Pith. sign in

REVIEW 4 major objections 5 minor 33 references

Model-free learning of probability flows: Elucidating the nonequilibrium dynamics of flocking

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper establishes that entropy production rates can be learned from trajectory data alone and localizes the breakdown of time-reversal symmetry to the interface of a flock.

desk verdict Solid analytic core and a plausible model-free estimator, but the high-dimensional results are under-validated; the paper needs a round of careful numerical checks before the conclusions can be trusted. read the letter →

arxiv 2411.14317 v1 pith:W2NSMT6Y submitted 2024-11-21 cond-mat.stat-mech cs.LGmath.PR

Nicholas M. Boffi , Eric Vanden-Eijnden This is my paper · ORCID
classification cond-mat.stat-mechcs.LGmath.PR
keywords entropyproductionrateactivematterflockingtime-reversalsymmetryprobabilitycurrentmachinelearninggraphneuralnetworknonequilibriumsteadystate
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 claims that for inertial active systems, the entropy production rate (EPR) can be estimated directly from stochastic trajectories, with no knowledge of the forces or equations of motion. The route is to learn the probability current velocity $g(x,v)=f(x,v)-\gamma v-\gamma v_*^2\nabla_v\log\rho(x,v)$, the conditional mean acceleration of particles on the nonequilibrium steady state; two identities tie $g$ to the local total EPR and the local system EPR. Applied to a Vicsek-like flocking model, the learned fields show that time-reversal symmetry is broken mostly at the interface between ordered flocks and the disordered gas, and that the system EPR is negative when particles align (order creation) and positive when they anti-align (order destruction). The paper's value is that it removes the need for a known dynamical model, opening entropy-production diagnostics to experimental trajectory data and giving spatially resolved maps of where a system is out of equilibrium.

What carries the argument

The load-bearing object is the current velocity $g(x,v)=f(x,v)-\gamma v-\gamma v_*^2\nabla_v\log\rho(x,v)$, which equals the conditional mean acceleration $\langle\dot v_t\,|\,(x_t,v_t)=(x,v)\rangle$ and whose flow lines $(v,g)\rho$ form the probability current on the steady state. The argument rides on two identities: $g$ squared gives the local total EPR and the velocity divergence of $g$ gives the local system EPR. The computational machinery is the variational objective $L[\hat g]=\frac{1}{T}\mathbb{E}\left[\int_0^T|\hat g(x_t,v_t)|^2\,dt-2\hat g(x_t,v_t)\circ dv_t\right]$, which has $g$ as its unique minimizer and is discretized over a single timestep, then minimized with a permutation-equivariant and translation-invariant graph neural network.

What would settle it

Run the same estimator on a system whose entropy production is exactly computable, such as the two-particle Vicsek pair reduced to one dimension by (60), solving the stationary Fokker-Planck equation for the true $g$ and comparing the learned $\dot s_{\rm sys}$ and $\dot s_{\rm tot}$ fields pointwise; if the learned fields deviate in low-probability regions or fail to reproduce the known macroscopic EPR, the 64-particle spatial-localization conclusions would be unsupported.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that two local measures of irreversibility in a nonequilibrium steady state are both determined by one vector field, the current velocity $g$. The total entropy production rate satisfies $\dot s_{\rm tot}(x,v)=|g^R(x,v)|^2/(\gamma v_*^2)$ with $g^R(x,v)=-g(x,-v)$, and the system entropy production rate satisfies $\dot s_{\rm sys}(x,v)=\nabla_v\cdot g(x,v)$. Because $g$ is the unique minimizer of a variational objective built from trajectory increments, it can be learned without specifying the active force $f$, and the authors do so for a 256-dimensional Vicsek-like system with a graph neural network. The learned fields indicate that entropy is produced and consumed on the spatial interface of a flock as alignment and fluctuation create and destroy order, with total EPR spikes marking flock breakup and mergers.

Load-bearing premise

The load-bearing premise is that the trained neural-network estimate of the current velocity is accurate enough in the 256-dimensional phase space that the computed entropy-production maps and their statistics reflect the true dynamics; the paper reports no validation against an exact solution, no error bars, and no convergence study.

