pith. sign in

arxiv: 2604.23952 · v1 · submitted 2026-04-27 · 📊 stat.ML · cs.LG· nlin.CD

Conditional Score-Based Modeling of Effective Langevin Dynamics

Ludovico T. Giorgini This is my paper

Pith reviewed 2026-05-08 01:15 UTC · model grok-4.3

classification 📊 stat.ML cs.LGnlin.CD
keywords stochastic reduced-order modelsconditional scoreeffective Langevin dynamicsdata-driven calibrationfinite-lag statisticsdrift diffusion estimationnonequilibrium systems
0
0 comments X p. Extension

The pith

An identity connects conditional scores of transition densities to drift and diffusion coefficients in reduced stochastic models.

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

The paper presents a method to calibrate stochastic reduced-order models by deriving an identity that expresses lagged correlation derivatives as expectations involving the conditional score and the model coefficients. This turns the problem into a least-squares fit using only stationary lagged data pairs. The approach avoids differentiating trajectories, partitioning states, or simulating candidate models repeatedly. Validation on tractable diffusions shows that the inferred models match invariant statistics and finite-lag correlations.

Core claim

Derivatives of lagged correlation functions equal stationary expectations over observed lagged pairs of the conditional score times the unknown drift and diffusion coefficients, allowing these coefficients to be recovered by least-squares fitting to finite-lag statistics.

What carries the argument

The conditional score, which is the gradient of the log of the finite-time transition density with respect to the starting state, used to link model coefficients to data expectations.

If this is right

  • The drift and diffusion can be constrained directly from finite-lag statistics.
  • Calibration requires no differentiation of trajectories.
  • No state-space partitioning is needed.
  • The method works for coarse temporal sampling and unevenly sampled data.
  • Inferred models preserve invariant statistics and reproduce dynamical correlations.

Where Pith is reading between the lines

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

  • Such score-based identities may generalize to other types of reduced dynamics beyond Langevin form.
  • Application to high-dimensional systems could reduce computational cost in model learning.

Load-bearing premise

The complex system's effective dynamics are faithfully represented by a stochastic reduced-order model with drift and diffusion coefficients related to the conditional score via the derived identity.

What would settle it

Generate data from a known reduced Langevin model with given coefficients, apply the method to infer coefficients from lagged pairs, and check if the inferred values recover the original coefficients within statistical error.

Figures

Figures reproduced from arXiv: 2604.23952 by Ludovico T. Giorgini.

Figure 1
Figure 1. Figure 1: Reference and learned score-based mobility fields for the two-dimensional affine multiplicative-noise view at source ↗
Figure 2
Figure 2. Figure 2: Forward validation for the learned reduced Langevin model. The black curve is the reference process, view at source ↗
Figure 3
Figure 3. Figure 3: Lagged observable correlations for the 18 channels used to constrain the mobility fit. view at source ↗
read the original abstract

Stochastic reduced-order models are widely used to represent the effective dynamics of complex systems, but estimating their drift and diffusion coefficients from data remains challenging. Standard approaches often rely on short-time trajectory increments, state-space partitioning, or repeated simulation of candidate models, which become unreliable or computationally expensive for high-dimensional systems, coarse temporal sampling, or unevenly sampled data. We introduce a data-driven calibration method based on a novel relationship between the coefficients of a stochastic reduced model and the conditional score of the finite-time transition density, defined as the gradient of the logarithm of the transition density with respect to the initial state. The resulting identity expresses derivatives of lagged correlation functions as stationary expectations over observed lagged pairs involving this conditional score and the unknown model coefficients. This formulation allows the drift and diffusion structure to be constrained directly from finite-lag statistics, without differentiating trajectories, partitioning state space, or repeatedly integrating candidate reduced models during calibration, yielding a least-squares fitting problem over stationary lagged pairs. We validate the approach on analytically tractable and data-driven nonequilibrium diffusions, demonstrating that the inferred models preserve the invariant statistics while accurately reproducing finite-lag dynamical correlations. The framework provides a scalable route for learning stochastic reduced-order models from data that reproduce prescribed statistical and dynamical properties.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 3 minor

