Pith. sign in

REVIEW 3 minor 66 references

Post-processed frozen-flow methods for the long time sampling of ergodic dynamics on Riemannian manifolds

T0 review · 0 major / 3 minor · reviewed 2026-06-28 · grok-4.3

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

Pith's one-line read Post-processed frozen-flow methods sample the invariant measure of ergodic dynamics on Riemannian manifolds to high order using only intrinsic operations.

desk verdict The paper gives an intrinsic post-processed integrator for long-time invariant measure sampling on manifolds via a new operation on exotic Lie-Butcher series, with the central construction looking consistent but the performance claims needing the experiments to hold up. read the letter →

arxiv 2606.06150 v1 pith:YTHWFL6U submitted 2026-06-04 math.NA cs.NAmath.COmath.DGmath.PR

Adrien Busnot Laurent , Sébastien Macé This is my paper · ORCID
classification math.NAcs.NAmath.COmath.DGmath.PR
keywords RiemannianmanifoldergodicSDEinvariantmeasurefrozenflowLie-Butcherseriesintrinsicintegratorpost-processingnumericalsampling
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 proposes post-processed frozen-flow methods as an intrinsic way to approximate ergodic stochastic differential equations on Riemannian manifolds. The methods rely on geodesics and parallel transport rather than embeddings or coordinates. A criterion for high-order accuracy in the invariant measure is given, derived through a new algebraic operation on exotic Lie-Butcher series. The approach emphasizes efficiency in long-time sampling of the invariant measure, showing better cost-accuracy tradeoffs than prior methods in experiments.

What carries the argument

The post-processed frozen-flow integrator, which applies a post-processing step to a frozen-flow discretization to boost the order of accuracy for the invariant measure, analyzed using a novel algebraic operation on exotic Lie-Butcher series.

What would settle it

Running the method on a simple manifold like the sphere with a known ergodic SDE and observing that the empirical measure converges at a lower order than predicted, or that the computed invariant measure deviates significantly from the theoretical one.

Watch

Extended reading notes

Core claim

The central discovery is that frozen-flow integrators, when post-processed appropriately, can be made to sample the invariant measure of an ergodic SDE on a Riemannian manifold to high order. This is achieved by developing new intrinsic schemes that use only natural geometric operations and by establishing high-order conditions via analysis with exotic Lie-Butcher series, prioritizing long-time measure accuracy over finite-time path accuracy.

Load-bearing premise

The manifold allows accurate evaluation of geodesics and parallel transport, and the SDE is ergodic with a unique invariant measure that the scheme is designed to preserve.

Editorial extensions

If this is right

  • These methods outperform previous extrinsic and intrinsic approaches in computational cost for a given accuracy level.
  • They apply directly to Riemannian Langevin dynamics without needing coordinate charts or embeddings.
  • High-order conditions for the invariant measure can be systematically derived using the new operation on Lie-Butcher series.
  • The schemes preserve the unique invariant measure to the designed order provided the underlying dynamics are ergodic.

Reading between the lines

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

  • The emphasis on invariant measure sampling could lead to more efficient algorithms in statistical mechanics simulations on curved spaces.
  • Similar post-processing ideas might apply to other geometric integrators where long-time behavior is key.
  • If the algebraic framework generalizes, it could enable higher-order methods for a broader class of manifold-valued processes.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

0 major / 3 minor

Summary. The paper proposes post-processed frozen-flow integrators as an intrinsic approach to long-time sampling of ergodic SDEs (including Riemannian Langevin dynamics) on Riemannian manifolds. It develops a criterion for high-order accuracy with respect to the invariant measure, introduces new intrinsic methods designed specifically for invariant-measure sampling (prioritizing this over finite-time accuracy in the spirit of Leimkuhler-Matthews), derives high-order conditions via a new algebraic operation on exotic Lie-Butcher series, and claims superior cost-for-accuracy performance relative to prior extrinsic (penalization/projection) and intrinsic schemes, as illustrated by numerical experiments.

