Pith. sign in

REVIEW 2 major objections 6 minor 2 cited by

Scale-Adaptive Generative Flows for Multiscale Scientific Data

T0 review · 2 major / 6 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Flow-based generative models of multiscale scientific fields require noise whose Fourier spectrum decays no faster than the target spectrum; otherwise the drift becomes unbounded near t=0, and a scale-adaptive schedule restores numerical ef

desk verdict Worth a real read: clean Gaussian analysis and a useful schedule, with the Navier-Stokes win resting on a hand-tuned parameter and a loose non-Gaussian bridge. read the letter →

arxiv 2509.02971 v2 pith:QRSTELTZ submitted 2025-09-03 stat.ML cs.LGcs.NAmath.NAmath.PR

Yifan Chen , Eric Vanden-Eijnden This is my paper · ORCID
classification stat.MLcs.LGcs.NAmath.NAmath.PR MSC 60H1562M4068T07
keywords stochasticinterpolantsmultiscaleFourierspectranoiseroughnessCameron–MartinspaceLipschitzdriftscale-adaptiveschedulesfunction-spacegenerativemodelsNavier–Stokesinvariantmeasure
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

Flow-based generative models build a field sample by integrating a learned drift from noise into data. The paper's core point is that for scientific data with wide Fourier spectra, the noise must not be smoother than the target: if the noise covariance decays faster in Fourier space than the data covariance, the drift operator becomes unbounded near the initial time and the generative ODE is ill-conditioned as resolution grows. For Gaussian targets this is a precise operator condition; for general targets, compact support in the noise's Cameron–Martin space guarantees a bounded, Lipschitz drift. When the correct roughness is known, matching the noise spectrum to the target spectrum cuts the number of integration steps needed to reproduce fine-scale spectra. When the target is too non-Gaussian to be matched by Gaussian noise, a scalar scale-adaptive schedule keeps the drift's Lipschitz constant at half the log of the spectral ratio and yields accurate Navier–Stokes enstrophy spectra with only 10 RK4 steps.

What carries the argument

The central object is the stochastic interpolant and its conditional-expectation drift b_t(x)=E[İ_t | I_t=x], which defines a generative ODE whose law matches the data law at t=1. Gaussian analysis reduces it to the operator B(t)=(α̇α C0 + β̇β C1)(α^2 C0 + β^2 C1)^{-1}; the ratio C1 C0^{-1} controls whether the drift stays bounded near t=0, and the Cameron–Martin space of the noise plays the analogous role for general targets. The other load-bearing piece is the scalar scale-adaptive schedule (5.1), α_t = sqrt((μ*−μ*^t)/(μ*−1)), β_t = sqrt((μ*^t−1)/(μ*−1)), built from the smallest eigenvalue ratio μ* of C1 C0^{-1}; it makes the drift's Lipschitz constant 1/2 |log μ*| instead of growing polyn

What would settle it

For a Matérn target with regularity s1 and noise with lower roughness s0 < s1, Proposition 3.1 predicts ∥B(t)∥ ~ 1/t near t=0; measuring the minimum RK4 step size needed to keep the generated energy spectrum within a fixed tolerance at N=32, 64, and 128 should show a clear resolution-dependent blow-up if the claim is right. Conversely, if a Gaussian-noise generative model with s0 > s1 still produces accurate fine-scale spectra with a fixed small number of steps, the drift-blowup mechanism is not the controlling factor.

Watch

Extended reading notes

Core claim

Working in the stochastic interpolant picture, the paper establishes a roughness-matching principle: for the interpolant I_t = α_t z + β_t x_1, the drift b_t(x)=E[İ_t | I_t=x] is well posed only if the noise z is at least as rough as the target x_1. For Gaussian measures this is exact (Proposition 3.1): b_t(x)=B(t)x with B(t)=(α̇α C0 + β̇β C1)(α^2 C0 + β^2 C1)^{-1}, and an unbounded C1 C0^{-1} forces ∥B(t)∥→∞ as t→0. For general targets, compact support of the data in the noise's Cameron–Martin space gives a bounded, Lipschitz drift (Proposition 3.3). The paper then shows two regimes: matched-spectrum noise reproduces fine scales with very few steps when the target's spectrum is known, while

Load-bearing premise

The general Lipschitz-drift guarantee assumes the target distribution is compactly supported inside the noise's Cameron–Martin space; when that assumption fails, as it does for the Navier–Stokes invariant measure under matched-spectrum noise, the paper falls back on rougher noise plus a schedule whose scale parameter μ* is fitted from the observed data spectrum rather than derived.