Summary. The paper introduces a conditional score-based method for calibrating stochastic reduced-order models (SROMs) of effective Langevin dynamics. It derives an identity linking the conditional score of the finite-time transition density to the drift and diffusion coefficients of an Itô SDE, allowing the coefficients to be fitted via least-squares minimization over derivatives of lagged correlation functions computed from stationary lagged data pairs. This avoids common pitfalls like trajectory differentiation or repeated model simulations. The method is validated on analytically tractable diffusions and data-driven nonequilibrium systems, showing preservation of invariant statistics and reproduction of dynamical correlations.

Significance. If the derived identity holds exactly for the reduced-order model and the validation generalizes, this offers a scalable alternative to existing SROM calibration techniques, especially for high-dimensional systems or coarsely sampled data. It connects score-based concepts from generative modeling to direct parameter estimation in stochastic dynamics without requiring state-space partitioning or repeated forward integrations. The approach is credited for yielding an exact identity (no detected circularity) that sets up a well-posed least-squares problem over observable lagged pairs, and for demonstrating both statistical invariance and finite-lag correlation reproduction in the reported cases.

minor comments (3)
  1. The abstract and introduction would benefit from an explicit statement of the assumed form of the reduced-order Itô SDE (e.g., state-dependent drift and diffusion) and the precise definition of the conditional score to aid readers in following the identity derivation.
  2. Validation sections should include quantitative error metrics (e.g., relative L2 errors on correlation functions) and details on data exclusion or train/test splits for the data-driven cases to strengthen reproducibility claims.
  3. Notation for lagged pairs and stationary expectations could be standardized with a single equation reference early in the methods to improve clarity across the derivation and numerical results.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive summary of our work and for recommending minor revision. The referee's description accurately reflects the core contribution: the derivation of an identity relating the conditional score of the finite-time transition density to the drift and diffusion of the reduced-order Itô SDE, which converts calibration into a least-squares problem over stationary lagged pairs. No specific major comments were raised in the report.

Circularity Check

0 steps flagged

No significant circularity detected in the derivation chain

full rationale

The paper derives a novel identity that relates derivatives of lagged correlation functions to stationary expectations involving the conditional score of the finite-time transition density and the unknown drift/diffusion coefficients of the reduced-order Itô model. This identity is then used to formulate an independent least-squares problem over observed lagged pairs from data. No step reduces by construction to a prior fit, self-citation, or renamed empirical pattern; the identity serves as an external bridge from the conditional score to the calibration procedure without forcing the coefficients. The derivation remains self-contained against the stated assumptions of faithful reduced-model representation and availability of stationary pairs, with validation on analytically tractable cases confirming independence from circular reductions.

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 0 invented entities

The central claim rests on the existence of a well-defined conditional score for the finite-time transition density and on the assumption that the target system admits an effective stochastic reduced-order model whose coefficients can be recovered via the derived identity. No explicit free parameters or invented entities are introduced in the abstract beyond the fitted drift and diffusion themselves.

axioms (3)
  • domain assumption The complex system possesses effective dynamics that can be represented by a stochastic reduced-order model with drift and diffusion coefficients.
    Invoked to justify the target of calibration.
  • domain assumption The finite-time transition density exists, is positive, and is differentiable with respect to the initial state so that the conditional score is well-defined.
    Required for the novel identity relating the score to model coefficients.
  • domain assumption Stationary lagged pairs can be sampled from data and used to estimate the relevant expectations.
    Underpins the least-squares formulation over observed pairs.

pith-pipeline@v0.9.0 · 5519 in / 1637 out tokens · 61861 ms · 2026-05-08T01:15:51.450072+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

68 extracted references · 49 canonical work pages

  1. [1]