Significance. If the high-order criterion and outperformance claims hold, the work would advance geometric stochastic integrators by supplying coordinate-free methods that exploit manifold structure (geodesics and parallel transport) for efficient ergodic sampling. The introduction of a new algebraic operation on exotic Lie-Butcher series for deriving invariant-measure order conditions is a notable technical contribution.

minor comments (3)
  1. [Numerical experiments] The abstract and introduction reference numerical experiments demonstrating outperformance, but the manuscript should include a dedicated section (e.g., §4 or §5) with explicit tables or figures comparing cost versus accuracy (e.g., wall-clock time or function evaluations versus error in invariant-measure statistics) against the cited extrinsic and intrinsic baselines.
  2. [§3 (or wherever the operation is defined)] Notation for the new algebraic operation on exotic Lie-Butcher series should be introduced with a self-contained definition or table early in the methods section to aid readability for readers unfamiliar with the extension of standard Butcher series.
  3. [Assumptions and experiments] The assumption that geodesics and parallel transport can be evaluated accurately is stated, but the paper should briefly discuss the impact of approximate geodesic solvers (e.g., via retraction approximations) on the observed order in the experiments.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their careful reading and positive evaluation of the manuscript, including the accurate summary of our contributions and the recommendation for minor revision. No specific major comments were raised in the report.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity detected

full rationale

The paper introduces a new intrinsic methodology for ergodic SDEs on Riemannian manifolds relying on geodesics and parallel transport, a criterion for invariant-measure accuracy derived from a novel algebraic operation on exotic Lie-Butcher series, and post-processed frozen-flow integrators. No load-bearing step reduces by construction to a fitted parameter, self-definition, or self-citation chain; the central claims rest on independently proposed operations and methods whose validity is to be verified externally via the referenced numerical experiments. The construction is explicitly conditioned on computable geometric operations and ergodicity assumptions that do not embed the target accuracy result. This matches the default expectation for a paper whose derivation chain remains self-contained against external benchmarks.

Assumptions & free parameters 0 free parameters · 2 assumptions · 0 invented entities

Based solely on abstract: methods rely on standard differential geometry (geodesics, parallel transport) and a new algebraic structure on Lie-Butcher series; no explicit free parameters or invented entities stated.

assumptions (2)
  • domain assumption Riemannian manifold admits sufficiently smooth geodesics and parallel transport that can be evaluated for numerical integration steps.
    Invoked when stating that the methodology relies only on natural geometric operations without embeddings.
  • domain assumption The SDE is ergodic and possesses a unique invariant measure that the integrator is designed to sample.
    Central to the claim of high-order accuracy for the invariant measure.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Post-processed frozen-flow methods for the long time sampling of ergodic dynamics on Riemannian manifolds." pith.science (2026). https://pith.science/paper/YTHWFL6U

@misc{pith2026260606150,
  author       = {Pith},
  title        = {Pith review of: Post-processed frozen-flow methods for the long time sampling of ergodic dynamics on Riemannian manifolds},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YTHWFL6U}},
  note         = {Machine review of arXiv:2606.06150}
}
read the original abstract

In this work, we propose a novel intrinsic approach to the approximation of ergodic SDEs on Riemannian manifolds, which include Riemannian Langevin dynamics. In opposition to the standard extrinsic approaches such as penalization methods and projection methods, our methodology does not use embeddings or coordinates and only relies on natural geometric operations: geodesics, parallel transport,... We give a criterion for high order of accuracy for the invariant measure, develop new intrinsic numerical methods designed solely for sampling the invariant measure, and derive high order conditions using a new algebraic operation on exotic Lie-Butcher series. In the spirit of the Leimkuhler-Matthews method, our approach prioritizes long time sampling efficiency over finite time accuracy, and outperforms the previous extrinsic and intrinsic approaches in terms of cost for a given accuracy, which we illustrate with several numerical experiments.

Figures

Figures reproduced from arXiv: 2606.06150 by the authors.