Editorial extensions

If this is right

  • Entropy-production diagnostics become available for systems whose dynamics are unknown, such as experimental active matter or animal groups, provided trajectories can be tracked.
  • The two identities give per-particle decompositions of both EPRs, so the spatial location of time-reversal symmetry breaking can be visualized even in a 256-dimensional phase space.
  • In the flocking model, particles on the boundary of a flock drive almost all entropy production; deep inside an aligned flock the system EPR vanishes.
  • The sign of the system EPR distinguishes order creation from order destruction, while the total EPR does not carry that signed information.
  • The EPR time series show heavy-tailed statistics and power spectra compatible with $1/f$ noise, indicating intermittency tied to flock formation and breakup.

Reading between the lines

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

  • If the learned current velocity is faithful at $N=64$, the paper's localization picture implies that the thermodynamic cost of maintaining a flock is paid at its boundary, so entropy production in larger flocks should scale with interface length rather than system volume.
  • The variational objective itself contains no noise-strength parameter, so $g$ can be learned even when friction or noise amplitude is unknown; converting $g$ into an EPR with physical units through (7) and (11) would then require estimating $\gamma$ and $v_*$ separately.
  • A finite-size test of the two-particle mechanism is natural: with open boundaries, where collisions are not forced by periodic conditions, the heavy negative tail of $\dot s_{\rm sys}$ should become less prominent if it is collision-driven.
  • Since the probability flow $\dot x=v,\ \dot v=g$ preserves the steady state, the learned $g$ is also a generative model of the nonequilibrium process, not just an EPR estimator.
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 Letter proposes a machine-learning method to estimate probability current velocities g from stochastic trajectories of inertial active-matter systems, and uses them to compute local total (s_tot) and system (s_sys) entropy production rates. The authors derive exact relations: s_tot = |g^R|^2/(γv_*^2) (Eq. 7) and s_sys = ∇_v·g (Eq. 11), and show that g is the unique minimizer of the variational objective (12). They demonstrate the method on a two-particle Vicsek-like system, where a two-dimensional phase-space visualization is possible, and on a 64-particle system in 256-dimensional phase space. The headline physical claim is that entropy is produced and consumed on the spatial interface of a flock, with intermittent dynamics and 1/f noise in the EPR time series. The analytical derivations appear coherent, but the numerical evidence for the high-dimensional results is not validated quantitatively, and several implementation details needed to assess the accuracy of the learned g are omitted.

Significance. If the numerical estimates of g are reliable, the paper offers a genuinely useful tool: it gives a trajectory-only route to local entropy production in high-dimensional inertial systems, with exact formulas (7) and (11) and a rigorously characterized variational objective whose unique minimizer is g (SI Section D). The low-dimensional phase-space pictures provide intuitive validation of the physics. However, the strength of the physical conclusions rests on unverified approximations: the discrete loss (23), the neural-network ansatz (24), and the absence of any quantitative comparison against an exact or reference solution. The paper's main contribution would be much stronger with a convergence or validation study. In its current form, the central claim that the method accurately reveals the spatial structure of entropy production in the N=64 flocking system is not yet supported by the evidence presented.

major comments (4)
  1. [§3 (High-dimensional system), Figs. 3–5] The N=64 results are not validated against any exact or reference solution, nor are error bars or convergence studies provided. The EPR fields, time series, and power spectra are all computed from the learned graph neural network ĝ of Eq. (24), and the reliability of these quantities is simply assumed. Since the identities (7) and (11) are exact only for the true g, the central physical claim about spatial localization on the flock interface requires either validation of ĝ in this setting or at least a capacity/convergence study showing that the estimator is not introducing artifacts.
  2. [End Matter, Eq. (23)] The discrete loss (23) replaces the Stratonovich integral in (12) by a one-step central-difference approximation, which introduces a bias of order Δt in the minimizer. The value of Δt used for the reported simulations is never given, and there is no demonstration that the estimated EPR quantities are insensitive to Δt or to the number of trajectories n. This is load-bearing because the quantitative statements in Figs. 4 and 5, such as the distribution asymmetries and the power-law exponents, depend on the accuracy of the learned g.
  3. [§3 (Low-dimensional system), Eq. (60)] For the N=2, d=1 system, where a full validation is feasible, no comparison is made against a numerical solution of the Fokker–Planck equation (60) or against an independently computed g, e.g. from conditional averages of the SDE. The phase portraits and EPR maps in Fig. 2 and the statistics in Fig. 6 are qualitative; reporting a quantitative error, such as a relative L2 error in the learned g or in ⟨s_tot⟩ and ⟨s_sys⟩, would establish that the neural-network minimization actually recovers the true probability current.
  4. [§3, Fig. 5] The power spectral density analysis quotes exponents p=1.16 and p=1.27 without error bars, fit ranges, or details of the fitting procedure. The claim that the EPR exhibits 1/f noise is a quantitative statement, and the current evidence does not establish its precision or its robustness with respect to the time-series length, binning, or detrending choices.
minor comments (5)
  1. [SI §F.1, Eq. (58)] The smoothing parameter β for the kernel is not reported; since β controls the sharpness of the interaction and affects the learning problem, a value should be specified for both the two-particle and the 64-particle simulations.
  2. [SI §F.2.1] There is a typo in the sentence 'we learn usingwa simple four-layer fully-connected network'; it should read 'using a simple'.
  3. [Throughout, SI §C–E] The notation for the velocity scale is inconsistent: the main text uses v_* while the SI uses v_0 (e.g., Eq. (25) and Eq. (47)). This should be harmonized.
  4. [End Matter, §Estimating the loss] The text says T is arbitrary, yet the discrete loss (23) uses T=Δt only. It should be clarified how stationarity is used to justify this reduction and how the initial samples (x_0,v_0)∼ρ are generated in practice.
  5. [Fig. 3 caption] The references to 'the accompanying movie here' and 'the longer movie here' are not self-contained; URLs or permanent DOIs should be provided in the caption or supplementary material.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: EPR identities and the loss minimizer are derived in-paper; numerical reports are conditional on the learned current, not predetermined.