Editorial extensions

If this is right

  • White noise is too smooth for smooth data: using it as the noise in a flow model on a Matérn-like field makes the drift unbounded as t→0, so errors at fine scales cannot be controlled as resolution increases.
  • When the data's fine-scale Fourier structure is known, noise with a matched spectrum reaches accurate spectra with as few as 5 RK4 steps, while white noise needs 20–80 steps and still degrades on finer grids.
  • The well-posedness criterion transfers to general targets: if the data lives compactly inside the noise's Cameron–Martin space, the drift is bounded and Lipschitz on [0,1−δ].
  • For rougher-than-data noise, the scale-adaptive schedule (5.1) caps the drift Lipschitz constant at 1/2 |log μ*|, turning a polynomially stiff integration problem into a logarithmically mild one.
  • Using white noise plus the designed schedule with μ* read from the spectrum gives accurate Navier–Stokes enstrophy spectra at 128×128 with only 10 RK4 steps.

Reading between the lines

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

  • The logarithmic dependence on μ* suggests the schedule's benefit should persist at higher resolution and in 3D, where the smallest resolved eigenvalue ratio shrinks only logarithmically; this is an extrapolation the paper does not run.
  • The heuristic that sets μ* from the ratio of the smallest target spectral value to the noise plateau could be automated, e.g. by estimating the spectrum and choosing μ* before training, rather than fixing it by hand.
  • The same Cameron–Martin reasoning may guide non-Gaussian noise families: choosing a noise whose Cameron–Martin space exactly contains the data support could combine the efficiency of spectrum-matched noise with the flexibility needed for non-Gaussian targets.
  • Integrated with per-scale or hierarchical modeling, the schedule could be applied separately to different wavenumber bands rather than globally, possibly improving accuracy on intermittent fields.
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

2 major / 6 minor

Summary. The paper studies noise distribution and interpolation schedule design within the stochastic interpolants / flow matching framework, targeting scientific data with multiscale Fourier spectra. For Gaussian targets, it shows that if the noise covariance C0 is smoother than the target covariance C1 in the sense that C1 C0^{-1} is unbounded, the drift field of the generative ODE has unbounded operator norm near t=0 (Prop. 3.1); conversely, matched-spectrum noise keeps the drift bounded and, with the proposed scalar schedule (5.1), gives a drift Jacobian bounded by (1/2)|log μ⋆| (Prop. 5.1). For general targets, a Lipschitz bound is stated under a compact-support assumption in the Cameron–Martin space of the noise (Prop. 3.3). Numerical experiments on Gaussian random fields, stochastic Allen–Cahn invariant measures, and stochastic Navier–Stokes invariant measures illustrate that spectrum-matched noise improves efficiency in the tractable cases, while for the Navier–Stokes case a scale-adaptive schedule with white noise improves the enstrophy spectrum with only 10 RK4 steps.

Significance. If the results hold, the paper makes a useful contribution by connecting noise roughness, drift regularity, and integration cost in flow-based generative models, and by giving an explicit, resolution-robust scalar schedule for Gaussian targets whose Lipschitz constant grows only logarithmically with the spectral range. The Gaussian derivations in Props. 3.1 and 5.1 are clean, and the release of code supports reproducibility. The function-space perspective and the demonstration that standard white noise can be inferior to spectrum-matched noise for smooth Gaussian/Allen–Cahn targets are valuable. However, the central practical claim for complex non-Gaussian targets such as Navier–Stokes is not backed by a quantitative theorem: Prop. 5.1 is Gaussian-only, while Prop. 3.3 only gives existence of a Lipschitz constant under assumptions that likely fail for the Navier–Stokes invariant measure. The Navier–Stokes experiment also uses a hand-fitted μ⋆ read from the data spectrum. These gaps limit the significance of the paper as a general theory, though they may be addressable by reframing and additional experiments.

