIAM Rev., 51(1):82–126, 2009
work page 2009
-
[32]
P. Souganidis. Approximation schemes for viscosity solutions of Hamilton-Jacobi equations.J. Differential Equations, 59(1):1–43, 1985
work page 1985
-
[33]
E. Tadmor. Local error estimates for discontinuous solutions of nonlinear hyperbolic equations.SIAM J. Numer. Anal., 28(4):891–906, 1991. Department of Applied Mathematics, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, SAR, China. Email address:jianbo.cui@polyu.edu.hk;tonghe.dang@polyu.edu.hk(Corresponding author)
work page 1991
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.