Pith. sign in

REVIEW 4 major objections 5 minor 63 references

A Dacorogna-Moser construction of transport maps on $\mathbb{R}^d$ with application to geodesics on the space of couplings

T0 review · 4 major / 5 minor · reviewed 2026-07-31 · deepseek-v4-flash

Pith's one-line read This paper extends the Dacorogna–Moser transport-map construction to all of R^d for strictly asymptotically log-concave measures, and derives optimality conditions for geodesics in the space of couplings.

desk verdict A genuinely useful R^d Dacorogna–Moser extension with a serious application to geodesics in coupling space, but the optimality proof contains a concrete false identity that must be repaired before the main theorems are accepted. read the letter →

arxiv 2607.23241 v1 pith:UCPCAT7R submitted 2026-07-25 math.AP

classification math.AP MSC 49Q2235B6560J60
keywords transportmapsDacorogna–Moserconstructionasymptoticallylog-concavemeasuresreflectioncouplinggeodesicsinthespaceofcouplingsLagrangemultipliersentropicregularizationSchrödingerbridgeproblem
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 proves that the classical Dacorogna–Moser construction of regular transport maps, previously confined to bounded domains, works on the whole space R^d for strictly asymptotically log-concave measures. The core technical advance is a global, uniform-in-time regularity estimate for the elliptic equation ∆ψ + ∇logµ·∇ψ = φ on R^d, obtained from parabolic semigroup estimates and reflection coupling. This yields an invertible transport map T with T#µ = (1+φ)µ and quantitative W^{n,∞} control on T−Id and T^{-1}−Id. The authors then use this construction to study geodesics in the space of probability measures on a product space with fixed marginals (couplings): they prove existence, derive optimality conditions with Lagrange multipliers, and introduce an entropic regularization whose minimizers converge as the regularization vanishes. Together these results answer an open question about the structure of geodesics in the space of couplings.

What carries the argument

Proposition 2.1, the parabolic–elliptic regularity estimate: the semigroup S_t generated by ½∆µ satisfies ∥∇S_tφ∥_{W^{n,∞}} ≤ c e^{−αt}∥φ∥ for t≥1 and ≤ c/√t ∥φ∥ for t≤1, and the unique solution ψ of ½∆µψ=φ, ∫ψ dµ=0 satisfies ∥∇ψ∥_{W^{n,∞}} ≤ C∥φ∥_{W^{n,∞}}. These estimates are proved by induction on the number of derivatives, using the Feynman–Kac formula, exponential contraction of the reflection-coupling probability, and standard small-time gradient estimates for diffusion semigroups. The strict asymptotic log-concavity of µ enters through the exponential contraction rate β>0 of the coupling.

What would settle it

Numerically compute the solution ψ of ∆µψ=φ on R^d for µ a Gaussian mixture and φ a sequence of compactly supported, mean-zero functions with growing support, and check whether ∥∇ψ∥_{W^{n,∞}} stays bounded by C∥φ∥_{W^{n,∞}} with C independent of the support; alternatively, estimate the reflection-coupling probability p_s(x,y) for large |x−y| and test the exponential decay with a positive rate required in the proof.

Watch

Extended reading notes

Core claim

Under Assumption (A.1) — µ strictly asymptotically log-concave with ∇^j logµ Lipschitz for 1≤j≤n+1 — every mean-zero φ∈W^{n,∞} with φ≥−1+δ admits an invertible transport map T such that T#µ = (1+φ)µ, with T−Id and T^{-1}−Id in W^{n,∞} and ∥T−Id∥_{W^{n,∞}} + ∥T^{-1}−Id∥_{W^{n,∞}} ≤ C∥φ∥_{W^{n,∞}}. The map is generated by the flow of the vector field v_s = −∇ψ/(1+sφ), where ψ solves ∆µψ=φ. The proof of the required global regularity of ψ rests on short- and long-time smoothing bounds for the semigroup of ½∆µ, proved by Feynman–Kac representation and reflection-coupling contraction. The same bounds supply the compactness and competitor arguments for the geodesic problem.