Figure 1
Figure 1. Trajectory on SO3 with quadratic potential at time T “ 2 and step time h “ 10´5 , represented by the three columns on the sphere S 2 (Left). Ergodic convergence computed with 104 trajectories and h “ 10´4 for the quadratic potential and the test function φ (Right). The method used is Method 2 in both computation. We observe the error curves for the invariant measure in [PITH_FULL_IMAGE:figures/full_fig_p029_1.png] view at source ↗
Figure 2
Figure 2. Order of convergence in long time for two potentials: quadratic (left) and sextic (right). The mean is computed at the final time T “ 10 with 108 trajectories and with the test function φ given by (5.1). 5.2 Von-Mises Fisher dynamics on the sphere S 2 Our second experiment focuses on the 2-dimensional sphere M “ S 2 . As our new methods rely on orthonormal bases, we consider the standard coordinates on S 2 minus 29 … view at source ↗
Figure 3
Figure 3. Trajectory for the Von-Mises Fisher dynamics on the sphere S 2 at time T “ 0.4 and step time h “ 4 ˆ 10´6 (Left). Convergence order for the Von-Mises Fisher dynamics on the sphere S 2 at time T “ 0.4 with 108 trajectories and with the test function φ : px, y, zq ÞÑ z 2 (Right). 30 [PITH_FULL_IMAGE:figures/full_fig_p030_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

66 extracted references · 6 canonical work pages

  1. [1]

    Abdulle, G

    A. Abdulle, G. A. Pavliotis, and G. Vilmart. Accelerated convergence to equilibrium and reduced asymptotic variance for Langevin dynamics using Stratonovich perturbations.C. R. Math. Acad. Sci. Paris, 357(4):349–354, 2019

    2019

  2. [2]

    Abdulle, G

    A. Abdulle, G. Vilmart, and K. C. Zygalakis. High order numerical approximation of the invariant measure of ergodic SDEs.SIAM J. Numer. Anal., 52(4):1600–1622, 2014

    2014

  3. [3]

    Abdulle, G

    A. Abdulle, G. Vilmart, and K. C. Zygalakis. Long time accuracy of Lie-Trotter splitting methods for Langevin dynamics.SIAM J. Numer. Anal., 53(1):1–16, 2015. 31

    2015

  4. [4]

    M. J. H. Al-Kaabi, K. Ebrahimi-Fard, D. Manchon, and H. Z. Munthe-Kaas. Algebraic aspects of connections: From torsion, curvature, and post-Lie algebras to Gavrilov’s double exponential and special polynomials.Journal of Noncommutative Geometry, 19(1):297–335, 2023

    2023

  5. [5]

    Alama Bronsard, Y

    Y. Alama Bronsard, Y. Bruned, and K. Schratz. Approximations of dispersive PDEs in the presence of low-regularity randomness.Found. Comput. Math., pages 1–51, 2024

    2024

  6. [6]

    Antonyuk and A

    A. Antonyuk and A. Antonyuk. Nonexplosion and solvability of nonlinear diffusion equations on noncompact manifolds.Ukr. Math. J., 59:1632–1652, 2007

    2007

  7. [7]

    D. Bakry. Un critère de non-explosion pour certaines diffusions sur une variété riemannienne com- plète.C.R. Acad. Sc. Paris, 303(1):23–26, 1986

    1986

  8. [8]

    Baxter et al

    G. Baxter et al. An analytic problem whose solution follows from a simple algebraic identity.Pacific J. Math, 10(3):731–742, 1960

    1960

Show all 66 references
  1. [9]

    Bharath, A

    K. Bharath, A. Lewis, A. Sharma, and M. V. Tretyakov. Sampling and Estimation on Manifolds using the Langevin Diffusion.Journal of Machine Learning Research, 26(71):1–50, 2025

    2025

  2. [10]

    Bonicelli

    A. Bonicelli. Exotic B-series representation of the Feller semigroup for Itô diffusions and the MSR path integral.arXiv preprint arXiv:2510.23102, 2025

    2025 arXiv

  3. [11]

    Long-runaccuracyofvariationalintegratorsinthestochasticcontext

    N.Bou-RabeeandH.Owhadi. Long-runaccuracyofvariationalintegratorsinthestochasticcontext. SIAM J. Numer. Anal., 48(1):278–297, 2010

    2010

  4. [12]

