Convergence of drift-diffusion PDEs arising as Wasserstein gradient flows of convex functions

L\'ena\"ic Chizat; Maria Colombo; Xavier Fern\'andez-Real

REVIEW 1 major objections 5 minor 1 cited by

Convergence of drift-diffusion PDEs arising as Wasserstein gradient flows of convex functions

T0 review · 1 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper proves that merely convex objectives, not just displacement-convex ones, yield quantitative convergence for Wasserstein gradient flows of drift-diffusion PDEs.

desk verdict Quantitative convergence for Wasserstein gradient flows at critical diffusion is genuinely new and mostly correct; the main caveat is that the theorems assume a WGF exists, which is only guaranteed under a stronger regularity assumption. read the letter →

arxiv 2507.12385 v1 pith:KV7GBAQ3 submitted 2025-07-16 math.OC math.AP

Lénaïc Chizat , Maria Colombo , Xavier Fernández-Real This is my paper · ORCID

classification math.OCmath.AP MSC 49Q2235Q8460J6090C25

keywords Wassersteingradientflowsmean-fieldLangevindynamicsdrift-diffusionPDEslinearconvexityentropyregularizationcriticaldiffusivityLojasiewiczinequalityGaussiandensityestimates

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

This paper establishes quantitative convergence guarantees for drift-diffusion PDEs that arise as Wasserstein gradient flows of objectives $F = G + \tau H$, where $H$ is the negative entropy and $G$ need only be linearly convex—not displacement convex, i.e., not convex along Wasserstein geodesics. It proves that whenever $F$ is convex, the suboptimality gap $F(\mu_t) - \inf F$ decays at least like $O(1/t)$ on $\mathbb{R}^d$, and that strict convexity relative to entropy ($\tau > \tau_c$) accelerates the decay to superpolynomial on $\mathbb{R}^d$ and exponential on the torus, with a rate linear in $\tau - \tau_c$. Because displacement convexity is rarely available in data-science applications, the result widens the class of objective functions for which global convergence is guaranteed, covering entropy-regularized nonconvex problems and a new approximate Fisher Information regularizer. The mechanism is diffusion itself: entropy produces uniform two-sided Gaussian (or uniform, on the torus) density bounds that turn linear convexity into a usable Lojasiewicz inequality—a bound linking the suboptimality gap to the gradient—in Wasserstein geometry.

What carries the argument

The load-bearing object is a two-sided Gaussian estimate for solutions of advection-diffusion equations with bounded drift: for any $\varepsilon > 0$ the density $\mu_t$ of the flow satisfies $c_{\varepsilon} e^{-(1+\varepsilon)|x|^2} \le \mu_t(x) \le c_{\varepsilon}^{-1} e^{-(1-\varepsilon)|x|^2}$ after an explicit waiting time, with constants computable from the drift bound and the subgaussian constant of the initial law. These estimates are proved from explicit two-sided bounds on Fokker–Planck transition kernels (Propositions 2.3 and 4.3), and they play the central role: with $\mu_t$ squeezed between Gaussians $\gamma_1$ and $\gamma_2$, the objective $F$ becomes strongly convex relative to the $L^2(\gamma_1)$ norm over the relevant densities (Lemma 2.1), and the Poincaré inequality for $\gamma_1$ converts $L^2$ gradients into Wasserstein gradients, yielding the differential inequality $h' \le -C h^{1+1/\kappa}$ that integrates to the stated rates. On the torus, uniform density bounds $m \le \mu_t \le M$ play the same role.

