pith. sign in

arxiv: 2605.24232 · v1 · pith:ISPUVPR5new · submitted 2026-05-22 · 🧮 math.AP

Stability of optimal transport maps and second variation of the 2-Monge-Kantorovich distance

Pith reviewed 2026-06-30 14:45 UTC · model grok-4.3

classification 🧮 math.AP
keywords optimal transportMonge-Ampère equationstability estimatesBrenier potentialsMonge-Kantorovich distancelinearizationsecond variationconvex domains
0
0 comments X

The pith

Under Hölder regularity, optimal transport maps between non-degenerate densities on uniformly convex domains are Lipschitz stable in L² with respect to the 2-Monge-Kantorovich distance.

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

The paper derives quantitative stability estimates for optimal transport maps with quadratic cost. When source and target densities are Hölder continuous and non-degenerate on uniformly convex domains, the maps satisfy a Lipschitz bound in L² controlled by the 2-Monge-Kantorovich distance between the densities. They also obtain C^{1,α} stability controlled by L^p distances. Without the Hölder assumption the Brenier potentials are still Lipschitz stable in L² with respect to the L² distance of the densities. The key step is a characterization of the linear response of the Brenier potential obtained by linearizing the Monge-Ampère equation, which additionally yields an explicit second variation formula for the Monge-Kantorovich distance.

Core claim

Under the stated assumptions the linear response of the Brenier potential to perturbations of the source and target densities is given by the solution of the linearized Monge-Ampère equation in divergence form. This response implies the Lipschitz stability estimates for the maps and potentials, and it produces the explicit formula for the second variation of the quadratic Monge-Kantorovich distance.

What carries the argument

The linearized Monge-Ampère equation in divergence form, used to characterize the first variation of the Brenier potential along interpolations of the densities.

If this is right

  • The optimal transport map depends Lipschitz continuously on the pair of densities when measured in the 2-Monge-Kantorovich metric and L² norm.
  • C^{1,α} regularity of the map is controlled by L^p distances of the densities under Hölder assumptions.
  • Brenier potentials remain stable in L² even if the densities lose continuity.
  • The second variation of the 2-Monge-Kantorovich distance admits an explicit integral expression derived from the linearization.

Where Pith is reading between the lines

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

  • This linearization technique could be applied to derive stability results for other cost functions in optimal transport.
  • The stability estimates suggest quantitative convergence rates for approximations of transport maps when densities are perturbed by discretization error.
  • Connections to the geometry of the Wasserstein space may follow from the second variation formula.
  • Testing the formula on explicit examples such as uniform densities on balls would verify the second variation explicitly.

Load-bearing premise

Densities are non-degenerate on uniformly convex domains, with Hölder continuity needed for the map stability claims.

What would settle it

Finding a counterexample consisting of non-degenerate Hölder densities on a uniformly convex domain where the L² difference of the optimal maps is not bounded by a constant times the 2-Monge-Kantorovich distance between the densities.

read the original abstract

We establish several quantitative stability estimates for optimal transport maps between non-degenerate densities on uniformly convex domains for the quadratic cost. Under H\"older regularity assumptions, we prove Lipschitz $L^2$ (respectively $C^{1,\alpha}$) stability estimates for optimal transport maps in terms of the 2-Monge-Kantorovich distance (respectively $L^{p}$ distances) between pairs of source and target densities. When the continuity assumption is removed, we obtain a Lipschitz $L^2$ stability estimate for the Brenier potentials in terms of the $L^2$ distance between the source and target densities. The proofs rely on a precise characterization of the linear response of the Brenier potential along smooth interpolations of the data, obtained by linearizing the Monge-Amp\`ere equation in divergence form. As a further application of this approach, we derive an explicit formula for the second variation of the quadratic Monge-Kantorovich distance.

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

2 major / 2 minor

Summary. The paper establishes quantitative stability estimates for optimal transport maps between non-degenerate densities on uniformly convex domains for the quadratic cost. Under Hölder regularity assumptions on the densities, it proves Lipschitz L² stability estimates for the maps in terms of the 2-Monge-Kantorovich distance between source and target densities, and C^{1,α} stability in terms of L^p distances. Removing the continuity assumption yields a Lipschitz L² stability estimate for the Brenier potentials in terms of the L² distance between the densities. The proofs rely on a precise characterization of the linear response of the Brenier potential obtained by linearizing the Monge-Ampère equation in divergence form along smooth interpolations; as an application, an explicit formula for the second variation of the quadratic Monge-Kantorovich distance is derived.