Load-bearing premise

The load-bearing premise is that µ is strictly asymptotically log-concave with enough Lipschitz regularity so that the reflection-coupling contraction rate β in the semigroup estimates is strictly positive; if the drift ½∇logµ fails to be eventually contracting at large scales, the exponential decay in Proposition 2.1 fails and Theorem 1.2's transport map is not obtained.

Editorial extensions

If this is right

  • For any mean-zero φ in W^{n,∞} with φ≥−1+δ, a regular invertible transport map T with T#µ=(1+φ)µ exists on R^d with explicit W^{n,∞} control, extending sampling and functional-inequality tools beyond bounded domains.
  • Geodesics in the space Π(µ,ν) of couplings with fixed marginals exist whenever µ and ν are strictly asymptotically log-concave, a broader class than the previously treated strongly log-concave marginals.
  • The marginal constraints admit a unique Lagrange multiplier in the relevant dual space, making the formal optimality conditions (continuity equation plus momentum equation) rigorous.
  • The entropically regularized geodesic problem has a unique minimizer, Γ-converges to the original problem as ε→0, and its Lagrange multipliers converge weak-* to the unregularized one.

Reading between the lines

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

  • A natural next step is to check whether the one-derivative loss between the bounded-domain and global theorems is intrinsic: if higher-order global uniform estimates are genuinely impossible, fractional regularity is likely the sharp endpoint.
  • The reflection-coupling estimates are probabilistic and depend only on tail contraction of the drift; the same strategy may extend the construction to non-Euclidean settings such as complete Riemannian manifolds with suitable curvature bounds.
  • The rigorous optimality conditions could support a gradient-flow theory in the space of couplings, potentially connecting the geodesic distance to projected Langevin dynamics and offering a quantitative handle on entropic regularization.
  • Numerical computation of ∥T−Id∥_{W^{n,∞}} for Gaussian mixtures (which are strictly asymptotically log-concave) could test the sharpness of the constants in the paper's estimates.
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

4 major / 5 minor

Summary. This paper proves an extension of the Dacorogna–Moser construction to the whole of R^d for strictly asymptotically log-concave reference measures with Lipschitz derivatives of the log-density (Theorem 1.2), based on global W^{n,∞} estimates for the elliptic operator Δ_μ = Δ + ∇log μ · ∇ obtained from parabolic semigroup estimates via reflection coupling (Proposition 2.1). It then applies this construction to the study of geodesics in the space of couplings Π(μ,ν): existence and Γ-convergence for an entropic regularization (Theorem 2.6), existence of Lagrange multipliers for the marginal constraints (Theorem 2.9), optimality conditions and uniqueness of the multipliers (Theorem 2.11), and weak-* convergence of the multipliers (Proposition 2.12). The paper also sketches a path-space interpretation of the regularized problem in Appendix A.

Significance. If the technical gaps are repaired, this would be a substantial contribution. Theorem 1.2 is a natural and nontrivial extension of the classical Dacorogna–Moser construction to unbounded domains, with quantitative global estimates, and the use of reflection-coupling semigroup estimates is well matched to the hypotheses. The application to geodesics in the coupling space addresses an open question from [CLP25], and the optimality/uniqueness results together with Γ-convergence and multiplier convergence provide a coherent variational framework. The paper also gives credit to several recent tools, including [CCE25] and [CL25]. However, several load-bearing computations in the proofs of Proposition 2.1 and Theorem 2.11 are currently invalid or incomplete, so the manuscript cannot be accepted without a substantive revision.