What would settle it

Simulate or analytically solve the Wasserstein gradient flow for a convex $F = G + \tau H$ with $\tau = \tau_c$ on $\mathbb{R}$ for a linearly convex but not displacement-convex $G$ satisfying Assumption A, starting from an $M_0$-subgaussian measure; if the suboptimality gap decays slower than $1/t$, or the density violates the two-sided Gaussian bounds stated in Proposition 4.1 at the given waiting time, the central claim fails.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the critical diffusivity $\tau_c = \inf\{\tau' \ge 0 : G + \tau' H \text{ is convex}\}$ is a sharp threshold. If $\tau \ge \tau_c > 0$, mere convexity of $F$ suffices for the Wasserstein gradient flow to converge with the Euclidean-style convex rate $O(1/t)$; if $\tau > \tau_c$, convergence is faster than any polynomial on $\mathbb{R}^d$ and exponential on $\mathbb{T}^d$ with a rate proportional to $\tau - \tau_c$. The paper identifies the previously missing mechanism: the diffusive entropy term forces the evolving density to be sandwiched between two Gaussians (or bounded above and below on the torus), and those bounds allow convexity of $F$ to be read as a Polyak–Lojasiewicz inequality in $L^2$ geometry, which then transfers to the Wasserstein metric.

Load-bearing premise

The results assume the drift-diffusion equation has a global solution whose energy decreases exactly at the dissipation rate stated in equation (5); the paper only proves such existence under a stronger regularity condition (Assumption A′) than the main theorems assume.

Editorial extensions

If this is right

Any G satisfying Assumption A′ on a compact domain can be made convex by adding enough entropy, so Wasserstein gradient flows of entropy-regularized nonconvex objectives converge with explicit rates.
The approximate Fisher Information regularizer Dτ(μ)=Tτ(μ,μ) is covered: Dτ+τH is convex, the flow converges at rate O(1/t), and regularized minimizers achieve O(τ²) approximation error when a minimizer of the unregularized objective has finite Fisher information.
The path-space trajectory-inference estimator (minimizing relative entropy with respect to the Wiener measure) fits the framework, so its Wasserstein gradient flow converges exponentially without extra regularization.
On the torus, the strong-convexity rate is linear in τ−τc, an exponential improvement over rates obtained by applying uniform log-Sobolev inequalities to G+τcH.

Reading between the lines

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

The same density-estimate strategy would likely transfer to compact manifolds or bounded domains with Neumann boundary conditions whenever explicit heat-kernel bounds are available; the paper notes this transfer but does not compute constants there.
If the Gaussian sandwich could be made with the two variances arbitrarily close, the proof would upgrade from superpolynomial to exponential in the noncompact case; the paper states that such a uniform Poincaré property appears out of reach of its method.
A numerical check of the explicit waiting time and rate constants in Theorem 1.3 on a simple non-displacement-convex example would reveal whether the stated constants are tight or only sufficient.
For diffusivity below the critical threshold the framework gives no guarantee, and the paper's own examples of stable non-optimal stationary points suggest some diffusion threshold is genuinely needed; identifying the sharpest such counterexample would delimit the theory.

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.

Desk Editor's Note

read the letter

Two things to know. The paper proves what it claims: for F = G + τH with F convex (not displacement convex), the suboptimality gap along the WGF decays at rate O(1/t) at the critical diffusion τ = τc, and at rate O(t^{−κ}) for any κ > 1 when τ > τc, with explicit constants. On the torus the rate is exponential and linear in τ − τc. That genuinely extends earlier mean-field Langevin results, which required convex G and a uniform log-Sobolev inequality. The proof strategy is clean: explicit two-sided Gaussian density estimates from transition kernel bounds, then PL-type inequalities in the Wasserstein geometry via Gaussian Poincaré. The applications—entropic regularization of nonconvex interactions, an approximate Fisher information regularizer, and path-space trajectory inference—are nontrivial and check out.

The soft spots. The main theorems are stated under Assumption A but take “let μt be a WGF (4)” as an assumption. Global existence of such a flow with the energy dissipation identity (5) is only guaranteed under the stronger Assumption A′. The paper acknowledges this in Section 1.1 but does not fold it into the theorem statements, so the advertised “if the objective is convex and suitably coercive” overreaches. This is a real caveat but an addressable one; the conditional statements remain valuable. The abstract’s “faster than any polynomial” is actually accurate, because the O(t^{−κ}) bound holds for every κ > 1, so for any polynomial degree k one can choose κ > k. The minor wrinkle in Lemma 4.2’s upper-bound exponent (it comes out as 1−10ε) is not a problem after Proposition 4.1’s proof rescales ε.