major comments (2)
  1. [§5.2, Prop. 5.1, Prop. 3.3] The quantitative efficiency claim for the scale-adaptive schedule is proved only for Gaussian targets with C0 and C1 mutually diagonalizable (Prop. 5.1). In the Navier–Stokes experiment (Sec. 4.2.3 and 5.2), the target is non-Gaussian, the drift is a trained UNet rather than the linear Gaussian map, and μ⋆ = 10^{-5} is read off the data-to-noise enstrophy ratio at k=26. Prop. 3.3 supplies only existence of a Lipschitz constant on [0,1−δ] under compact support in the Cameron–Martin space V; it does not quantify the constant, and the compact-support assumption is not verified for the Navier–Stokes invariant measure. Indeed, the paper itself notes in Sec. 4.2.3 that the NS invariant measure is not in the Cameron–Martin space of the spectrum noise, and no analogous statement is made for white noise. Thus no stated theorem justifies applying the ½|log μ⋆| bound to the learned non-Gaussian dri
  2. [§5.2, Fig. 6] The parameter μ⋆ = 10^{-5} is selected directly from the data spectrum (the enstrophy ratio at the highest resolved mode). This is a fitted value, not a prediction, and Section 6 lists automated spectral estimation as future work. The observed 10-step RK4 improvement may be specific to this μ⋆, this network, and this noise normalization. To make the scale-adaptive claim credible, the paper should report sensitivity of the results to μ⋆ (e.g., varying it over several orders of magnitude) and, ideally, propose a data-driven estimator for μ⋆. Without this, the Navier–Stokes experiment is a demonstration on a hand-tuned parameter rather than a validation of a design principle.
minor comments (6)
  1. [§4.2.1, §4.2.2] In the experiment descriptions, the interpolant is written as It = αt z + βt z; the second factor should be βt x1 (or βt x0, depending on notation). This typo appears in both sections.
  2. [Appendix B, proof of Prop. 3.1] The displayed formula for Cov(˙I_t, I_t) Cov(I_t)^{-1} omits the C0 factors: it should be (˙α_t α_t C0 + ˙β_t β_t C1)(α_t^2 C0 + β_t^2 C1)^{-1}. The subsequent line is correct, but the intermediate expression is inconsistent.
  3. [Appendix C, Eq. (C.7)–(C.10)] The Lipschitz bound for the conditional expectation F is missing the factor β_t/α_t^2 that appears in (C.7). While this factor can be absorbed into the constant Cδ in (C.3), the displayed bound (C.10) is not correct as written and the proof should be adjusted.
  4. [§4.2.3] The statement that 'with more RK4 steps, the result using spectrum noise does not improve' is not supported by a figure or table. If this is an important negative result, it should be documented; otherwise it should be removed or qualified.
  5. [Prop. 5.1] The proposition assumes the eigenvalues of C1 C0^{-1} satisfy 1 ≥ μ1 ≥ ... ≥ μd. The paper should state what happens if the largest eigenvalue exceeds 1, since the schedule formula (5.1) and the monotonicity step in the proof rely on μ ≤ 1.
  6. [Figures 2–6] The energy/enstrophy spectra are averaged over ensembles, but no error bars or ensemble sizes are reported. Given that some differences between methods appear modest at low wavenumbers, error bars would strengthen the empirical claims.

Circularity Check

1 steps flagged · score 2.0 of 10

Core Gaussian and general-measure results are derived from first principles; only a minor, non-load-bearing self-citation and a hand-set NS schedule parameter keep this from a clean bill.

  1. other [Appendix B.2, final sentence after proof of Proposition 5.1]
    "In fact, the choice (5.1) minimizes the averaged squared 2-norm of the gradient over all βt; see [10]."