full rationale

The paper's derivation chain is self-contained. The key identities are derived in the manuscript itself: the reverse-time dynamics (3) follows from the Fokker-Planck equation in Appendix A; the representation g(x,v) = <dv/dt | (x,v)> is proved in Appendix B; the local and global total EPR relations (7) and (8) follow from an explicit path-integral/Girsanov computation in Appendices C and C.1; the system EPR relation (11) follows from the transport form of the stationary Fokker-Planck equation in Appendix E; and the variational loss (12) is shown in Appendix D to be a strongly convex regression objective with unique minimizer g. None of these steps uses the EPR values it claims to produce as training targets or as assumed inputs. The learned field g is obtained by minimizing an empirical version of (12) on trajectory increments, and the reported EPR maps are deterministic functions of that learned field, so the numerical claims inherit the approximation quality of the network; this is a validation and robustness concern, not a circularity. The manuscript contains self-citations to Boffi and Vanden-Eijnden (2023, 2024) for the terminology 'probability flow', the phase-space system EPR definition, and prior machine-learning context, but the load-bearing derivations are reproduced in the present paper, so these citations are not load-bearing. Cross-references to 'SI Appendix' for proofs that actually appear in the End Matter/Appendices are minor organizational inconsistencies rather than circular steps. Overall, no equation or prediction in the paper reduces by construction to its own inputs.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The paper introduces no new physical entities. The free parameters are computational and modeling choices not reported in sufficient detail to assess their influence. The domain assumptions define the class of systems to which the method applies; the ad hoc assumption about the neural network's accuracy is the most fragile element, since the high-dimensional results are not validated.

free parameters (2)
  • neural network training hyperparameters (learning rate, batch size, number of trajectories, integration timestep Δt) = not reported
    These choices determine the accuracy of ĝ and therefore all N=64 numerical results; the paper only reports layer sizes (512/2048 neurons).
  • kernel smoothing parameter β = not reported
    Chosen by hand in Eq (58); the paper states qualitative dynamics are unaffected but does not give the value used in simulations.
assumptions (5)
  • domain assumption The system dynamics are given by (1) with constant, isotropic white velocity noise of strength 2γv*^2; the time-reversed dynamics (3) and the EPR formulas (7) and (11) are derived for this class.
    Invoked at Eq (1); all subsequent derivations of g, the reverse process, and the EPR identities assume this specific SDE structure.
  • domain assumption The NESS density ρ exists, is smooth, strictly positive, and satisfies the stationary Fokker-Planck equation (20); ∇_v log ρ is well-defined.
    Needed for the score in Eq (4), the integration by parts in Appendix D, and the derivation of Eq (11) in Appendix E.
  • domain assumption The dynamics are ergodic, so time averages equal stationary averages as used in Eq (42).
    Used to convert the time average in the macroscopic EPR derivation into a phase-space average over ρ.
  • ad hoc to paper The discrete loss (23) with a central difference accurately approximates the continuous loss (12) for the chosen timestep, and the trained graph neural network (24) approximates g well enough that the N=64 EPR fields are reliable.
    No convergence analysis, error bars, or validation against ground truth is provided for the high-dimensional estimator.
  • standard math Standard Ito-Stratonovich conversion and integration by parts identities.
    Used in Appendix D (Eq 51-53) and Appendix B; these are classical results, not assumptions specific to the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Model-free learning of probability flows: Elucidating the nonequilibrium dynamics of flocking." pith.science (2026). https://pith.science/paper/W2NSMT6Y