    Memory effects in irreversible thermodynamics.Phys

    Robert Zwanzig. Memory effects in irreversible thermodynamics.Physical Review, 124(4): 983–992, 1961. doi: 10.1103/PhysRev.124.983. 30 Figure S1: Training residual channels for the two-dimensional affine multiplicative-noise benchmark. The black curves are the data-derived residual targets, the orange dashed curves are ex-post evaluations of the same empi...

    work page doi:10.1103/physrev.124.983 1961

  2. [2]

    Transport, collective motion, and brownian motion

    Hazime Mori. Transport, collective motion, and Brownian motion.Progress of Theoretical Physics, 33(3):423–455, 1965. doi: 10.1143/PTP.33.423

    work page doi:10.1143/ptp.33.423 1965

  3. [3]

    Chorin, Ole H

    Alexandre J. Chorin, Ole H. Hald, and Raz Kupferman. Optimal prediction and the mori– zwanzig representation of irreversible processes.Proceedings of the National Academy of Sciences of the United States of America, 97(7):2968–2973, 2000. doi: 10.1073/pnas.97.7.2968

    work page doi:10.1073/pnas.97.7.2968 2000

  4. [4]

    Extracting macroscopic dynamics: Model problems and algorithms.Nonlinearity, 17(6):R55–R127, 2004

    Dror Givon, Raz Kupferman, and Andrew Stuart. Extracting macroscopic dynamics: Model problems and algorithms.Nonlinearity, 17(6):R55–R127, 2004. doi: 10.1088/0951-7715/17/6/ R01

    work page doi:10.1088/0951-7715/17/6/ 2004

  5. [5]

    Disentangling multi-level systems: Averaging, correlations and memory.Journal of Statistical Mechanics: Theory and Experiment, 2012(03):P03003, 2012

    Jeroen Wouters and Valerio Lucarini. Disentangling multi-level systems: Averaging, correlations and memory.Journal of Statistical Mechanics: Theory and Experiment, 2012(03):P03003, 2012. doi: 10.1088/1742-5468/2012/03/P03003

    work page doi:10.1088/1742-5468/2012/03/p03003 2012

  6. [6]

    Multi-level dynamical systems: Connecting the Ruelle response theory and the Mori–Zwanzig approach.Journal of Statistical Physics, 151(5):850–860,

    Jeroen Wouters and Valerio Lucarini. Multi-level dynamical systems: Connecting the Ruelle response theory and the Mori–Zwanzig approach.Journal of Statistical Physics, 151(5):850–860,

  7. [7]

    doi: 10.1007/s10955-013-0726-8

    work page doi:10.1007/s10955-013-0726-8

  8. [8]

    Springer Science & Business Media, 2008

    Grigorios A Pavliotis and Andrew Stuart.Multiscale methods: averaging and homogenization, volume 53. Springer Science & Business Media, 2008

    2008

  9. [9]

    A. J. Majda, I. Timofeyev, and E. Vanden-Eijnden. Models for stochastic climate prediction. Proc. Natl. Acad. Sci. USA, 96(26):14687–14691, 1999. doi: 10.1073/pnas.96.26.14687

    work page doi:10.1073/pnas.96.26.14687 1999

  10. [10]

    A. J. Majda, I. Timofeyev, and E. Vanden-Eijnden. A mathematical framework for stochastic climate models.Communications on Pure and Applied Mathematics, 54(8):891–974, 2001. doi: 10.1002/cpa.1014

    work page doi:10.1002/cpa.1014 2001

  11. [11]

    1976 , title =

    Klaus Hasselmann. Stochastic climate models. part I. Theory.Tellus, 28(6):473–484, 1976. doi: 10.3402/tellusa.v28i6.11316

    work page doi:10.3402/tellusa.v28i6.11316 1976

  12. [12]

    Monthly Weather Review117(10), 2165–2185 (1989) https://doi.org/ 10.1175/1520-0493(1989)117⟨2165:RFAFUP⟩2.0.CO;2