Overall, the central argument holds up. The proofs are long but careful, and the hardest estimates—explicit transition kernel bounds—are real work that checks out. This paper deserves a serious referee. I would cite it if I worked on mean-field Langevin dynamics or sampling, and I would bring it to our reading group.

Referee Report

1 major / 5 minor

Summary. The paper studies Wasserstein gradient flows of linearly convex objectives of the form F = G + τH, where G need not be displacement convex. The main theorems (Theorem 1.3 on R^d and Theorem 1.5 on the torus) establish quantitative convergence of the suboptimality gap for any global Wasserstein gradient flow satisfying the energy-dissipation identity (5): an O(1/t) rate under mere convexity (τ ≥ τc), a superpolynomial rate with exponent κ > 1 when τ > τc, and an exponential rate on the torus with a rate linear in τ - τc. The proofs combine explicit two-sided Gaussian density estimates for the drift-diffusion equation (Propositions 2.3, 4.1, 4.3) with Łojasiewicz-type inequalities in the L^2 geometry and Poincaré inequalities to transfer to the Wasserstein geometry. Applications are given to entropic regularization of nonconvex interaction energies, an approximate Fisher information regularization, and a trajectory inference estimator in path space (Section 5).

Significance. If the results hold, they substantially extend the quantitative convergence theory of mean-field Langevin dynamics beyond displacement convexity, covering objectives that are merely linearly convex and including nonconvex interaction energies. The explicit transition-kernel estimates with computable constants (Section 4) are a valuable technical contribution. The applications in Section 5, especially the approximate Fisher information regularizer (Proposition 5.7) and the path-space trajectory inference estimator (Propositions 5.9–5.11), demonstrate genuine new use cases. The proofs are detailed and the main derivation steps are coherent; no fitted parameters or circular definitions appear, and τc is defined by a convexity threshold rather than chosen to force the conclusion.

major comments (1)

[Section 1.1, Theorems 1.3 and 1.5, Abstract] The theorems assume the existence of a global Wasserstein gradient flow satisfying the energy-dissipation identity (5), but the paper only guarantees such existence under the stronger Assumption A′ (stated after Assumption A), not under Assumption A alone. As a result, the abstract's claim that convergence holds 'if the objective is convex and suitably coercive' is broader than what is proved: the results are conditional on an additional existence hypothesis that is not highlighted. Please either state the existence hypothesis explicitly in the abstract and theorems, prove existence under Assumption A, or restrict the main statements to Assumption A′ and mention that Assumption A suffices for the convergence mechanism once a flow exists. This is load-bearing for the advertised scope of the paper.

minor comments (5)

[Proof of Theorem 1.3, convex case] The proof says 'take ε = 1/4 in Proposition 4.1', but Proposition 4.1 requires ε ∈ (0, 1/4), and the proof of Lemma 4.2 only yields an upper-bound exponent of the form 1−10ε before redefining ε. The argument is repairable by choosing an ε < 1/4 (for instance ε = 1/8), which still satisfies the condition σ1²/σ2² > 1/2 needed for Lemma 3.3.
[Proposition 5.7(1)] The text says 'the convergence guarantee of Theorem 1.3 applies' in the setting Ω = T^d; the correct reference is Theorem 1.5 (point 2 for the critical case).
[Lemma 2.1] The statement says λ ∈ L1_+(Ω) with H(λ) < +∞, but the proof uses H(µ) = H(µ|λ) + ∫ log(λ) dµ and identifies λ with a probability density; please clarify that λ is a probability density (or normalize it) in the lemma statement.
[Section 1.3 and References] The citation keys [CR W22] and [CRL VP20] contain stray spaces in the text (e.g., 'CR W22'); this should be fixed for consistency with the reference list.
[Lemma 4.2] The proof obtains an upper bound of the form e−(1−10ε)|z|^2 and then says 'up to redefining ε'; for a reader, it would be clearer to state explicitly in the lemma that the constant and the exponent are after such a redefinition, so that the admissible range of ε is transparent.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the convergence theorems are conditional on an assumed WGF and derive their rates from convexity, Assumption A, and explicit two-sided density estimates, with no fitted parameter renamed as a prediction.