    This optimality claim is supported only by a self-citation to [10], a companion paper with two of the present authors (Chen, Vanden-Eijnden, Xu). However, the preceding calculation already proves the central quantitative consequence, ∥∇bt(x)∥₂ = ½|log μ⋆| for the schedule (5.1), entirely within this paper. The self-citation therefore adds an optimality gloss but is not load-bearing for the paper's main efficiency bound. It is a minor self-citation rather than a circular reduction.

full rationale

Proposition 3.1 is proven in Appendix B from Gaussian conditioning: the drift operator is computed explicitly, and the divergence lim_{t→0}‖B(t)‖=∞ follows from β_0=0 and β_t∈C¹. The Matérn example is then a direct application, not an input. Proposition 3.3 is proven in Appendix C using Cameron–Martin/Girsanov densities and a compact-support assumption; boundedness and Lipschitzness are derived, not assumed. Proposition 5.1 likewise computes the Jacobian norm of the drift for schedule (5.1) and obtains ½|log μ⋆| by direct algebra. The Navier-Stokes experiment in Sec. 5.2 sets μ⋆ = 10⁻⁵ from the data-to-noise enstrophy ratio at the finest resolved mode; this is a hand-tuned, spectrum-adaptive parameter, and Sec. 6 lists automated spectral estimation as future work. But the paper does not present μ⋆ as a predicted quantity, nor does it claim Proposition 5.1 applies verbatim to the non-Gaussian NS invariant measure; it explicitly says the schedule is “motivated by the Gaussian case.” The Gaussian and Allen-Cahn experiments use known covariances as noise, which is a demonstration rather than a circular prediction. The only noteworthy circularity-adjacent issue is the non-load-bearing self-citation to [10], yielding a score of 2.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The central method introduces no new physical entities. The only fitted free parameter is μ⋆ for the Navier-Stokes schedule. The main axioms are standard independence and smoothness assumptions, plus the compact-support-in-Cameron-Martin-space condition for the general theory.

free parameters (1)
  • μ⋆ (schedule parameter) = 1e-5
    In Sec. 5.2, μ⋆ is set to 10^-5 = 10^-4 / 10, the ratio of the data enstrophy at the finest scale to the white-noise level, read off the observed spectrum.
assumptions (5)
  • standard math z ⊥ x1 (independence of noise and data)
    Used throughout the stochastic interpolant construction in Definition 2.1 and Proposition 3.1.
  • standard math αt, βt ∈ C^1([0,1]) with α0=β1=1, α1=β0=0, ˙βt>0, ˙αt<0
    Standard boundary conditions and monotonicity for interpolant schedules, assumed throughout.
  • domain assumption µ∗ is compactly supported in the Cameron-Martin space V of the noise
    Proposition 3.3 requires ∥x1∥V ≤ R almost surely. The authors acknowledge this fails for Navier-Stokes with spectrum-matched noise (Sec. 4.2.3).
  • domain assumption C0 and C1 are mutually diagonalizable with eigenvalues of C1C0^{-1} in [µ⋆,1]
    Proposition 5.1 assumes this to derive the scalar schedule (5.1). It holds for the Matern-like example but not for general non-Gaussian targets.
  • standard math Cameron-Martin theorem for conditional expectation formula (C.2)
    The proof of Prop. 3.3 uses a Radon-Nikodym derivative and cites [43, Theorem 12] for rigor. The paper's derivation is a sketch.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Scale-Adaptive Generative Flows for Multiscale Scientific Data." pith.science (2026). https://pith.science/paper/QRSTELTZ

@misc{pith2026250902971,
  author       = {Pith},
  title        = {Pith review of: Scale-Adaptive Generative Flows for Multiscale Scientific Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QRSTELTZ}},
  note         = {Machine review of arXiv:2509.02971}
}
read the original abstract

Flow-based generative models can face numerical challenges on scientific data with multiscale Fourier spectra, often producing large errors at fine scales. We approach this problem within the flow matching and stochastic interpolants framework, through the principled design of noise distributions and interpolation schedules. Working in function space ensures that the generative model remains well defined as the resolution is refined; the Lipschitz regularity of the drift is important to both this function-space well-posedness and the integration cost at fixed resolution. The central observation is that the noise should be at least as rough as the target distribution -- measured by Fourier-spectrum decay -- in order to keep the Lipschitz constant finite. For Gaussian and near-Gaussian targets whose fine-scale structure is known, matched-spectrum noise improves numerical efficiency over standard white-noise choices. For more complex non-Gaussian targets, matched-spectrum noise may not be sufficient, and we propose scale-adaptive interpolation schedules to mitigate the terminal-time stiffness that arises when the noise is rougher than the data. Numerical experiments on synthetic Gaussian random fields and on invariant measures of the stochastic Allen--Cahn and Navier--Stokes equations illustrate the approach and demonstrate its ability to generate high-fidelity samples at lower computational cost than traditional approaches.