    Cecile Penland. Random forcing and forecasting using principal oscillation pattern analysis. Monthly Weather Review, 117(10):2165–2185, 1989. doi: 10.1175/1520-0493(1989)117<2165: RFAFUP>2.0.CO;2

    work page doi:10.1175/1520-0493(1989)117 1989

  13. [13]

    Sardeshmukh

    Cécile Penland and Prashant D. Sardeshmukh. The optimal growth of tropical sea surface temperature anomalies.Journal of Climate, 8(8):1999–2024, 1995. doi: 10.1175/1520-0442(1995) 008<1999:TOGOTS>2.0.CO;2

    work page doi:10.1175/1520-0442(1995 1999

  14. [14]

    Kondrashov, M.D

    D. Kondrashov, M.D. Chekroun, and M. Ghil. Data-driven non-markovian closure models. Physica D: Nonlinear Phenomena, 297:33–55, 2015. ISSN 0167-2789. doi: 10.1016/j.physd.2014. 12.005. URLhttps://www.sciencedirect.com/science/article/pii/S0167278914002413

    work page doi:10.1016/j.physd.2014 2015

  15. [15]

    2011 Stochastic climate dynamics: Random attractors and time-dependent invariant measures.Physica D: Nonlinear Phenomena240, 1685–1700

    Mickaël D. Chekroun, Eric Simonnet, and Michael Ghil. Stochastic climate dynamics: Random attractors and time-dependent invariant measures.Physica D: Nonlinear Phenomena, 240(21): 1685–1700, 2011. doi: 10.1016/j.physd.2011.06.005

    work page doi:10.1016/j.physd.2011.06.005 2011

  16. [16]

    Lucarini, M

    V. Lucarini and M.D. Chekroun. Theoretical tools for understanding the climate crisis from Hasselmann’s programme and beyond.Nature Reviews Physics, 5(12):744–765, 2023. doi: 10.1038/s42254-023-00650-8. URLhttps://doi.org/10.1038/s42254-023-00650-8. 32

    work page doi:10.1038/s42254-023-00650-8 2023

  17. [17]

    Nonequilibrium statistical mechanics near equilibrium: Computing higher-order terms.Nonlinearity, 11(1):5–18, 1998

    David Ruelle. Nonequilibrium statistical mechanics near equilibrium: Computing higher-order terms.Nonlinearity, 11(1):5–18, 1998. doi: 10.1088/0951-7715/11/1/002

    work page doi:10.1088/0951-7715/11/1/002 1998

  18. [18]

    2009 A review of linear response theory for general differentiable dynamical systems.Nonlinearity22, 855–870

    David Ruelle. A review of linear response theory for general differentiable dynamical systems. Nonlinearity, 22(4):855–870, 2009. doi: 10.1088/0951-7715/22/4/009

    work page doi:10.1088/0951-7715/22/4/009 2009

  19. [19]

    Probing forced responses and causality in data-driven climate emulators: Conceptual limitations and the role of reduced-order models.Physical Review Research, 7(4): 043314, 2025

    Fabrizio Falasca. Probing forced responses and causality in data-driven climate emulators: Conceptual limitations and the role of reduced-order models.Physical Review Research, 7(4): 043314, 2025. doi: 10.1103/2f4r-k8lr. URLhttps://link.aps.org/doi/10.1103/2f4r-k8lr

    work page doi:10.1103/2f4r-k8lr 2025

  20. [20]

    Data-driven dimensionality reduction and causal inference for spatiotemporal climate fields.Physical Review E, 109(4):044202, 2024

    Fabrizio Falasca, Pavel Perezhogin, and Laure Zanna. Data-driven dimensionality reduction and causal inference for spatiotemporal climate fields.Physical Review E, 109(4):044202, 2024. doi: 10.1103/PhysRevE.109.044202

    work page doi:10.1103/physreve.109.044202 2024

  21. [21]

    Causally constrained reduced-order neural models of complex turbulent dynamical systems, 2026

    Fabrizio Falasca and Laure Zanna. Causally constrained reduced-order neural models of complex turbulent dynamical systems, 2026