full rationale

The derivation chain is self-contained in the relevant sense. The main theorems are explicitly conditional on the existence of a Wasserstein gradient flow satisfying the energy dissipation identity (5): Theorem 1.3 states 'let µt be a WGF (4) of F' and Theorem 1.5 similarly assumes 'let µt be a WGF (4) starting from µ0'. The convergence rates are then proved from the assumed convexity of F, Assumption A, the definition of τc, and the paper's own two-sided density estimates (Proposition 4.1, Lemmas 2.2, 3.1, and 3.3). The critical diffusivity τc = inf{τ′ ≥ 0 : G + τ′H is convex} is a property of the input G and H, not a parameter fitted to the convergence conclusion; the theorem's dependence on τ − τc is a derived consequence, not an assumed rate. No quantity appearing in the bounds is fitted to a subset of data and then reported as a prediction. The paper does cite prior work by the authors, including [Chi22], [CZHS22], [CCL24], and [CT21], but these are used as external mathematical tools: regularity of displacement semiconvexity, entropic optimal transport identities, and EOT potential smoothness. These citations do not by themselves contain the O(1/t), superpolynomial, or exponential rates that the paper proves, so they are not load-bearing in a circular way. The manuscript also transparently notes a hypothesis-coverage limitation: 'The global-in-time existence of the WGF is satisfied in general under fairly general assumptions, but slightly more restricitive than Assumption A in terms of regularity: for instance, Assumption A′ below directly ensures that the functional is displacement semiconvex and imply the existence of a WGF.' This means the theorems, stated under Assumption A with the WGF taken as given, may have narrower scope than the abstract's informal phrasing suggests; that is a possible correctness or scope gap, but it is not circularity, because the theorems do not claim to prove that existence. Similarly, the reader's note that Lemma 4.2's upper-bound proof yields an exponent around 1−10ε rather than the 1−ε eventually used is a technical repairability detail, not a reduction of the conclusion to the hypotheses. No circular step satisfying the required standard can be quoted and exhibited, so the appropriate verdict is no significant circularity.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No free parameters or invented entities. All constants are explicit and depend only on problem inputs. The most fragile item is the global existence of the WGF, which is assumed rather than proven under Assumption A.

assumptions (5)

domain assumption G admits a C^1 first variation with gradient of the form alpha x + V[mu] on R^d, with |grad V| <= L (Assumption A).
Core hypothesis of Theorems 1.3 and 1.5; used to bound the drift and to prove the Gaussian density estimates.
domain assumption The WGF (mu_t) exists globally and satisfies the energy dissipation identity (5).
Assumed in Definition 1.1; the paper notes after Assumption A that existence is guaranteed under the stronger Assumption A'.
domain assumption The initial measure mu0 is M0-subgaussian.
Required for the finite-time Gaussian density bounds in Proposition 4.1.
standard math Poincare inequality holds on the domain (the torus, or Gaussian measures on R^d).
Used in Lemmas 2.2, 3.1, 3.3 to convert L2 gradient estimates into Wasserstein PL inequalities.
standard math Girsanov theorem and Brownian-bridge estimates for Fokker-Planck transition kernels.
Used in Proposition 2.3 to derive explicit two-sided transition kernel bounds.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Convergence of drift-diffusion PDEs arising as Wasserstein gradient flows of convex functions." pith.science (2026). https://pith.science/paper/KV7GBAQ3