    Bronasco

    E. Bronasco. Exotic B-series and S-series: algebraic structures and order conditions for invariant measure sampling.Found. Comput. Math., pages 1–31, 2024

    2024

  5. [13]

    Bronasco and A

    E. Bronasco and A. Busnot Laurent. Hopf algebra structures for the backward error analysis of ergodic stochastic differential equations.Numer. Math., pages 1–61, 2026

    2026

  6. [14]

    Bronasco, A

    E. Bronasco, A. Busnot Laurent, and B. Huguet. High order integration of stochastic dynamics on Riemannian manifolds with frozen-flow methods.arXiv:2503.21855, 2025

    2025

  7. [15]

    Bronasco, B

    E. Bronasco, B. Leimkuhler, D. Phillips, and G. Vilmart. Efficient Langevin sampling with position- dependent diffusion.arXiv:2501.02943, 2025

    2025

  8. [16]

    C. Brouder. Runge–Kutta methods and renormalization.Eur. Phys. J. C, 12(3):521–534, 2000

    2000

  9. [17]

    Busnot Laurent, K

    A. Busnot Laurent, K. Debrabant, and A. Kværnø. Optimal stochastic Runge-Kutta methods for the weak integration of stochastic dynamics.arXiv:2603.24255, 2026

    2026

  10. [18]

    Busnot Laurent, Y

    A. Busnot Laurent, Y. Li, and Y. Sheng. Post-Hopf algebroids, post-Lie-Rinehart algebras and geometric numerical integration.arXiv:2512.21971, 2025

    2025

  11. [19]

    Busnot Laurent, H

    A. Busnot Laurent, H. Munthe-Kaas, and G. S. Venkatesh. The free tracial post-Lie-Rinehart alge- bra of planar aromatic trees for the design of divergence-free Lie-group methods.arXiv:2603.28437, 2026

    2026 arXiv

  12. [20]

    J. C. Butcher. An algebraic theory of integration methods.Math. Comp., 26:79–106, 1972

    1972

  13. [21]

    J. C. Butcher.B-series: algebraic analysis of numerical methods. Springer, 2021

    2021

  14. [22]

    Celledoni, A

    E. Celledoni, A. Marthinsen, and B. Owren. Commutator-free Lie group methods.Future Genera- tion Computer Systems, 19(3):341–352, 2003

    2003

  15. [23]

    P. E. Crouch and R. Grossman. Numerical integration of ordinary differential equations on mani- folds.Journal of Nonlinear Science, 3:1–33, 1993

    1993

  16. [24]

    Debussche and E

    A. Debussche and E. Faou. Weak backward error analysis for SDEs.SIAM J. Numer. Anal., 50(3):1735–1752, 2012

    2012

  17. [25]

    Deng and Z

    Y. Deng and Z. Hani. Full derivation of the wave kinetic equation.Inventiones mathematicae, 233(2):543–724, 2023

    2023

  18. [26]

    A. B. Duncan, T. Lelièvre, and G. A. Pavliotis. Variance reduction using nonreversible Langevin samplers.J. Stat. Phys., 163(3):457–491, 2016. 32

    2016

  19. [27]

    Ebrahimi-Fard, A

    K. Ebrahimi-Fard, A. Lundervold, and H. Z. Munthe-Kaas. On the Lie enveloping algebra of a post-Lie algebra.J. Lie Theory, 25(4):1139–1165, 2015

    2015

  20. [28]

    Ebrahimi-Fard and D

    K. Ebrahimi-Fard and D. Manchon. The Magnus expansion, trees and Knuth’s rotation correspon- dence.Found. Comput. Math., 14(1):1–25, 2014

    2014

  21. [29]

    Fløystad, D

    G. Fløystad, D. Manchon, and H. Z. Munthe-Kaas. The universal pre-Lie-Rinehart algebras of aromatic trees. InGeometric and harmonic analysis on homogeneous spaces and applications, volume 366 ofSpringer Proc. Math. Stat., pages 137–159. Springer, Cham, [2021]©2021

    2021

  22. [30]

    Grong, H

    E. Grong, H. Z. Munthe-Kaas, and J. Stava. Post-Lie algebra structure of manifolds with constant curvature and torsion.Journal of Lie Theory, 34(2):339–352, 2024