Figures

Figures reproduced from arXiv: 2509.02971 by the authors.

Figure 1
Figure 1. Examples of samples from 2D Mat´ern Gaussian measures (left panel) and the invariant measure of the stochastically forced Navier￾Stokes (middle panel) at a resolution of 128 ×128; the right panel shows their energy and enstrophy spectra. We develop strategies to address these numerical challenges through the principled design of generative flow processes within the framework of stochastic interpolants [2, 3], which … view at source ↗
Figure 2
Figure 2. Energy spectra of Gaussian fields: comparison between ground truth, spectrum noise generation (5 RK4 steps), and white noise generation (20, 40, or 80 RK4 steps). Left: 32 × 32 resolution; middle: 64 × 64; right: 128 × 128 [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Energy spectra of stochastic Allen-Cahn invariant distribu￾tions: comparison between ground truth, spectrum noise generation (5 RK4 steps), and white noise generation (10, 20, or 40 RK4 steps). Left: N = 32; middle: N = 64; right: N = 128. that remains robust across different resolutions, while white noise performance degrades and requires substantially more integration steps. 4.2.3. Failure example: Invariant distr… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Up: Three samples drawn from the invariant distribution of the stochastically forced Navier-Stokes. Down: three types of noises used for constructing generative models: Gaussian with the same spec￾trum behavior as the invariant distribution, Gaussian with a rougher spe…
Figure 5
Figure 5. Figure 5: Enstrophy spectrum of truth, noises, and generated samples using the three different types of noises in [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]
Figure 6
Figure 6. Figure 6: Experiments on stochastically forced Navier-Stokes using white noise. Left: generated samples using linear schedule; middle: generated samples using designed schedule; right: comparison of their estimated enstrophy spectra along with the truth and the noise. In all cas…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. Discretization and Statistical Consistency of Functional Flow Matching

    cs.LG 2026-08 accept novelty 7.0 of 10

    Finite conditional velocity targets in functional flow matching converge in L2 to the continuum target under nonnested, strongly consistent reconstructions, with end-to-end Wasserstein control of the generated laws.

  2. Telegrapher's Generative Model via Kac Flows

    math.AP 2025-06 conditional novelty 6.0 of 10

    Kac flows, driven by the telegrapher's equation, provide a generative model with an explicit conditional velocity field and bounded velocity norm.

Reference graph

Works this paper leans on

60 extracted references · 36 canonical work pages · cited by 2 Pith papers

  1. [1]

    Ballard, Joshua Bambrick, Sebastian W

    Josh Abramson, Jonas Adler, Jack Dunger, Richard Evans, Tim Green, Alexander Pritzel, Olaf Ronneberger, Lindsay Willmore, Andrew J. Ballard, Joshua Bambrick, Sebastian W. Bodenstein, David A. Evans, Chia-Chun Hung, Michael O’Neill, David Reiman, Kathryn Tunyasuvunakool, Zachary Wu, Akvil˙ e ˇZemgulyt˙ e, Eirini Arvaniti, Charles Beattie, Ottavia Bertolli,...

    work page 2024

  2. [2]

    Building normalizing flows with stochastic interpolants

    Michael Albergo and Eric Vanden-Eijnden. Building normalizing flows with stochastic interpolants. In ICLR 2023 Conference , 2023

    work page 2023

  3. [3]

    Stochastic interpolants: A unifying framework for flows and diffusions

    Michael S Albergo, Nicholas M Boffi, and Eric Vanden-Eijnden. Stochastic interpolants: A unifying framework for flows and diffusions. arXiv preprint arXiv:2303.08797 , 2023

    arXiv 2023

  4. [4]

    Condi- tional score-based diffusion models for bayesian inference in infinite dimensions.Advances in Neural Information Processing Systems, 36, 2024

    Lorenzo Baldassari, Ali Siahkoohi, Josselin Garnier, Knut Solna, and Maarten V de Hoop. Condi- tional score-based diffusion models for bayesian inference in infinite dimensions.Advances in Neural Information Processing Systems, 36, 2024

    work page 2024

  5. [5]