@misc{pith2026250712385,
  author       = {Pith},
  title        = {Pith review of: Convergence of drift-diffusion PDEs arising as Wasserstein gradient flows of convex functions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KV7GBAQ3}},
  note         = {Machine review of arXiv:2507.12385}
}

abstract

We study the quantitative convergence of drift-diffusion PDEs that arise as Wasserstein gradient flows of linearly convex functions over the space of probability measures on ${\mathbb R}^d$. In this setting, the objective is in general not displacement convex, so it is not clear a priori whether global convergence even holds. Still, our analysis reveals that diffusion {allows} a favorable interaction between Wasserstein geometry and linear convexity, leading to a general quantitative convergence theory, analogous to that of gradient flows in convex settings in the Euclidean space. Specifically, we prove that if the objective is convex and suitably coercive, the suboptimality gap decreases at a rate $O(1/t)$. This improves to a rate faster than any polynomial -- or even exponential in compact settings -- when the objective is strongly convex relative to the entropy. Our results extend the range of mean-field Langevin dynamics that enjoy quantitative convergence guarantees, and enable new applications to optimization over the space of probability measures. To illustrate this, we show quantitative convergence results for the minimization of entropy-regularized nonconvex problems, we propose and study an \emph{approximate Fisher Information} regularization covered by our setting, and we apply our results to an estimator for trajectory inference which involves the minimization of the relative entropy with respect to the Wiener measure in path space.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Convergence Rates for Distribution Matching with Sliced Optimal Transport
stat.ML 2026-02 conditional novelty 7.0 of 10

For Gaussian distributions, slice-matching to an isotropic target with decaying step sizes converges at rate O(k^{-(2α-1)}) in expectation.

Reference graph

Works this paper leans on

13 extracted references · 9 canonical work pages · cited by 1 Pith paper

[3]

Global well-posedness of master equa- tions for deterministic displacement convex potential mean field games

2017, pp. 569–600. [GM22] Wilfrid Gangbo and Alp´ ar R M´ esz´ aros. “Global well-posedness of master equa- tions for deterministic displacement convex potential mean field games”. In: Communications on Pure and Applied Mathematics 75.12 (2022), pp. 2685–

work page 2022
[4]

Weighted ultrafast diffusion equations: from well-posedness to long-time behaviour

Institut Henri Poincar´ e. 2021, pp. 2043–2065. [IPS19] Mikaela Iacobelli, Francesco S Patacchini, and Filippo Santambrogio. “Weighted ultrafast diffusion equations: from well-posedness to long-time behaviour”. In: Archive for Rational Mechanics and Analysis 232 (2019), pp. 1165–1206. [JKO98] Richard Jordan, David Kinderlehrer, and Felix Otto. “The variat...

arXiv 2019
[25]

A gradient flow approach to quantization of measures

[CGI15] Emanuele Caglioti, Fran¸ cois Golse, and Mikaela Iacobelli. “A gradient flow approach to quantization of measures”. In: Mathematical Models and Methods in Applied Sciences 25.10 (2015), pp. 1845–1885. [CGP16] Yongxin Chen, Tryphon T Georgiou, and Michele Pavon. “On the relation between optimal transport and Schr¨ odinger bridges: A stochastic cont...

work page 2015
[27]

Improved Particle Approximation Error for Mean Field Neural Networks

[MMM19] Song Mei, Theodor Misiakiewicz, and Andrea Montanari. “Mean-field theory of two-layers neural networks: dimension-free bounds and kernel limit”. In: Conference on Learning Theory. PMLR. 2019, pp. 2388–2464. [MMN18] Song Mei, Andrea Montanari, and Phan-Minh Nguyen. “A mean field view of the landscape of two-layer neural networks”. In: Proceedings o...

work page Pith review arXiv 2018
[94]

Some estimates of the transition density of a nondegenerate diffusion Markov process

[She91] Shuenn-Jyi Sheu. “Some estimates of the transition density of a nondegenerate diffusion Markov process”. In: The Annals of Probability (1991), pp. 538–561. [SS20] Justin Sirignano and Konstantinos Spiliopoulos. “Mean field analysis of neural networks: A law of large numbers”. In: SIAM Journal on Applied Mathematics 80.2 (2020), pp. 725–752. [SWN23...

work page 1991
[691]

Long-time behaviour and phase transitions for the McKean– Vlasov equation on the torus