    2024

  23. [31]

    Hairer, C

    E. Hairer, C. Lubich, and G. Wanner.Geometric numerical integration, volume 31 ofSpringer Series in Computational Mathematics. Springer-Verlag, Berlin, second edition, 2006. Structure-preserving algorithms for ordinary differential equations

    2006

  24. [32]

    Hairer and D

    M. Hairer and D. Kelly. Geometric versus non-geometric rough paths.Ann. Inst. Henri Poincaré Probab. Stat., 51(1):207–251, 2015

    2015

  25. [33]

    E. P. Hsu.Stochastic analysis on manifolds, volume 38 ofGraduate Studies in Mathematics. Amer- ican Mathematical Society, Providence, RI, 2002

    2002

  26. [34]

    Iserles, H

    A. Iserles, H. Z. Munthe-Kaas, S. P. Nørsett, and A. Zanna. Lie-group methods. InActa numerica, 2000, volume 9 ofActa Numer., pages 215–365. Cambridge Univ. Press, Cambridge, 2000

    2000

  27. [35]

    Laurent.Algebraic Tools and Multiscale Methods for the Numerical Integration of Stochastic Evolutionary Problems

    A. Laurent.Algebraic Tools and Multiscale Methods for the Numerical Integration of Stochastic Evolutionary Problems. PhD thesis, University of Geneva, 2021

    2021

  28. [36]

    A. Laurent. The Lie derivative and Noether’s theorem on the aromatic bicomplex for the study of volume-preserving numerical integrators.J. Comput. Dyn., 11(1):10–22, 2024

    2024

  29. [37]

    Laurent, R

    A. Laurent, R. I. McLachlan, H. Z. Munthe-Kaas, and O. Verdier. The aromatic bicomplex for the description of divergence-free aromatic forms and volume-preserving integrators.Forum Math. Sigma, 11:Paper No. e69, 2023

    2023

  30. [38]

    Laurent and H

    A. Laurent and H. Munthe-Kaas. The universal equivariance properties of exotic aromatic B-series. Found. Comput. Math., 25(5):1595–1626, 2025

    2025

  31. [39]

    Laurent and G

    A. Laurent and G. Vilmart. Exotic aromatic B-series for the study of long time integrators for a class of ergodic SDEs.Math. Comp., 89(321):169–202, 2020

    2020

  32. [40]

    Laurent and G

    A. Laurent and G. Vilmart. Order conditions for sampling the invariant measure of ergodic stochas- tic differential equations on manifolds.Found. Comput. Math., 22(3):649–695, 2022

    2022

  33. [41]

    Lee.Introduction to Riemannian Manifolds

    J. Lee.Introduction to Riemannian Manifolds. Graduate Texts in Mathematics. Springer Interna- tional Publishing, 2019

    2019

  34. [42]

    Leimkuhler and C

    B. Leimkuhler and C. Matthews. Rational construction of stochastic numerical methods for molec- ular sampling.Appl. Math. Res. Express. AMRX, 2013(1):34–56, 2013

    2013

  35. [43]

    Leimkuhler, C

    B. Leimkuhler, C. Matthews, and G. Stoltz. The computation of averages from equilibrium and nonequilibrium Langevin molecular dynamics.IMA J. Numer. Anal., 36(1):13–79, 2016

    2016

  36. [44]

    A. Lejay. Constructing general rough differential equations through flow approximations.Electron. J. Probab., 27:Paper No. 7, 24, 2022

    2022

  37. [45]

    Lelièvre, F

    T. Lelièvre, F. Nier, and G. A. Pavliotis. Optimal non-reversible linear drift for the convergence to equilibrium of a diffusion.J. Stat. Phys., 152(2):237–274, 2013

    2013

  38. [46]

    Lelièvre, M

    T. Lelièvre, M. Rousset, and G. Stoltz.Free energy computations. Imperial College Press, London,

  39. [47]

    A mathematical perspective

  40. [48]

    X.-M. Li. Stochastic differential equations on noncompact manifolds: moment stability and its topological consequences.Probab. Theory Relat. Fields, 100:417–428, 1994