@misc{pith2026241114317,
  author       = {Pith},
  title        = {Pith review of: Model-free learning of probability flows: Elucidating the nonequilibrium dynamics of flocking},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W2NSMT6Y}},
  note         = {Machine review of arXiv:2411.14317}
}
read the original abstract

Active systems comprise a class of nonequilibrium dynamics in which individual components autonomously dissipate energy. Efforts towards understanding the role played by activity have centered on computation of the entropy production rate (EPR), which quantifies the breakdown of time reversal symmetry. A fundamental difficulty in this program is that high dimensionality of the phase space renders traditional computational techniques infeasible for estimating the EPR. Here, we overcome this challenge with a novel deep learning approach that estimates probability currents directly from stochastic system trajectories. We derive a new physical connection between the probability current and two local definitions of the EPR for inertial systems, which we apply to characterize the departure from equilibrium in a canonical model of flocking. Our results highlight that entropy is produced and consumed on the spatial interface of a flock as the interplay between alignment and fluctuation dynamically creates and annihilates order. By enabling the direct visualization of when and where a given system is out of equilibrium, we anticipate that our methodology will advance the understanding of a broad class of complex nonequilibrium dynamics.

Figures

Figures reproduced from arXiv: 2411.14317 by the authors.

Figure 1
Figure 1. The nonequilibrium dynamics of flocking. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Two particles: Phase portrait and EPR. (A) Flow lines of the probability current on the NESS. Pendulum-like limit cycles of alignment and anti-alignment can be seen. Comparison with (B) and (C) highlights alternating regions of ˙stot ̸= 0 and ˙stot = 0, as well as alternating regions of system entropy production ( ˙ssys > 0) and system entropy consumption ( ˙ssys < 0) as the particles oscillate in and out of the ali… view at source ↗
Figure 3
Figure 3. Sixty-four particles in two dimensions. The reader is strongly encouraged to view the accompanying movie here, as well as the longer movie here. (A) Uncolored reference depiction of the particle trajectories. Frames chosen based on large spikes in ˙stot, shown by vertical lines in (E) and (F). (B) Particle contributions to ˙ssys. Particles exhibit negative system EPR during alignment and positive system EPR during a… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: EPR statistics. (A) Density of the per-particle system EPR. The distribution exhibits asym￾metry between entropy production and entropy consumption. (B) Density of the per-particle total EPR. The distribution has a heavy tail, but does not distinguish between productio…
Figure 5
Figure 5. Figure 5: Power spectral density. Time series and PSD for the system EPR (A) and the total EPR (B). Dashed line on PSD indicates nonlinear fit to c/ωp . the realistic experimental setting where only a subset of dynamical variables can be observed, and to understand how the choic…
Figure 6
Figure 6. Figure 6: Two Vicsek particles in one dimension: statistics of the stationary distribution [PITH_FULL_IMAGE:figures/full_fig_p021_6.png]
Figure 7
Figure 7. Figure 7: (Left) Example trajectories from the SDE (60). (Right) Stationary density for the SDE (60), obtained by computing a histogram from samples. F.2.2 Additional results Statistics of the learned quantities on the NESS distribution are shown in [PITH_FULL_IMAGE:figures/ful…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 29 canonical work pages

  1. [1]

    The Mechanics and Statistics of Active Matter

    Sriram Ramaswamy. The Mechanics and Statistics of Active Matter . Annual Review of Condensed Matter Physics, 1 0 (1): 0 323--345, August 2010

    work page 2010

  2. [2]

    M. C. Marchetti, J. F. Joanny, S. Ramaswamy, T. B. Liverpool, J. Prost, Madan Rao, and R. Aditi Simha. Hydrodynamics of soft active matter. Reviews of Modern Physics, 85 0 (3): 0 1143--1189, July 2013

    2013

  3. [3]

    Bowick, Nikta Fakhri, M

    Mark J. Bowick, Nikta Fakhri, M. Cristina Marchetti, and Sriram Ramaswamy. Symmetry, Thermodynamics , and Topology in Active Matter . Physical Review X, 12 0 (1): 0 010501, February 2022

    work page 2022

  4. [4]