major comments (4)
  1. [3.2, Theorem 2.11, Eq. (48)] The displayed chain (48) is false as written. For μ=N(0,1) and f(x)=x²/2−1/2, one has Δ_μ f = 1−x², ∫ Δ²f dμ = ∫ f'''' dμ = 0, while ∫ ∇log μ · ∇f dμ = ∫ (−x)·x dμ = −1; hence the asserted equality ∫ Δ²f_t dμ = ∫ ∇log μ · ∇f_t dμ is not an identity. There is also a sign error two lines earlier: integration by parts gives ∫ ∇ρ · ∇Δf = −∫ ρ Δ²f, not +. The intended identities (47) are in fact true for ρ_t ∈ Π(μ,ν) because ∫ Δ_μ h dμ = 0 and the first marginal of ρ_t is μ, so the step is likely repairable, but the proof of Theorem 2.11 must be rewritten here.
  2. [3.2, Theorem 2.11, Eq. (45)] The expansion of A₀ uses pairings defined as ⟨∂t(ρv),(∇f,0)⟩ = ∫(∇f_t,0)·v_t dρ_t dt and ⟨div(ρv⊗v),(∇f,0)⟩ = ∫(∇²f_t v_t)·v_t dρ_t dt. This is not the first variation of A₀ along the perturbed curve ρ^δ_t = ((T^δ_t)^{-1},Id)_#ρ_t. The correct expansion contains −∫ ρ v·∂_t∇f dt and −∫ ρ (∇²f v)·v dt; the printed definitions have the wrong sign/content, and the subsequent 'substituting φ by −φ' does not repair the computation. This is a load-bearing step in the derivation of the optimality conditions.
  3. [3.1, Proposition 2.1, parabolic induction around Eq. (26)] The proof claims ∥∂_I v_t∥∞ ≤ ∥∂_I φ∥∞ for the nonhomogeneous parabolic equation (26), whose right-hand side contains the source term −Σ ∂_J b · ∇∂_{I\J} v. For such an equation the maximum principle gives an additional integral of the source; the displayed bound is false in general. Although the subsequent stochastic estimates may not require this exact bound, it is asserted as a justification for applying Itô's formula. Moreover, the integrals in the source term are bounded using the coupling probability p_s(x,y)=P(X_s≠Y_s), but the integrands involve Lipschitz functions of the processes; controlling these differences requires an estimate on E|X_s−Y_s| (or an equivalent coupling contraction bound), not merely on P(X_s≠Y_s). The manuscript should state and prove the needed coupling contraction estimate from [EZ19].
  4. [3.1, Proposition 2.1, elliptic part] The line d/dt ∫ φ_t dμ = (1/2)∫ ∇·(μ φ_t) dx is wrong in two ways: it should involve μ∇φ_t rather than μφ_t, and the vanishing of the integral of a divergence over R^d requires spatial decay/boundary conditions that are not established. The conclusion ∫φ_t dμ = 0 is true and follows immediately from the fact that S_t is the symmetric Markov semigroup of ½Δ_μ in L²(μ); the proof should be rewritten accordingly. Without this step the pointwise definition of ψ = −∫₀^∞ φ_t dt and the mean-zero condition are not justified by the text.
minor comments (5)
  1. [3.1, elliptic part] The notation 'φ_t(x)∈L^1(dt⊗dμ(x))' is imprecise; it should be stated as a joint integrability condition for the function (t,x)↦φ_t(x).
  2. [Appendix A, Proposition A.1] Labeling an 'Informal' statement as a Proposition may confuse readers; consider calling it a Remark or stating precise hypotheses.
  3. [References] The reference [MACJC17] has a garbled author list; it should be M. Arnaudon, A.B. Cruzeiro, C. Léonard, and J.-C. Zambrini.
  4. [3.2, Theorem 2.6] The construction of the velocity field v_t by pushforward under Lipschitz maps T,S should spell out the weak/a.e. differentiability interpretation if T,S are only Lipschitz; this is used later in the continuity equation.
  5. [3.2, Theorem 2.11] Several steps are described only as 'straightforward adaptation of [Bar20]' without checking the endpoint or compactness conditions specific to this setting. Please make these adaptations explicit, especially for the expansions leading to (45)–(49).