    2026

  22. [23]

    Singleton

    Darrell Duffie and Kenneth J. Singleton. Simulated moments estimation of markov models of asset prices.Econometrica, 61(4):929–952, 1993. doi: 10.2307/2951768

    work page doi:10.2307/2951768 1993

  23. [24]

    Indirect inference.Journal of Applied Econometrics, 8(S1):S85–S118, 1993

    Christian Gouriéroux, Alain Monfort, and Eric Renault. Indirect inference.Journal of Applied Econometrics, 8(S1):S85–S118, 1993. doi: 10.1002/jae.3950080507

    work page doi:10.1002/jae.3950080507 1993

  24. [25]

    Gutmann and Jukka Corander

    Michael U. Gutmann and Jukka Corander. Bayesian optimization for likelihood-free inference of simulator-based statistical models.Journal of Machine Learning Research, 17(125):1–47, 2016

    2016

  25. [26]

    Description of a turbulent cascade by a Fokker–Planck equation.Physical Review Letters, 78(5):863–866, 1997

    Rudolf Friedrich and Joachim Peinke. Description of a turbulent cascade by a Fokker–Planck equation.Physical Review Letters, 78(5):863–866, 1997. doi: 10.1103/PhysRevLett.78.863

    work page doi:10.1103/physrevlett.78.863 1997

  26. [27]

    Siegert, R

    S. Siegert, R. Friedrich, and J. Peinke. Analysis of data sets of stochastic systems.Physics Letters A, 243(5–6):275–280, 1998. doi: 10.1016/S0375-9601(98)00283-7

    work page doi:10.1016/s0375-9601(98)00283-7 1998

  27. [28]

    Indispensable finite time corrections for Fokker–Planck equations from time series data.Physical Review Letters, 87:254501, 2001

    Mario Ragwitz and Holger Kantz. Indispensable finite time corrections for Fokker–Planck equations from time series data.Physical Review Letters, 87:254501, 2001. doi: 10.1103/ PhysRevLett.87.254501

    2001

  28. [29]

    Non-gaussian stochastic dynamical model for the el niño southern oscillation.Physical Review Research, 4(2):L022065, 2022

    Ludovico T Giorgini, W Moon, N Chen, and J Wettlaufer. Non-gaussian stochastic dynamical model for the el niño southern oscillation.Physical Review Research, 4(2):L022065, 2022. doi: 10.1103/PhysRevResearch.4.L022065

    work page doi:10.1103/physrevresearch.4.l022065 2022

  29. [30]

    Stochastic paleoclimatology: Modeling the epica ice core climate records.Chaos, 33(9):093132, 2023

    N D Keyes, L T Giorgini, and J S Wettlaufer. Stochastic paleoclimatology: Modeling the epica ice core climate records.Chaos, 33(9):093132, 2023. doi: 10.1063/5.0128814

    work page doi:10.1063/5.0128814 2023

  30. [31]

    Contribution of vi- brational excited molecular nitrogen to ammonia synthesis using an atmospheric-pressure plasma jet,

    Jonah Botvinick-Greenhouse, Robert Martin, and Yunan Yang. Learning dynamics on invariant measures using pde-constrained optimization.Chaos, 33(6):063152, 2023. doi: 10.1063/5. 0149673

    work page doi:10.1063/5 2023

  31. [32]

    Invariant measures in time-delay coordinates for unique dynamical system identification.Physical Review Letters, 135(16):167202,

    Jonah Botvinick-Greenhouse, Robert Martin, and Yunan Yang. Invariant measures in time-delay coordinates for unique dynamical system identification.Physical Review Letters, 135(16):167202,

  32. [33]

    doi: 10.1103/ppys-lx68. 33

    work page doi:10.1103/ppys-lx68

  33. [34]