[CGPS20] Jose Antonio Carrillo, Rishabh S Gvalani, Grigorios A Pavliotis, and Andre Schlichting. “Long-time behaviour and phase transitions for the McKean– Vlasov equation on the torus”. In: Archive for Rational Mechanics and Anal- ysis 235.1 (2020), pp. 635–690. [Che23] Sinho Chewi. An optimization perspective on log-concave sampling and beyond. Massachu...

work page 2020
[2005]

Neural Wasser- stein Gradient Flows for Discrepancies with Riesz Kernels

[AHS23] Fabian Altekr¨ uger, Johannes Hertrich, and Gabriele Steidl. “Neural Wasser- stein Gradient Flows for Discrepancies with Riesz Kernels”. In: International Conference on Machine Learning . PMLR. 2023, pp. 664–690. [AKSG19] Michael Arbel, Anna Korba, Adil Salim, and Arthur Gretton. “Maximum mean discrepancy gradient flow”. In: Advances in Neural Inf...

arXiv 2019
[2007]

Quantita- tive uniform stability of the iterative proportional fitting procedure

isbn: 9788876423130. [DDD24] George Deligiannidis, Valentin De Bortoli, and Arnaud Doucet. “Quantita- tive uniform stability of the iterative proportional fitting procedure”. In: The Annals of Applied Probability 34.1A (2024), pp. 501–516. [Fey+19] Jean Feydy, Thibault S´ ejourn´ e, Fran¸ cois-Xavier Vialard, Shun-ichi Amari, Alain Trouv´ e, and Gabriel P...

work page 2024

Show all 13 references

[2019]

Sharp explicit lower bounds of heat kernels

[Wan97] Feng-Yu Wang. “Sharp explicit lower bounds of heat kernels”. In: The Annals of Probability 25.4 (1997), pp. 1995–2006. [Wib18] Andre Wibisono. “Sampling as optimization in the space of measures: The Langevin dynamics as a composite optimization problem”. In: Conference...

1997
[2020]

Sharp uniform-in-time mean-field convergence for singular periodic Riesz flows

[CRS23] Antonin Chodron de Courcel, Matthew Rosenzweig, and Sylvia Serfaty. “Sharp uniform-in-time mean-field convergence for singular periodic Riesz flows”. In: Annales de l’Institut Henri Poincar´ e C (2023). [CR W22] Fan Chen, Zhenjie Ren, and Songbo Wang. “Uniform-in-time ...

2023
[2021]

Sample complexity of Sinkhorn divergences

[GCBCP19] Aude Genevay, L´ ena ¨ ıc Chizat, Francis Bach, Marco Cuturi, and Gabriel Peyr´ e. “Sample complexity of Sinkhorn divergences”. In:The 22nd International Con- ference on Artificial Intelligence and Statistics . PMLR. 2019, pp. 1574–1583. [GLR17] Ivan Gentil, Christia...

2019
[2023]

Mean-Field Langevin Dynamics: Exponential Convergence and Annealing

[Chi22] L´ ena ¨ ıc Chizat. “Mean-Field Langevin Dynamics: Exponential Convergence and Annealing”. In: Transactions on Machine Learning Research (2022). [Chi25] L´ ena ¨ ıc Chizat. “Doubly regularized entropic Wasserstein barycenters”. In: To appear in Foundations of Computati...

2022 arXiv
[2801]

The Wasserstein gra- dient flow of the Fisher information and the quantum drift-diffusion equation

[GST09] Ugo Gianazza, Giuseppe Savar´ e, and Giuseppe Toscani. “The Wasserstein gra- dient flow of the Fisher information and the quantum drift-diffusion equation”. In: Archive for rational mechanics and analysis 194.1 (2009), pp. 133–220. [Hag+24] Paul Hagemann, Johannes Hert...