    Gaussian Measures

    Vladimir Igorevich Bogachev. Gaussian Measures . Number 62. American Mathematical Society, 1998

    work page 1998

  6. [6]

    $\infty$-Diff: Infinite Resolution Diffusion with Subsampled Mollified States

    Sam Bond-Taylor and Chris G Willcocks. ∞-diff: Infinite resolution diffusion with subsampled mollified states. arXiv preprint arXiv:2303.18242 , 2023

    work page Pith review arXiv 2023

  7. [7]

    Sampling is as easy as learning the score: Theory for diffusion models with minimal data assumptions

    Sitan Chen, Sinho Chewi, Jerry Li, Yuanzhi Li, Adil Salim, and Anru R Zhang. Sampling is as easy as learning the score: Theory for diffusion models with minimal data assumptions. In International Conference on Learning Representations, 2023. SCALE-ADAPTIVE GENERATIVE FLOWS FOR MULTISCALE SCIENTIFIC DATA 17

    work page 2023

  8. [8]

    New affine invariant ensemble samplers and their dimensional scaling

    Yifan Chen. New affine invariant ensemble samplers and their dimensional scaling. arXiv preprint arXiv:2505.02987, 2025

    arXiv 2025

Show all 60 references
  1. [9]

    Probabilistic forecasting with stochastic interpolants and F¨ ollmer processes

    Yifan Chen, Mark Goldstein, Mengjian Hua, Michael Samuel Albergo, Nicholas Matthew Boffi, and Eric Vanden-Eijnden. Probabilistic forecasting with stochastic interpolants and F¨ ollmer processes. In International Conference on Machine Learning , pages 6728–6756. PMLR, 2024

    2024

  2. [10]

    Lipschitz-guided design of interpolation schedules in generative models

    Yifan Chen, Eric Vanden-Eijnden, and Jiawei Xu. Lipschitz-guided design of interpolation schedules in generative models. arXiv preprint arXiv:2509.01629 , 2025

    2025 arXiv

  3. [11]

    Renormalization group flow as optimal transport

    Jordan Cotler and Semon Rezchikov. Renormalization group flow as optimal transport. Physical Review D, 108(2):025003, 2023

    2023

  4. [12]

    Renormalizing diffusion models

    Jordan Cotler and Semon Rezchikov. Renormalizing diffusion models. arXiv preprint arXiv:2308.12355, 2023

    2023 arXiv

  5. [13]

    Diffusion models beat GANs on image synthesis.Advances in Neural Information Processing Systems , 34:8780–8794, 2021

    Prafulla Dhariwal and Alexander Nichol. Diffusion models beat GANs on image synthesis.Advances in Neural Information Processing Systems , 34:8780–8794, 2021

    2021

  6. [14]

    Generative diffusion models in infinite dimensions: A survey

    Giulio Franzese and Pietro Michiardi. Generative diffusion models in infinite dimensions: A survey. Philosophical Transactions A, 383(2299):20240322, 2025

    2025

  7. [15]

    Gaussian interpolation flows

    Yuan Gao, Jian Huang, and Yuling Jiao. Gaussian interpolation flows. arXiv preprint arXiv:2311.11475, 2023

    2023 arXiv

  8. [16]

    Wavelet score-based generative modeling

    Florentin Guth, Simon Coste, Valentin De Bortoli, and Stephane Mallat. Wavelet score-based generative modeling. Advances in Neural Information Processing Systems , 35:478–491, 2022

    2022

  9. [17]

    Studies in the history of probability and statistics XLIX on the Mat´ ern correlation family.Biometrika, 93(4):989–995, 2006

    Peter Guttorp and Tilmann Gneiting. Studies in the history of probability and statistics XLIX on the Mat´ ern correlation family.Biometrika, 93(4):989–995, 2006

    2006

  10. [18]

    Mimicking the one-dimensional marginal distributions of processes having an Itˆ o differential

    Istv´ an Gy¨ ongy. Mimicking the one-dimensional marginal distributions of processes having an Itˆ o differential. Probability Theory and Related Fields , 71(4):501–516, 1986

    1986

  11. [19]

    Multilevel diffusion: Infinite dimensional score-based diffusion models for image generation