    Chekroun, and Michael Ghil

    Manuel Santos Gutiérrez, Valerio Lucarini, Mickaël D. Chekroun, and Michael Ghil. Reduced- order models for coupled dynamical systems: Data-driven methods and the koopman operator. Chaos, 31(5):053116, 2021. doi: 10.1063/5.0039496

    work page doi:10.1063/5.0039496 2021

  34. [35]

    A fluctuation-dissipation theorem perspective on radiative responses to temperature perturbations.Journal of Climate, 38(14):3239–3260, 2025

    Fabrizio Falasca, Aurora Basinski-Ferris, Laure Zanna, and Ming Zhao. A fluctuation-dissipation theorem perspective on radiative responses to temperature perturbations.Journal of Climate, 38(14):3239–3260, 2025. doi: 10.1175/JCLI-D-24-0479.1. URLhttps://doi.org/10.1175/ JCLI-D-24-0479.1

    work page doi:10.1175/jcli-d-24-0479.1 2025

  35. [36]

    Giorgini, Andre N

    Ludovico T. Giorgini, Andre N. Souza, Domenico Lippolis, Predrag Cvitanović, and Peter Schmid. Learning dissipation and instability fields from chaotic dynamics.Physica D: Nonlinear Phenomena, 481:134865, 2025. ISSN 0167-2789. doi: 10.1016/j.physd.2025.134865. URL https://www.sciencedirect.com/science/article/pii/S0167278925003422

    work page doi:10.1016/j.physd.2025.134865 2025

  36. [37]

    Reduced markovian models of dynamical systems.Physica D: Nonlinear Phenomena, 470:134393, 2024

    Ludovico T Giorgini, Andre N Souza, and Peter J Schmid. Reduced markovian models of dynamical systems.Physica D: Nonlinear Phenomena, 470:134393, 2024. doi: 10.1016/j.physd. 2024.134393

    work page doi:10.1016/j.physd 2024

  37. [38]

    Representing turbulent statistics with partitions of state space

    Andre N Souza. Representing turbulent statistics with partitions of state space. part 1. theory and methodology.Journal of Fluid Mechanics, 997:A1, 2024. doi: 10.1017/jfm.2024.658

    work page doi:10.1017/jfm.2024.658 2024

  38. [39]

    Black, and George Ilhwan Park

    Andre N Souza. Representing turbulent statistics with partitions of state space. part 2. the compressible euler equations.Journal of Fluid Mechanics, 997:A2, 2024. doi: 10.1017/jfm.2024. 657

    work page doi:10.1017/jfm.2024 2024

  39. [40]

    Souza and Simone Silvestri

    Andre N. Souza and Simone Silvestri. A modified bisecting k-means for approximating transfer operators: Application to the lorenz equations, 2024

    2024

  40. [41]

    Takens, Detecting strange attractors in turbulence, in: Dynamical Systems and Turbulence, Vol

    Floris Takens. Detecting strange attractors in turbulence. In D. A. Rand and L.-S. Young, editors,Dynamical Systems and Turbulence, Warwick 1980, volume 898 ofLecture Notes in Mathematics, pages 366–381. Springer, Berlin, Heidelberg, 1981. doi: 10.1007/BFb0091924

    work page doi:10.1007/bfb0091924 1980

  41. [42]

    Estimation of non-normalized statistical models by score matching.Journal of Machine Learning Research, 6:695–709, 2005

    Aapo Hyvärinen. Estimation of non-normalized statistical models by score matching.Journal of Machine Learning Research, 6:695–709, 2005

    2005

  42. [43]

    A connection between score matching and denoising autoencoders.Neural computation, 23(7):1661–1674, 2011

    Pascal Vincent. A connection between score matching and denoising autoencoders.Neural Computation, 23(7):1661–1674, 2011. doi: 10.1162/NECO_a_00142

    work page doi:10.1162/neco_a_00142 2011

  43. [44]

    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. In International Conference on Learning Representations, 2021

    2021

  44. [45]

    Giorgini, Tobias Bischoff, and Andre N

    Ludovico T. Giorgini, Tobias Bischoff, and Andre N. Souza. Kgmm: A k-means clustering approach to gaussian mixture modeling for score function estimation, 2025