Significance. If the central claims hold, the results supply useful quantitative stability controls for optimal transport maps and potentials under standard structural assumptions (uniform convexity of domains and non-degeneracy of densities). The linearization approach to the Monge-Ampère equation and the explicit second-variation formula constitute concrete contributions that can be applied in stability analysis and related PDE problems. The distinction between map stability under Hölder regularity and potential stability without continuity is clearly delineated.

major comments (2)
  1. [§3] §3 (linearization step): the derivation of the linearized Monge-Ampère equation in divergence form assumes sufficient smoothness of the interpolation to justify differentiation under the integral; the manuscript should explicitly verify that the error terms arising from this differentiation remain controlled by the 2-MK distance under the stated Hölder assumptions.
  2. [Theorem 1.2] Theorem 1.2 (map stability): the Lipschitz constant in the L² estimate appears to depend on the uniform convexity constants and the non-degeneracy bounds; the proof should confirm that this dependence is explicit and does not deteriorate when the Hölder exponent α approaches zero.
minor comments (2)
  1. [Abstract] The abstract states the results for 'pairs of source and target densities' but the precise interpolation path between densities is only defined later; a brief sentence in the abstract or introduction clarifying the interpolation would improve readability.
  2. [Introduction] Notation for the 2-Monge-Kantorovich distance is introduced without an explicit formula in the opening paragraph; adding the standard definition W_2^2(μ,ν) = inf ∫ |x-y|^2 dγ would help readers unfamiliar with the notation.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and constructive comments. Below we address the two major comments point by point, indicating the changes we will make to the manuscript.

read point-by-point responses
  1. Referee: [§3] §3 (linearization step): the derivation of the linearized Monge-Ampère equation in divergence form assumes sufficient smoothness of the interpolation to justify differentiation under the integral; the manuscript should explicitly verify that the error terms arising from this differentiation remain controlled by the 2-MK distance under the stated Hölder assumptions.

    Authors: We agree that the passage from the smooth interpolation to the linearized equation requires explicit control of the error terms arising from differentiation under the integral. In the revised manuscript we will add a short lemma in §3 that bounds these remainder terms in terms of the 2-MK distance, using only the Hölder regularity of the densities and the uniform convexity of the domains. The argument proceeds by approximating the given Hölder densities by smooth ones whose 2-MK distance is controlled, applying the smooth linearization, and passing to the limit with the error estimate. revision: yes

  2. Referee: [Theorem 1.2] Theorem 1.2 (map stability): the Lipschitz constant in the L² estimate appears to depend on the uniform convexity constants and the non-degeneracy bounds; the proof should confirm that this dependence is explicit and does not deteriorate when the Hölder exponent α approaches zero.

    Authors: The Lipschitz constant in Theorem 1.2 is explicit and depends on the uniform convexity constants, the non-degeneracy bounds, and the Hölder exponent α. As α → 0 the constant necessarily deteriorates; this is consistent with the paper’s separation of results, since map stability in L² requires positive Hölder regularity while the continuous case yields only potential stability. We will insert a remark immediately after the statement of Theorem 1.2 that records the explicit dependence on all parameters and notes the blow-up as α → 0. No change to the theorem statement itself is required. revision: partial

Circularity Check

0 steps flagged

No significant circularity

full rationale

The derivation relies on linearizing the Monge-Ampère equation in divergence form to characterize the linear response of the Brenier potential, followed by standard estimates under Hölder or continuity assumptions on non-degenerate densities. No quoted step reduces a claimed stability estimate or second-variation formula to a fitted parameter, self-defined quantity, or load-bearing self-citation; the argument is a direct PDE analysis whose inputs (uniform convexity, non-degeneracy) are independent of the output estimates.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The work rests on standard domain assumptions of optimal transport theory; no free parameters, new entities, or ad-hoc axioms are introduced in the abstract.