2009

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [3]

Global well-posedness of master equa- tions for deterministic displacement convex potential mean field games

2017, pp. 569–600. [GM22] Wilfrid Gangbo and Alp´ ar R M´ esz´ aros. “Global well-posedness of master equa- tions for deterministic displacement convex potential mean field games”. In: Communications on Pure and Applied Mathematics 75.12 (2022), pp. 2685–

work page 2022

[2] [4]

Weighted ultrafast diffusion equations: from well-posedness to long-time behaviour

Institut Henri Poincar´ e. 2021, pp. 2043–2065. [IPS19] Mikaela Iacobelli, Francesco S Patacchini, and Filippo Santambrogio. “Weighted ultrafast diffusion equations: from well-posedness to long-time behaviour”. In: Archive for Rational Mechanics and Analysis 232 (2019), pp. 1165–1206. [JKO98] Richard Jordan, David Kinderlehrer, and Felix Otto. “The variat...

arXiv 2019

[3] [25]

A gradient flow approach to quantization of measures

[CGI15] Emanuele Caglioti, Fran¸ cois Golse, and Mikaela Iacobelli. “A gradient flow approach to quantization of measures”. In: Mathematical Models and Methods in Applied Sciences 25.10 (2015), pp. 1845–1885. [CGP16] Yongxin Chen, Tryphon T Georgiou, and Michele Pavon. “On the relation between optimal transport and Schr¨ odinger bridges: A stochastic cont...

work page 2015

[4] [27]

Improved Particle Approximation Error for Mean Field Neural Networks

[MMM19] Song Mei, Theodor Misiakiewicz, and Andrea Montanari. “Mean-field theory of two-layers neural networks: dimension-free bounds and kernel limit”. In: Conference on Learning Theory. PMLR. 2019, pp. 2388–2464. [MMN18] Song Mei, Andrea Montanari, and Phan-Minh Nguyen. “A mean field view of the landscape of two-layer neural networks”. In: Proceedings o...

work page Pith review arXiv 2018

[5] [94]

Some estimates of the transition density of a nondegenerate diffusion Markov process

[She91] Shuenn-Jyi Sheu. “Some estimates of the transition density of a nondegenerate diffusion Markov process”. In: The Annals of Probability (1991), pp. 538–561. [SS20] Justin Sirignano and Konstantinos Spiliopoulos. “Mean field analysis of neural networks: A law of large numbers”. In: SIAM Journal on Applied Mathematics 80.2 (2020), pp. 725–752. [SWN23...

work page 1991

[6] [691]

Long-time behaviour and phase transitions for the McKean– Vlasov equation on the torus

[CGPS20] Jose Antonio Carrillo, Rishabh S Gvalani, Grigorios A Pavliotis, and Andre Schlichting. “Long-time behaviour and phase transitions for the McKean– Vlasov equation on the torus”. In: Archive for Rational Mechanics and Anal- ysis 235.1 (2020), pp. 635–690. [Che23] Sinho Chewi. An optimization perspective on log-concave sampling and beyond. Massachu...

work page 2020

[7] [2005]

Neural Wasser- stein Gradient Flows for Discrepancies with Riesz Kernels

[AHS23] Fabian Altekr¨ uger, Johannes Hertrich, and Gabriele Steidl. “Neural Wasser- stein Gradient Flows for Discrepancies with Riesz Kernels”. In: International Conference on Machine Learning . PMLR. 2023, pp. 664–690. [AKSG19] Michael Arbel, Anna Korba, Adil Salim, and Arthur Gretton. “Maximum mean discrepancy gradient flow”. In: Advances in Neural Inf...

arXiv 2019

[8] [2007]

Quantita- tive uniform stability of the iterative proportional fitting procedure

isbn: 9788876423130. [DDD24] George Deligiannidis, Valentin De Bortoli, and Arnaud Doucet. “Quantita- tive uniform stability of the iterative proportional fitting procedure”. In: The Annals of Applied Probability 34.1A (2024), pp. 501–516. [Fey+19] Jean Feydy, Thibault S´ ejourn´ e, Fran¸ cois-Xavier Vialard, Shun-ichi Amari, Alain Trouv´ e, and Gabriel P...

work page 2024