    2025

  45. [46]

    R. Kubo. The fluctuation-dissipation theorem.Reports on Progress in Physics, 29(1):255–284,

  46. [47]

    doi: 10.1088/0034-4885/29/1/306

    work page doi:10.1088/0034-4885/29/1/306

  47. [48]

    U. M. B. Marconi, A. Puglisi, L. Rondoni, and A. Vulpiani. Fluctuation-dissipation: Response theory in statistical physics.Physics Reports, 461(4-6):111–195, 2008. doi: 10.1016/j.physrep. 2008.02.002. 34

    work page doi:10.1016/j.physrep 2008

  48. [49]

    A. J. Majda, B. Gershgorin, and Y. Yuan. Low-frequency climate response and fluctuation– dissipation theorems: Theory and practice.Journal of the Atmospheric Sciences, 67:1186–1201, 2010

    2010

  49. [50]

    Majda, R

    A.J. Majda, R. Abramov, and B. Gershgorin. High skill in low-frequency climate response through fluctuation dissipation theorems despite structural instability.Proceedings of the National Academy of Sciences, 107(2):581–586, 2010. doi: 10.1073/pnas.0912997107

    work page doi:10.1073/pnas.0912997107 2010

  50. [51]

    A. J. Majda, R. V. Abramov, and M. J. Grote.Information Theory and Stochastics for Multiscale Nonlinear Systems. CRM Monograph Series, American Mathematical Society, 2005

    2005

  51. [52]

    Gritsun and G

    A. Gritsun and G. Branstator. Climate response using a three-dimensional operator based on the fluctuation–dissipation theorem.Journal of the Atmospheric Sciences, pages 2558–2575,

  52. [53]

    doi: 10.1175/JAS3943.1

    work page doi:10.1175/jas3943.1

  53. [54]

    Climate sensitivity via a nonparametric fluctuation– dissipation theorem.Journal of the Atmospheric Sciences, 68(5):937–953, 2011

    Fenwick C Cooper and Peter H Haynes. Climate sensitivity via a nonparametric fluctuation– dissipation theorem.Journal of the Atmospheric Sciences, 68(5):937–953, 2011

    2011

  54. [55]

    Response theory via generative score modeling.Physical Review Letters, 133(26):267302, 2024

    Ludovico Theo Giorgini, Katherine Deck, Tobias Bischoff, and Andre Souza. Response theory via generative score modeling.Physical Review Letters, 133(26):267302, 2024. doi: 10.1103/ PhysRevLett.133.267302

    2024

  55. [56]

    Giorgini, Fabrizio Falasca, and Andre N

    Ludovico T. Giorgini, Fabrizio Falasca, and Andre N. Souza. Predicting forced responses of probability distributions via the fluctuation-dissipation theorem and generative modeling. Proceedings of the National Academy of Sciences of the United States of America, 122(41): e2509578122, 2025. doi: 10.1073/pnas.2509578122

    work page doi:10.1073/pnas.2509578122 2025

  56. [57]

    Giorgini, Tobias Bischoff, and Andre N

    Ludovico T. Giorgini, Tobias Bischoff, and Andre N. Souza. Statistical parameter calibration with the generalized fluctuation dissipation theorem and generative modeling, 2025

    2025

  57. [58]

    Giorgini, Score-Based Modeling of Effective Langevin Dynamics, Arxiv https://doi.org/10.48550/arXiv.2505.01895 (2025)

    Ludovico Theo Giorgini. Score-based modeling of effective langevin dynamics, 2025. URL https://arxiv.org/abs/2505.01895

    work page arXiv 2025

  58. [59]

    Integrating score-based generative modeling and neural odes for accurate representation of multiscale chaotic dynamics.arXiv preprint arXiv:2511.03862, 2025