    Paul Hagemann, Sophie Mildenberger, Lars Ruthotto, Gabriele Steidl, and Nicole Tianjiao Yang. Multilevel diffusion: Infinite dimensional score-based diffusion models for image generation. arXiv preprint arXiv:2303.04772, 2023

    2023 arXiv

  12. [20]

    An introduction to stochastic PDEs

    Martin Hairer. An introduction to stochastic PDEs. arXiv preprint arXiv:0907.4178 , 2009

    2009 arXiv

  13. [21]

    Ergodicity of the 2D Navier-Stokes equations with de- generate stochastic forcing

    Martin Hairer and Jonathan C Mattingly. Ergodicity of the 2D Navier-Stokes equations with de- generate stochastic forcing. Annals of Mathematics , pages 993–1032, 2006

    2006

  14. [22]

    Denoising diffusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models. In Advances in Neural Information Processing Systems , volume 33, pages 6840–6851, 2020

    2020

  15. [23]

    Cascaded diffusion models for high fidelity image generation

    Jonathan Ho, Chitwan Saharia, William Chan, David J Fleet, Mohammad Norouzi, and Tim Sal- imans. Cascaded diffusion models for high fidelity image generation. Journal of Machine Learning Research, 23(47):1–33, 2022

    2022

  16. [24]

    Video diffusion models

    Jonathan Ho, Tim Salimans, Alexey Gritsenko, William Chan, Mohammad Norouzi, and David J Fleet. Video diffusion models. Advances in Neural Information Processing Systems , 35:8633–8646, 2022

    2022

  17. [25]

    Blurring diffusion models

    Emiel Hoogeboom and Tim Salimans. Blurring diffusion models. In The Eleventh International Conference on Learning Representations, 2023

    2023

  18. [26]

    Subspace diffusion gener- ative models

    Bowen Jing, Gabriele Corso, Renato Berlinghieri, and Tommi Jaakkola. Subspace diffusion gener- ative models. In European Conference on Computer Vision , pages 274–289. Springer, 2022

    2022

  19. [27]

    Gotta go fast when generating data with score-based models

    Alexia Jolicoeur-Martineau, Ke Li, R´ emi Pich´ e-Taillefer, Tal Kachman, and Ioannis Mitliagkas. Gotta go fast when generating data with score-based models. arXiv preprint arXiv:2105.14080 , 2021

    2021 arXiv

  20. [28]

    Elucidating the design space of diffusion- based generative models

    Tero Karras, Miika Aittala, Timo Aila, and Samuli Laine. Elucidating the design space of diffusion- based generative models. Advances in Neural Information Processing Systems , 35:26565–26577, 2022

    2022

  21. [29]

    Diffusion generative models in infinite dimen- sions

    Gavin Kerrigan, Justin Ley, and Padhraic Smyth. Diffusion generative models in infinite dimen- sions. In International Conference on Artificial Intelligence and Statistics, pages 9538–9563. PMLR, 2023

    2023

  22. [30]

    Functional flow matching

    Gavin Kerrigan, Giosue Migliorini, and Padhraic Smyth. Functional flow matching. arXiv preprint arXiv:2305.17209, 2023

    2023 arXiv

  23. [31]

    Learning skillful medium-range global weather forecasting

    Remi Lam, Alvaro Sanchez-Gonzalez, Matthew Willson, Peter Wirnsberger, Meire Fortunato, Fer- ran Alet, Suman Ravuri, Timo Ewalds, Zach Eaton-Rosen, Weihua Hu, Alexander Merose, Stephan 18 SCALE-ADAPTIVE GENERATIVE FLOWS FOR MULTISCALE SCIENTIFIC DATA Hoyer, George Holland, Ori...