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the main construction is proved from PDE/semigroup estimates and external coupling results; self-citations are auxiliary, not assumed conclusions.

full rationale

Walking the derivation chain, I find no circular step. Theorem 1.2 is proved by reducing the transport equation to the elliptic problem Δ_μ ψ = φ; Proposition 2.1 solves this by the semigroup representation ψ = −∫_0^∞ S_t φ dt and obtains the required W^{n,∞} gradient bounds from the external coupling estimates [PW06, Theorem 3.4] and [EZ19, Theorem 1] plus an induction on derivatives. The only self-citation in this part, [CCE25, Theorem 1.4], is used solely to get a Poincaré inequality from existence of a Lipschitz Gaussian→μ map; that theorem's assumptions do not include Theorem 1.2, so it is independent support rather than a circular premise. Similarly, Theorem 2.6 derives existence of geodesics for strictly asymptotically log-concave marginals by pulling the problem back to Gaussian marginals via [CCE25] and invoking [CL25] for the Gaussian/strongly log-concave special case; this is a reduction to a special case proved elsewhere, not to the theorem being claimed. The Lagrange-multiplier results use Theorem 1.2 as a competitor-building device and adapt [Bar20]; the optimality conditions are derived from first-order expansions, not from the desired conclusion. There are no fitted parameters renamed as predictions, no uniqueness theorem imported from the same authors to force the conclusion, and no ansatz smuggled in by citation. One non-circular defect should be flagged: in the proof of Theorem 2.11, the displayed chain in Eq. (48) contains an invalid integration-by-parts identity as written (it asserts ∫ Δ²f dμ = ∫ ∇log μ·∇f dμ, which is false for the standard Gaussian and f(x)=x²/2−1/2, and the preceding line has a sign error); the skeptic's own note says the cancellation may be salvageable with corrected signs. Because this is a proof error rather than a reduction of the target result to its own inputs, it does not raise the circularity score.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

There are no fitted parameters and no new ontological entities. All constants are structural functions of the semiconvexity profile, Lipschitz constants, n, M, and δ. The main hidden inputs are imported theorems listed above, several of which come from papers with overlapping authorship.

assumptions (7)
  • domain assumption Strict asymptotic log-concavity of µ and ν with Lipschitz derivatives of log densities up to order n+1 (Assumption (A.1)).
    This is the main hypothesis of Theorem 1.2 and of all the geodesic results.
  • domain assumption Perturbation φ satisfies ∫φdµ=0, ∥φ∥_{W^{n,∞}}≤M, and φ≥−1+δ.
    The transported density (1+φ)µ must be nonnegative and preserve normalization; the theorem is stated for this class.
  • standard math External reflection-coupling estimates [EZ19, Theorem 1] and [PW06, Theorem 3.4] for the semigroup of ½∆_µ.
    Used in the base case and induction step of the parabolic estimates in Proposition 2.1; if these fail, the gradient bounds collapse.
  • standard math Existence of Lipschitz transport maps from the standard Gaussian to strictly asymptotically log-concave measures [CCE25, Theorem 1.4].
    Used to obtain the Poincaré inequality for µ and finite-length curves in Theorem 2.6; this is an overlapping-author prior result.
  • standard math Finite-length curves and compactness for couplings with Gaussian marginals [CL25, Section 3].
    Used in Theorem 2.6 to transfer curves from Gaussian marginals to asymptotically log-concave marginals.
  • standard math Estimates for the dynamic Schrödinger problem in metric spaces [MTV23, Theorem 3.12] and the EVI gradient-flow framework.
    Used to prove finiteness and Γ-convergence of the regularized problem.
  • standard math Standard PDE and SDE existence theory plus Feynman–Kac representation [Fri08, Fri75], and Lax–Milgram for H^1(µ).
    Basis for the semigroup construction and elliptic solvability.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Dacorogna-Moser construction of transport maps on $\mathbb{R}^d$ with application to geodesics on the space of couplings." pith.science (2026). https://pith.science/paper/UCPCAT7R

@misc{pith2026260723241,
  author       = {Pith},
  title        = {Pith review of: A Dacorogna-Moser construction of transport maps on $\mathbbR^d$ with application to geodesics on the space of couplings},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UCPCAT7R}},
  note         = {Machine review of arXiv:2607.23241}
}
abstract