axioms (2)
  • domain assumption Densities are non-degenerate on uniformly convex domains
    Required setting for the stability statements (abstract opening sentence).
  • domain assumption Hölder regularity of densities for the C^{1,α} map estimates
    Explicitly assumed to obtain the stronger stability result.

pith-pipeline@v0.9.1-grok · 5711 in / 1440 out tokens · 53638 ms · 2026-06-30T14:45:49.331771+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

46 extracted references · 4 canonical work pages

  1. [1]

    D´ ecomposition polaire et r´ earrangement monotone des champs de vecteurs.CR Acad

    Yann Brenier. D´ ecomposition polaire et r´ earrangement monotone des champs de vecteurs.CR Acad. Sci. Paris S´ er. I Math., 305:805–808, 1987

  2. [2]

    InteriorW 2,p estimates for solutions of the monge-ampere equation.Annals of Mathematics, pages 135–150, 1990

    Luis A Caffarelli. InteriorW 2,p estimates for solutions of the monge-ampere equation.Annals of Mathematics, pages 135–150, 1990

  3. [3]

    Classical solvability in dimension two of the second boundary-value problem as- sociated with the Monge-Amp` ere operator.Ann

    Philippe Delano¨ e. Classical solvability in dimension two of the second boundary-value problem as- sociated with the Monge-Amp` ere operator.Ann. Inst. H. Poincar´ e Anal. Non Lin´ eaire, 8(5):443– 457, 1991

  4. [4]

    The regularity of mappings with a convex potential.Journal of the American Mathematical Society, 5(1):99–104, 1992

    Luis A Caffarelli. The regularity of mappings with a convex potential.Journal of the American Mathematical Society, 5(1):99–104, 1992

  5. [5]

    Boundary regularity of maps with convex potentials.Communications on pure and applied mathematics, 45(9):1141–1151, 1992

    Luis A Caffarelli. Boundary regularity of maps with convex potentials.Communications on pure and applied mathematics, 45(9):1141–1151, 1992

  6. [6]

    Boundary regularity of maps with convex potentials–ii.Annals of mathematics, 144(3):453–496, 1996

    Luis A Caffarelli. Boundary regularity of maps with convex potentials–ii.Annals of mathematics, 144(3):453–496, 1996

  7. [7]

    On the second boundary value problem for equations of Monge-Amp` ere type.J

    John Urbas. On the second boundary value problem for equations of Monge-Amp` ere type.J. Reine Angew. Math., 487:115–124, 1997

  8. [8]

    Regularity of optimal transport on compact, locally nearly spher- ical, manifolds.J

    Philippe Delano¨ e and Yuxin Ge. Regularity of optimal transport on compact, locally nearly spher- ical, manifolds.J. Reine Angew. Math., 646:65–115, 2010

  9. [9]

    Locally nearly spherical surfaces are almost-positivelyc-curved

    Philippe Delano¨ e and Yuxin Ge. Locally nearly spherical surfaces are almost-positivelyc-curved. Methods Appl. Anal., 18(3):269–302, 2011

  10. [10]

    Inventiones mathematicae, 192(1):55–69, 2013

    Guido De Philippis and Alessio Figalli.W 2,1 regularity for solutions of the monge–amp` ere equation. Inventiones mathematicae, 192(1):55–69, 2013. STABILITY OF OPTIMAL TRANSPORT AND SECOND VARIATION OF THE DISTANCE 31

  11. [11]

    Alessio Figalli, Young-Heon Kim, and Robert J. McCann. H¨ older continuity and injectivity of optimal maps.Arch. Ration. Mech. Anal., 209(3):747–795, 2013

  12. [12]

    On the local geometry of maps with c-convex potentials.Calc

    Nestor Guillen and Jun Kitagawa. On the local geometry of maps with c-convex potentials.Calc. Var. Partial Differential Equations, 52(1-2):345–387, 2015

  13. [13]

    Regularity of monotone transport maps between un- bounded domains.Discrete and Continuous Dynamical Systems, 2019

    Dario Cordero-Erausquin and Alessio Figalli. Regularity of monotone transport maps between un- bounded domains.Discrete and Continuous Dynamical Systems, 2019

  14. [14]

    Global regularity for the monge-ampere equation with natural boundary condition.Annals of Mathematics, 194(3):745–793, 2021