    Giulio Del Felice and Ludovico Theo Giorgini. Integrating score-based generative modeling and neural odes for accurate representation of multiscale chaotic dynamics.arXiv preprint arXiv:2511.03862, 2025

    work page arXiv 2025

  59. [60]

    Reduced-order modeling of cyclo-stationary time series using score-based generative methods, 2025

    Ludovico Theo Giorgini, Tobias Bischoff, and Andre Nogueira Souza. Reduced-order modeling of cyclo-stationary time series using score-based generative methods, 2025

    2025

  60. [61]

    Bernard O. Koopman. Hamiltonian systems and transformation in hilbert space.Proceedings of the National Academy of Sciences of the United States of America, 17(5):315–318, 1931. doi: 10.1073/pnas.17.5.315

    work page doi:10.1073/pnas.17.5.315 1931

  61. [62]

    Nonlinear Dynamics 41, 309–325

    Igor Mezić. Spectral properties of dynamical systems, model reduction and decompositions. Nonlinear Dynamics, 41(1–3):309–325, 2005. doi: 10.1007/s11071-005-2824-x

    work page doi:10.1007/s11071-005-2824-x 2005

  62. [63]

    Budiˇ si´ c, R

    Marko Budišić, Ryan Mohr, and Igor Mezić. Applied koopmanism.Chaos: An Interdisciplinary Journal of Nonlinear Science, 22(4):047510, 2012. doi: 10.1063/1.4772195

    work page doi:10.1063/1.4772195 2012

  63. [64]

    and Kevrekidis, Ioannis G

    Matthew O. Williams, Ioannis G. Kevrekidis, and Clarence W. Rowley. A data–driven ap- proximation of the koopman operator: Extending dynamic mode decomposition.Journal of Nonlinear Science, 25(6):1307–1346, 2015. doi: 10.1007/s00332-015-9258-5. 35

    work page doi:10.1007/s00332-015-9258-5 2015

  64. [65]

    Data-driven model reduction and transfer operator approximation.Journal of Nonlinear Science, 28:985–1010, 2018

    Stefan Klus, Feliks Nüske, Péter Koltai, Hao Wu, Ioannis Kevrekidis, Christof Schütte, and Frank Noé. Data-driven model reduction and transfer operator approximation.Journal of Nonlinear Science, 28:985–1010, 2018. doi: 10.1007/s00332-017-9437-7

    work page doi:10.1007/s00332-017-9437-7 2018

  65. [66]

    Colbrook, Valerio Lucarini, Igor Mezić, and John Moroney

    Niccolò Zagli, Matthew J. Colbrook, Valerio Lucarini, Igor Mezić, and John Moroney. Bridging the gap between koopmanism and response theory: Using natural variability to predict forced response.SIAM Journal on Applied Dynamical Systems, 25(1):196–229, 2026. doi: 10.1137/ 24M1699206

    2026

  66. [67]

    Interpretable and equation-free response theory: From fingerprinting to forcing design.Chaos, Solitons & Fractals, 202:117540, 2026

    Valerio Lucarini, Manuel Santos Gutiérrez, John Moroney, and Niccolò Zagli. Interpretable and equation-free response theory: From fingerprinting to forcing design.Chaos, Solitons & Fractals, 202:117540, 2026. doi: 10.1016/j.chaos.2025.117540

    work page doi:10.1016/j.chaos.2025.117540 2026

  67. [68]

    C., INGERSOLL, J

    John C. Cox, Jonathan E. Ingersoll, and Stephen A. Ross. A theory of the term structure of interest rates.Econometrica, 53(2):385–407, 1985. doi: 10.2307/1911242

    work page doi:10.2307/1911242 1985

  68. [69]

    Feller ,\ title title Two singular diffusion problems , \ https://doi.org/10.2307/1969318 journal journal Annals of Mathematics \ volume 54 ,\ pages 173--182 ( year 1951 ) NoStop

    William Feller. Two singular diffusion problems.Annals of Mathematics, 54(1):173–182, 1951. doi: 10.2307/1969318. 36

    work page doi:10.2307/1969318 1951