A seminal work by Dacorogna and Moser introduced a way of constructing regular transport maps from a probability distribution on a bounded domain to another one. In this work, we extend this construction to the whole $\mathbb{R}^d$ for strictly asymptotically log-concave measures, a wide class of distributions that encompasses Lipschitz-perturbations of log-concave measures. We then leverage this construction to study geodesics in the space of probability measures on a product set with imposed marginal laws (couplings), for which we derive optimality conditions, answering an open question in a recent work by Conforti, Lacker and Pal. Taking inspiration from Brenier's variational model for incompressible fluids and its regularization, we further introduce an entropic regularization of the geodesic problem, which can be seen as the Schr\"odinger bridge problem on the space of couplings, for which we also derive optimality conditions. We eventually study convergence of minimizers as the regularization vanishes, and we prove convergence of the related Lagrange multipliers. Our approach involves proving uniform-in-time global regularity estimates on elliptic and parabolic equations on $\mathbb{R}^d$, by exploiting the structure of asymptotically log-concave measures using the probabilistic notion of reflection coupling.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

63 extracted references · 4 linked inside Pith

  1. [1]

    Journal of Functional Analysis , volume=

    Gradient estimates for diffusion semigroups with singular coefficients , author=. Journal of Functional Analysis , volume=. 2006 , publisher=

  2. [2]

    Optimal Transport: Theory and Applications , publisher=

    Ambrosio, Luigi and Figalli, Alessio , title=. Optimal Transport: Theory and Applications , publisher=. 2014 , pages=

  3. [3]

    A survey of the

    L. A survey of the. Discrete and Continuous Dynamical Systems , volume=. 2013 , publisher=

  4. [4]

    Girsanov theory under a finite entropy condition , author=. S. 2012 , publisher=

  5. [5]

    2024 , journal =

    López-Rivera, Pablo , title =. 2024 , journal =

  6. [6]

    Variational models for the incompressible

    Daneri, Sara and Figalli, Alessio , journal=. Variational models for the incompressible

  7. [7]

    Examples of hidden convexity in nonlinear PDEs , author=

  8. [8]

    Journal of the American Mathematical Society , volume=

    The least action principle and the related concept of generalized flows for incompressible perfect fluids , author=. Journal of the American Mathematical Society , volume=

Show all 63 references
  1. [9]

    Communications in Partial Differential Equations , volume =

    The geometry of dissipative evolution equations: the porous medium equation , author=. Communications in Partial Differential Equations , volume =. 2001 , publisher=

  2. [10]

    NeurIPS , year =

    Ho, Jonathan and Jain, Ajay and Abbeel, Pieter , title =. NeurIPS , year =

  3. [11]

    Book draft available at https://chewisinho

    Log-concave sampling , author=. Book draft available at https://chewisinho. github. io , volume=

  4. [12]

    and Kumar, Abhishek and Ermon, Stefano and Poole, Ben , title =

    Song, Yang and Sohl-Dickstein, Jascha and Kingma, Diederik P. and Kumar, Abhishek and Ermon, Stefano and Poole, Ben , title =. ICLR , year =

  5. [13]

    Probability Theory and Related Fields , volume=

    Eberle, Andreas , title =. Probability Theory and Related Fields , volume=

  6. [14]

    Annales de l'Institut Henry-Poincaré - Probabilité et Statistiques , volume =

    Eberle, Andreas and Zimmer, Raphael , title =. Annales de l'Institut Henry-Poincaré - Probabilité et Statistiques , volume =

  7. [15]

    arXiv: 2508.07931 , year =

    Chaintron, Louis-Pierre and Conforti, Giovanni and Eichinger, Katharina , title =. arXiv: 2508.07931 , year =

  8. [16]

    The Annals of Applied Probability , volume=

    Conforti, Giovanni , title =. The Annals of Applied Probability , volume=

  9. [17]

    Generalized incompressible flows, multi-marginal transport and

    Benamou, Jean-David and Carlier, Guillaume and Nenna, Luca , journal=. Generalized incompressible flows, multi-marginal transport and. 2019 , publisher=

  10. [18]

    Projected

    Conforti, Giovanni and Lacker, Daniel and Pal, Soumik , journal =. Projected

  11. [19]