    Shibing Chen, Jiakun Liu, and Xu-Jia Wang. Global regularity for the monge-ampere equation with natural boundary condition.Annals of Mathematics, 194(3):745–793, 2021

  15. [15]

    Springer, 2009

    C´ edric Villani et al.Optimal transport: old and new, volume 338. Springer, 2009

  16. [16]

    Polar factorization and monotone rearrangement of vector-valued functions.Comm

    Yann Brenier. Polar factorization and monotone rearrangement of vector-valued functions.Comm. Pure Appl. Math., 44(4):375–417, 1991

  17. [17]

    Robert J. McCann. A convexity principle for interacting gases.Adv. Math., 128(1):153–179, 1997

  18. [18]

    On h¨ older continuity-in-time of the optimal transport map towards measures along a curve.Proceedings of the Edinburgh Mathematical Society, 54(2):401–409, 2011

    Nicola Gigli. On h¨ older continuity-in-time of the optimal transport map towards measures along a curve.Proceedings of the Edinburgh Mathematical Society, 54(2):401–409, 2011

  19. [19]

    On the regularity of the polar factorization for time dependent maps.Calculus of Variations and Partial Differential Equations, 22(3):343–374, 2005

    Gr´ egoire Loeper. On the regularity of the polar factorization for time dependent maps.Calculus of Variations and Partial Differential Equations, 22(3):343–374, 2005

  20. [20]

    Robert J. Berman. Convergence rates for discretized Monge–Amp` ere equations and quantitative stability of optimal transport.Foundations of Computational Mathematics, 21(4):1099–1140, 2021

  21. [21]

    Quantitative stability of optimal transport maps and linearization of the 2-wasserstein space

    Quentin M´ erigot, Alex Delalande, and Frederic Chazal. Quantitative stability of optimal transport maps and linearization of the 2-wasserstein space. In Silvia Chiappa and Roberto Calandra, ed- itors,Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics, volume 108 ofProceedings of Machine Learning Research, p...

  22. [22]

    Quantitative stability of optimal transport maps under vari- ations of the target measure.Duke Math

    Alex Delalande and Quentin M´ erigot. Quantitative stability of optimal transport maps under vari- ations of the target measure.Duke Math. J., 172(17):3321 – 3357, 2023

  23. [23]

    Gluing methods for quantitative stability of optimal transport maps.arXiv preprint arXiv:2411.04908, 2024

    Cyril Letrouit and Quentin M´ erigot. Gluing methods for quantitative stability of optimal transport maps.arXiv preprint arXiv:2411.04908, 2024

  24. [24]

    Stability of optimal transport maps on

    Jun Kitagawa, Cyril Letrouit, and Quentin M´ erigot. Stability of optimal transport maps on rie- mannian manifolds.arXiv preprint arXiv:2504.05412, 2025

  25. [25]

    Quantitative stability in quadratic optimal transport

    Alex Delalande. Quantitative stability in quadratic optimal transport. PhD thesis, Universit´ e Paris- Saclay, 2022

  26. [26]

    Quantitative stability in discrete optimal transport.arXiv preprint arXiv:2510.17407, 2025

    William Ford. Quantitative stability in discrete optimal transport.arXiv preprint arXiv:2510.17407, 2025

  27. [27]

    A class of sobolev type inequalities.Methods Appl

    Gu-Ji Tian and Xu-Jia Wang. A class of sobolev type inequalities.Methods Appl. Anal., 15(2):263– 276, 2008

  28. [28]

    Regularity for the optimal transportation problem with eu- clidean distance squared cost on the embedded sphere.SIAM Journal on Mathematical Analysis, 44(4):2871–2887, 2012

    Jun Kitagawa and Micah Warren. Regularity for the optimal transportation problem with eu- clidean distance squared cost on the embedded sphere.SIAM Journal on Mathematical Analysis, 44(4):2871–2887, 2012

  29. [29]

    Conditions for existence of single valued optimal trans- port maps on convex boundaries with nontwisted cost.Calc

    Seonghyeon Jeong and Jun Kitagawa. Conditions for existence of single valued optimal trans- port maps on convex boundaries with nontwisted cost.Calc. Var. Partial Differential Equations, 64(4):Paper No. 122, 44, 2025

  30. [30]