    1994

  41. [49]

    Y. Li, Y. Sheng, and R. Tang. Post-Hopf algebras, relative Rota–Baxter operators and solutions to the Yang–Baxter equation.Journal of Noncommutative Geometry, 18(2):605–630, 2023. 33

    2023

  42. [50]

    Luesink and O

    E. Luesink and O. D. Street. Symplectic techniques for stochastic differential equations on reductive Lie groups with applications to Langevin diffusions.Journal of Differential Equations, 458:114034, 2026

    2026

  43. [51]

    Lundervold and H

    A. Lundervold and H. Munthe-Kaas. Hopf algebras of formal diffeomorphisms and numerical inte- gration on manifolds. InCombinatorics and physics, volume 539 ofContemp. Math., pages 295–324. Amer. Math. Soc., Providence, RI, 2011

    2011

  44. [52]

    S. J. A. Malham and A. Wiese. Stochastic Lie group integrators.SIAM J. Sci. Comput., 30(2):597– 617, 2008

    2008

  45. [53]

    R. I. McLachlan, K. Modin, H. Munthe-Kaas, and O. Verdier. B-series methods are exactly the affine equivariant methods.Numer. Math., 133(3):599–622, 2016

    2016

  46. [54]

    Muniz, M

    M. Muniz, M. Ehrhardt, M. Günther, and R. Winkler. Higher strong order methods for linear Itô SDEs on matrix Lie groups.BIT Numer. Math., 62(4):1095–1119, 2022

    2022

  47. [55]

    Munthe-Kaas

    H. Munthe-Kaas. Geometric integration on symmetric spaces.J. Comput. Dyn., 11(1):43–58, 2024

    2024

  48. [56]

    Munthe-Kaas and J

    H. Munthe-Kaas and J. Stava. Lie admissible triple algebras: The connection algebra of symmetric spaces.Submitted, 2023

    2023

  49. [57]

    Munthe-Kaas and O

    H. Munthe-Kaas and O. Verdier. Aromatic Butcher series.Found. Comput. Math., 16(1):183–215, 2016

    2016

  50. [58]

    H. Z. Munthe-Kaas and A. Lundervold. On post-Lie algebras, Lie–Butcher series and moving frames.Found. Comput. Math., 13:583–613, 2013

    2013

  51. [59]

    H. Z. Munthe-Kaas and W. M. Wright. On the Hopf algebraic structure of Lie group integrators. Found. Comput. Math., 8(2):227–257, 2008

    2008

  52. [60]

    O’Neill.Semi-Riemannian Geometry With Applications to Relativity

    B. O’Neill.Semi-Riemannian Geometry With Applications to Relativity. Pure and Applied Math- ematics. Academic Press, 1983

    1983

  53. [61]

    Oudom and D

    J.-M. Oudom and D. Guin. On the Lie enveloping algebra of a pre-Lie algebra.J. K-Theory, 2(1):147–167, 2008

    2008

  54. [62]

    B. Owren. Order conditions for commutator-free Lie group methods.Journal of Physics A: Math- ematical and General, 39(19):5585, 2006

    2006

  55. [63]

    Owren and A

    B. Owren and A. Marthinsen. Runge-Kutta methods adapted to manifolds and based on rigid frames.BIT Numer. Math., 39(1):116–142, 1999

    1999

  56. [64]

    Stava.On connection algebras of symmetric spaces and reductive homogeneous spaces

    J. Stava.On connection algebras of symmetric spaces and reductive homogeneous spaces. PhD thesis, University of Bergen, 2024

    2024

  57. [65]

    Talay and L

    D. Talay and L. Tubaro. Expansion of the global error for numerical schemes solving stochastic differential equations.Stochastic Anal. Appl., 8(4):483–509 (1991), 1990

    1991

  58. [66]

    G. Vilmart. Postprocessed integrators for the high order integration of ergodic SDEs.SIAM J. Sci. Comput., 37(1):A201–A220, 2015. 34

    2015

Pith tools

Reviewed June 28, 2026 · model on record in the stance chip above.