    Lebowitz and Herbert Spohn

    Joel L. Lebowitz and Herbert Spohn. A Gallavotti – Cohen - Type Symmetry in the Large Deviation Functional for Stochastic Dynamics . Journal of Statistical Physics, 95 0 (1): 0 333--365, April 1999. ISSN 1572-9613

    work page 1999

  5. [5]

    Entropy production along a stochastic trajectory and an integral fluctuation theorem

    Udo Seifert. Entropy production along a stochastic trajectory and an integral fluctuation theorem. Physical Review Letters, 95 0 (4): 0 040602, 2005

    work page 2005

  6. [6]

    Stochastic thermodynamics, fluctuation theorems, and molecular machines

    Udo Seifert. Stochastic thermodynamics, fluctuation theorems, and molecular machines. Reports on Progress in Physics, 75 0 (12): 0 126001, December 2012

    work page 2012

  7. [7]

    Cates, Julien Tailleur, Paolo Visco, and Fr\' e d\' e ric Van Wijland

    \' E tienne Fodor, Cesare Nardini, Michael E. Cates, Julien Tailleur, Paolo Visco, and Fr\' e d\' e ric Van Wijland. How Far from Equilibrium Is Active Matter ? Physical Review Letters, 117 0 (3): 0 038103, July 2016

    work page 2016

  8. [8]

    Cesare Nardini, \' E tienne Fodor, Elsen Tjhung, Fr\' e d\' e ric van Wijland, Julien Tailleur, and Michael E. Cates. Entropy Production in Field Theories without Time - Reversal Symmetry : Quantifying the Non - Equilibrium Character of Active Matter . Physical Review X, 7 0 (2): 0 021007, April 2017

    work page 2017

Show all 33 references
  1. [9]

    Phan, Robert H

    Sunghan Ro, Buming Guo, Aaron Shih, Trung V. Phan, Robert H. Austin, Dov Levine, Paul M. Chaikin, and Stefano Martiniani. Model- Free Measurement of Local Entropy Production and Extractable Work in Active Matter . Physical Review Letters, 129 0 (22): 0 220601, November 2022

    2022

  2. [10]

    Chaikin, and Dov Levine

    Stefano Martiniani, Paul M. Chaikin, and Dov Levine. Quantifying Hidden Order out of Equilibrium . Physical Review X, 9 0 (1): 0 011031, February 2019

    2019

  3. [11]

    Boffi and Eric Vanden-Eijnden

    Nicholas M. Boffi and Eric Vanden-Eijnden. Deep learning probability flows and entropy production rates in active matter. Proceedings of the National Academy of Sciences, 121 0 (25): 0 e2318106121, June 2024

    2024

  4. [12]

    Boffi and Eric Vanden-Eijnden

    Nicholas M. Boffi and Eric Vanden-Eijnden. Probability flow solution of the Fokker – Planck equation. Machine Learning: Science and Technology, 4 0 (3): 0 035012, July 2023

    2023

  5. [13]

    Empirical investigation of starling flocks: a benchmark study in collective animal behaviour

    Michele Ballerini, Nicola Cabibbo, Raphael Candelier, Andrea Cavagna, Evaristo Cisbani, Irene Giardina, Alberto Orlandi, Giorgio Parisi, Andrea Procaccini, Massimiliano Viale, and Vladimir Zdravkovic. Empirical investigation of starling flocks: a benchmark study in collective ...

    2008

  6. [14]

    Pine, and Paul M

    Jeremie Palacci, Stefano Sacanna, Asher Preska Steinberg, David J. Pine, and Paul M. Chaikin. Living crystals of light-activated colloidal surfers. Science (New York, N.Y.), 339 0 (6122): 0 936--940, February 2013

    2013

  7. [15]

    Flocks, herds, and schools: A quantitative theory of flocking

    John Toner and Yuhai Tu. Flocks, herds, and schools: A quantitative theory of flocking. Physical Review E, 58 0 (4): 0 4828--4858, October 1998

    1998

  8. [16]

    Hydrodynamics and phases of flocks

    John Toner, Yuhai Tu, and Sriram Ramaswamy. Hydrodynamics and phases of flocks. Annals of Physics, 318 0 (1): 0 170--244, July 2005

    2005

  9. [17]

    Long- Range Order in a Two - Dimensional Dynamical XY Model : How Birds Fly Together