    Generalization of an inequality by talagrand and links with the logarithmic sobolev inequality.Journal of Functional Analysis, 2000

    Felix Otto and C´ edric Villani. Generalization of an inequality by talagrand and links with the logarithmic sobolev inequality.Journal of Functional Analysis, 2000

  31. [31]

    Linearization of monge-amp` ere equations and statistical applications, 2024

    Alberto Gonz´ alez-Sanz and Shunan Sheng. Linearization of monge-amp` ere equations and statistical applications, 2024

  32. [32]

    Collins and Freid Tong

    Tristan C. Collins and Freid Tong. Boundary regularity of optimal transport maps on convex do- mains. https://arxiv.org/abs/2507.05395, 2025

  33. [33]

    De La Llave and R

    R. De La Llave and R. Obaya. Regularity of the composition operator in spaces of h¨ older functions. Discrete and Continuous Dynamical Systems, 5, 1999

  34. [34]

    American Mathematical Society, 2003

    C´ edric Villani.Topics in optimal transportation, volume 58. American Mathematical Society, 2003

  35. [35]

    Lectures in Mathematics ETH Z¨ urich

    Luigi Ambrosio, Nicola Gigli, and Giuseppe Savar´ e.Gradient flows in metric spaces and in the space of probability measures. Lectures in Mathematics ETH Z¨ urich. Birkh¨ auser Verlag, Basel, second edition, 2008

  36. [36]

    Springer Cham, 2021

    Luigi Ambrosio, Elia Bru´ e, and Daniele Semola.Lectures on optimal transport. Springer Cham, 2021

  37. [37]

    Springer, 1977

    David Gilbarg, Neil S Trudinger, David Gilbarg, and NS Trudinger.Elliptic partial differential equations of second order, volume 224. Springer, 1977

  38. [38]

    Springer, 2015

    Filippo Santambrogio.Optimal transport for applied mathematicians, volume 55. Springer, 2015

  39. [39]

    Herm Jan Brascamp and Elliott H Lieb. On extensions of the brunn-minkowski and pr´ ekopa-leindler theorems, including inequalities for log concave functions, and with an application to the diffusion equation.Journal of Functional Analysis, 22(4):366–389, 1976

  40. [40]

    On Witten Laplacians and Brascamp-Lieb’s inequality on manifolds with bound- ary.Integral Equations Operator Theory, 87(3):411–434, 2017

    Dorian Le Peutrec. On Witten Laplacians and Brascamp-Lieb’s inequality on manifolds with bound- ary.Integral Equations Operator Theory, 87(3):411–434, 2017

  41. [41]

    Kolesnikov and Emanuel Milman

    Alexander V. Kolesnikov and Emanuel Milman. Brascamp-Lieb-type inequalities on weighted Rie- mannian manifolds with boundary.J. Geom. Anal., 27(2):1680–1702, 2017

  42. [42]

    Interpolations, convexity and geometric inequalities

    Dario Cordero-Erausquin and Bo’az Klartag. Interpolations, convexity and geometric inequalities. InGeometric aspects of functional analysis, volume 2050 ofLecture Notes in Math., pages 151–

  43. [43]

    Springer, Heidelberg, 2012

  44. [44]

    Society for Industrial and Applied Mathematics (SIAM), Philadelphia, PA, 2011

    Pierre Grisvard.Elliptic problems in nonsmooth domains, volume 69 ofClassics in Applied Math- ematics. Society for Industrial and Applied Mathematics (SIAM), Philadelphia, PA, 2011. Reprint of the 1985 original [MR0775683], With a foreword by Susanne C. Brenner

  45. [45]

    World Scientific, 2013

    Gary M Lieberman.Oblique derivative problems for elliptic equations. World Scientific, 2013. 32 F.-U. CAJA-LOPEZ, MATIAS G. DELGADINO, AND JUN KITAGAWA

  46. [46]

    Princeton university press, 1970

    Elias M Stein.Singular integrals and differentiability properties of functions. Princeton university press, 1970. Department of Mathematics, The University of Texas at Austin (USA) Email address:funai.caja@utexas.edu Department of Mathematics, The University of Texas at Austin (USA) Email address:matias.delgadino@math.utexas.edu Department of Mathematics,...