    Geodesic convexity and strengthened functional inequalities in submanifolds of

    Chaintron, Louis-Pierre and Lacker, Daniel , journal =. Geodesic convexity and strengthened functional inequalities in submanifolds of

  12. [20]

    Small Noise Limit and Convexity for Generalized Incompressible Flows,

    Baradat, Aymeric and Monsaingeon, Léonard , journal =. Small Noise Limit and Convexity for Generalized Incompressible Flows,. 2020 , volume =

  13. [21]

    The dynamical

    Léonard Monsaingeon and Luca Tamanini and Dmitry Vorotnikov , journal =. The dynamical. 2023 , volume =

  14. [22]

    A computational fluid mechanics solution to the

    Benamou, Jean-David and Brenier, Yann , journal =. A computational fluid mechanics solution to the. 2000 , pages =

  15. [23]

    I: Structural properties , author =

    Gradient flows and Evolution Variational Inequalities in metric spaces. I: Structural properties , author =. Journal of Functional Analysis , year =

  16. [24]

    2008 , publisher =

    Gradient Flows in Metric Spaces and in the Space of Probability Measures , author =. 2008 , publisher =

  17. [25]

    2002 , publisher=

    Gamma-convergence for Beginners , author =. 2002 , publisher=

  18. [26]

    On the Existence of a Scalar Pressure Field in the

    Baradat, Aymeric , year =. On the Existence of a Scalar Pressure Field in the. SIAM Journal on Mathematical Analysis , volume =

  19. [27]

    2019 , journal =

    Continuous dependence of the pressure field with respect to endpoints for ideal incompressible fluids , author =. 2019 , journal =

  20. [28]

    1970 , publisher =

    Convex Analysis , author =. 1970 , publisher =

  21. [29]

    On a partial differential equation involving the

    Dacorogna, Bernard and Moser, Jürgen , year =. On a partial differential equation involving the. Annales de l'I.H.P. Analyse non linéaire , volume=

  22. [30]

    On the regularity of the pressure field of

    Figalli, Alessio and Ambrosio, Luigi , journal =. On the regularity of the pressure field of

  23. [31]

    Minimal geodesics on groups of volume-preserving maps and generalized solutions of the

    Yann Brenier , journal=. Minimal geodesics on groups of volume-preserving maps and generalized solutions of the. 1999 , volume=

  24. [32]

    Calculus of Variations and Partial Differential Equations , volume=

    Remarks on the minimizing geodesic problem in inviscid incompressible fluid mechanics , author =. Calculus of Variations and Partial Differential Equations , volume=

  25. [33]

    Annales de l'Institut Henri Poincaré, Probabilités et Statistiques , year =

    An entropic interpolation problem for incompressible viscous fluids , author =. Annales de l'Institut Henri Poincaré, Probabilités et Statistiques , year =

  26. [34]

    Sur la g\'eom\'etrie diff\'erentielle des groupes de

    Arnold, Vladimir , journal =. Sur la g\'eom\'etrie diff\'erentielle des groupes de. 1966 , publisher =

  27. [35]

    ESAIM, Proceedings and Surveys , year =

    Santambrogio, Filippo , title =. ESAIM, Proceedings and Surveys , year =

  28. [36]

    Lipschitz regularity in Flow Matching and Diffusion Models: sharp sampling rates and functional inequalities , journal =

    St. Lipschitz regularity in Flow Matching and Diffusion Models: sharp sampling rates and functional inequalities , journal =

  29. [37]

    Nonlinear Differential Equations and Applications NoDEA , year =

    Rivière, Tristan and Ye, Dong , title =. Nonlinear Differential Equations and Applications NoDEA , year =

  30. [38]

    and Gangbo, W

    Evans, L.C. and Gangbo, W. , title =. 1999 , publisher=

  31. [39]

    , title =

    McMullen, Curtis T. , title =. Geometric and Functional Analysis , year =

  32. [40]

    arXiv: 2605.06172 , year =

    Iske, Meira and Sch\"onlieb Carola-Bibiane , title =. arXiv: 2605.06172 , year =

  33. [41]