    John Toner and Yuhai Tu. Long- Range Order in a Two - Dimensional Dynamical XY Model : How Birds Fly Together . Physical Review Letters, 75 0 (23): 0 4326--4329, December 1995

    1995

  10. [18]

    Novel Type of Phase Transition in a System of Self - Driven Particles

    Tamás Vicsek, András Czirók, Eshel Ben-Jacob, Inon Cohen, and Ofer Shochet. Novel Type of Phase Transition in a System of Self - Driven Particles . Physical Review Letters, 75 0 (6): 0 1226--1229, August 1995

    1995

  11. [19]

    Chaté, F

    H. Chaté, F. Ginelli, G. Grégoire, F. Peruani, and F. Raynaud. Modeling collective motion: variations on the Vicsek model. The European Physical Journal B, 64 0 (3-4): 0 451--456, August 2008

    2008

  12. [20]

    J. E. Hirsch, B. A. Huberman, and D. J. Scalapino. Theory of intermittency. Physical Review A, 25 0 (1): 0 519--532, January 1982

    1982

  13. [21]

    Manneville

    P. Manneville. Intermittency, self-similarity and 1/f spectrum in dissipative dynamical systems. Journal de Physique, 41 0 (11): 0 1235--1243, 1980

    1980

  14. [22]

    H. E. Stanley. Dependence of Critical Properties on Dimensionality of Spins . Physical Review Letters, 20 0 (12): 0 589--592, March 1968

    1968

  15. [23]

    Shivers, Irene Giardina, Thierry Mora, and Aleksandra Walczak

    Federica Ferretti, Simon Grosse-Holz, Caroline Holmes, Jordan L. Shivers, Irene Giardina, Thierry Mora, and Aleksandra Walczak. Signatures of irreversibility in microscopic models of flocking. Physical Review E, 106 0 (3): 0 034608, September 2022

    2022

  16. [24]

    Fluctuation Relations for Diffusion Processes

    Raphael Chetrite and Krzysztof Gawedzki. Fluctuation Relations for Diffusion Processes . Communications in Mathematical Physics, 282 0 (2): 0 469--518, September 2008

    2008

  17. [25]

    Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole

    Yang Song, Jascha Sohl-Dickstein, Diederik P. Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole. Score- Based Generative Modeling through Stochastic Differential Equations . arXiv:2011.13456, February 2021

    2011 arXiv

  18. [26]

    Estimation of Non - Normalized Statistical Models by Score Matching

    Aapo Hyvärinen. Estimation of Non - Normalized Statistical Models by Score Matching . Journal of Machine Learning Research, 6 0 (24): 0 695--709, 2005

    2005

  19. [27]

    Gavin E. Crooks. Entropy production fluctuation theorem and the nonequilibrium work relation for free energy differences. Physical Review E, 60 0 (3): 0 2721--2726, September 1999

    1999

  20. [28]

    Onsager and S

    L. Onsager and S. Machlup. Fluctuations and Irreversible Processes . Physical Review, 91 0 (6): 0 1505--1512, September 1953

    1953

  21. [29]

    Machlup and L

    S. Machlup and L. Onsager. Fluctuations and Irreversible Process . II . Systems with Kinetic Energy . Physical Review, 91 0 (6): 0 1512--1515, September 1953

    1953

  22. [30]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. Adam: A Method for Stochastic Optimization . arXiv:1412.6980, January 2017

    2017 arXiv

  23. [31]

    Peter Battaglia, Jessica Blake Chandler Hamrick, Victor Bapst, Alvaro Sanchez, Vinicius Zambaldi, Mateusz Malinowski, Andrea Tacchetti, David Raposo, Adam Santoro, Ryan Faulkner, Caglar Gulcehre, Francis Song, Andy Ballard, Justin Gilmer, George E. Dahl, Ashish Vaswani, Kelsey...

    2018 arXiv

  24. [32]

    Intermittency and Clustering in a System of Self - Driven Particles

    Cristián Huepe and Maximino Aldana. Intermittency and Clustering in a System of Self - Driven Particles . Physical Review Letters, 92 0 (16): 0 168701, April 2004

    2004

  25. [33]

    Interacting particle solutions of Fokker - Planck equations through gradient-log-density estimation

    Dimitra Maoutsa, Sebastian Reich, and Manfred Opper. Interacting particle solutions of Fokker - Planck equations through gradient-log-density estimation. Entropy, 22 0 (8): 0 802, July 2020. ISSN 1099-4300

    2020

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.