    Calculus of Variations and Partial Differential Equations , year =

    Fathi, Max and Mikulincer, Dan and Shenfeld, Yair , title =. Calculus of Variations and Partial Differential Equations , year =

  34. [42]

    Mathematische Annalen , volume =

    Kim, Young-Heon and Milman, Emanuel , title =. Mathematische Annalen , volume =

  35. [43]

    Geometric Aspects of Functional Analysis , year =

    Dan Mikulincer and Yair Shenfeld , title =. Geometric Aspects of Functional Analysis , year =. doi:10.1007/978-3-031-26300-2_9 , url =

  36. [44]

    Time reversal on

    F. Time reversal on. Stochastic Processes—Mathematics and Physics: Proceedings of the 1st BiBoS-Symposium held in Bielefeld , pages=

  37. [45]

    Smooth transport map via diffusion process , journal =

    St. Smooth transport map via diffusion process , journal =

  38. [46]

    Journal of the American Mathematical Society , year =

    Andrews, Ben and Clutterbuck, Julie , title =. Journal of the American Mathematical Society , year =

  39. [47]

    , title =

    Lindvall, Torgny and Rogers, L.C.G. , title =. The Annals of Probability , volume=

  40. [48]

    , title =

    Caffarelli, L.A. , title =. Communications in Mathematical Physics , volume=

  41. [49]

    and Ben-Hamu, Heli and Nickel, Maximilian and Le, Matt , title =

    Lipman, Yaron and Chen, Ricky T.Q. and Ben-Hamu, Heli and Nickel, Maximilian and Le, Matt , title =. International Conference on Learning Representations (ICLR) , year =

  42. [50]

    Michael and Vanden-Eijnden, Eric , title =

    Albergo, S. Michael and Vanden-Eijnden, Eric , title =. International Conference on Learning Representations (ICLR) , year =

  43. [51]

    Michael and Boffi, Nicholas M

    Albergo, S. Michael and Boffi, Nicholas M. and Vanden-Eijnden, Eric , title =. Journal of Machine Learning Research , volume=

  44. [52]

    Handbook of Uncertainty Quantification , year=

    Marzouk, Youssef and Moselhy, Tarek and Parno, Matthew and Spantini, Alessio , title =. Handbook of Uncertainty Quantification , year=

  45. [53]

    , title =

    Chewi, Sinho and Pooladian, Alexandre-Aram and Zhang, Matthew S. , title =. arXiv: 2511.01154 , year =

  46. [54]

    arXiv: 2606.23383 , year =

    Chewi, Sinho and Eichinger, Katharina and Pooladian, Alexandre-Aram , title =. arXiv: 2606.23383 , year =

  47. [55]

    Shnirel'man, A. I. , title =. Mathematics of the USSR-Sbornik , volume =

  48. [56]

    Shnirel'man, A. I. , title =. Geometric and Functional Analysis GAFA , volume=

  49. [57]

    Nonlinear Analysis , volume=

    Ambrosio, Luigi and Erbar, Matthias and Savaré, Giuseppe , title =. Nonlinear Analysis , volume=

  50. [58]

    Lecture notes on gradient flows and optimal transport , booktitle=

    Daneri, Sara and Savaré, Giuseppe , series=. Lecture notes on gradient flows and optimal transport , booktitle=. 2014 , pages=

  51. [59]

    Transactions of the American Mathematical Society , volume=

    Kac, Mark , title =. Transactions of the American Mathematical Society , volume=

  52. [60]

    Friedman, Avner , title =

  53. [61]

    Stochastic differential equations , publisher =

    Friedman, Avner , title =. Stochastic differential equations , publisher =

  54. [62]

    Inventiones mathematicae , volume=

    Ambrosio, Luigi and Gigli, Nicola and Savaré, Giuseppe , title =. Inventiones mathematicae , volume=

  55. [63]

    and Pavon, Michele , title =

    Chen, Yongxin and Georgiou, Tryphon T. and Pavon, Michele , title =. Journal of Optimization Theory and Applications , volume=

Pith tools

Reviewed July 31, 2026 · model on record in the stance chip above.