    2023

  24. [32]

    Hierarchic flows to estimate and sample high-dimensional probabilities

    Etienne Lempereur and St´ ephane Mallat. Hierarchic flows to estimate and sample high-dimensional probabilities. arXiv preprint arXiv:2405.03468 , 2024

    2024

  25. [33]

    Fourier neural operator for parametric partial differential equations

    Zongyi Li, Nikola Kovachki, Kamyar Azizzadenesheli, Burigede Liu, Kaushik Bhattacharya, An- drew Stuart, and Anima Anandkumar. Fourier neural operator for parametric partial differential equations. arXiv preprint arXiv:2010.08895 , 2020

    2010 arXiv

  26. [34]

    Score-based diffusion models in function space

    Jae Hyun Lim, Nikola B Kovachki, Ricardo Baptista, Christopher Beckham, Kamyar Azizzade- nesheli, Jean Kossaifi, Vikram Voleti, Jiaming Song, Karsten Kreis, Jan Kautz, et al. Score-based diffusion models in function space. arXiv preprint arXiv:2302.07400 , 2023

    2023 arXiv

  27. [35]

    An explicit link between Gaussian fields and Gaussian Markov random fields: The stochastic partial differential equation approach

    Finn Lindgren, H ˚ avard Rue, and Johan Lindstr¨ om. An explicit link between Gaussian fields and Gaussian Markov random fields: The stochastic partial differential equation approach. Journal of the Royal Statistical Society Series B: Statistical Methodology , 73(4):423–498, 2011

    2011

  28. [36]

    Flow match- ing for generative modeling

    Yaron Lipman, Ricky TQ Chen, Heli Ben-Hamu, Maximilian Nickel, and Matthew Le. Flow match- ing for generative modeling. InThe Eleventh International Conference on Learning Representations, 2022

    2022

  29. [37]

    Flow straight and fast: Learning to generate and transfer data with rectified flow

    Xingchao Liu, Chengyue Gong, and Qiang Liu. Flow straight and fast: Learning to generate and transfer data with rectified flow. In The Eleventh International Conference on Learning Represen- tations, 2022

    2022

  30. [38]

    Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators

    Lu Lu, Pengzhan Jin, Guofei Pang, Zhongqiang Zhang, and George Em Karniadakis. Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators. Nature Machine Intelligence , 3(3):218–229, 2021

    2021

  31. [39]

    Wavelet conditional renor- malization group

    Tanguy Marchand, Misaki Ozawa, Giulio Biroli, and St´ ephane Mallat. Wavelet conditional renor- malization group. arXiv preprint arXiv:2207.04941 , 2022

    2022 arXiv

  32. [40]

    Improved denoising diffusion probabilistic models

    Alexander Quinn Nichol and Prafulla Dhariwal. Improved denoising diffusion probabilistic models. In International Conference on Machine Learning , pages 8162–8171. PMLR, 2021

    2021

  33. [41]

    Spectral diffusion processes

    Angus Phillips, Thomas Seror, Michael Hutchinson, Valentin De Bortoli, Arnaud Doucet, and Emile Mathieu. Spectral diffusion processes. arXiv preprint arXiv:2209.14125 , 2022

    2022 arXiv

  34. [42]

    Wavelet diffusion models are fast and scalable image gener- ators

    Hao Phung, Quan Dao, and Anh Tran. Wavelet diffusion models are fast and scalable image gener- ators. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 10199–10208, 2023

    2023

  35. [43]

    Infinite-dimensional diffu- sion models

    Jakiw Pidstrigach, Youssef Marzouk, Sebastian Reich, and Sven Wang. Infinite-dimensional diffu- sion models. arXiv preprint arXiv:2302.10130 , 2023

    2023 arXiv

  36. [44]

    Generative modelling with inverse heat dissi- pation

    Severi Rissanen, Markus Heinonen, and Arno Solin. Generative modelling with inverse heat dissi- pation. In The Eleventh International Conference on Learning Representations , 2022

    2022

  37. [45]

    Image super-resolution via iterative refinement

    Chitwan Saharia, Jonathan Ho, William Chan, Tim Salimans, David J Fleet, and Mohammad Norouzi. Image super-resolution via iterative refinement. IEEE Transactions on Pattern Analysis and Machine Intelligence , 45(4):4713–4726, 2022

    2022

  38. [46]

    Noise estimation for generative diffusion mod- els

    Robin San-Roman, Eliya Nachmani, and Lior Wolf. Noise estimation for generative diffusion mod- els. arXiv preprint arXiv:2104.02600 , 2021

    2021 arXiv

  39. [47]

    Blundell, Pietro Lio, Max Welling, Michael Bronstein, and Bruno Correia

    Arne Schneuing, Charles Harris, Yuanqi Du, Kieran Didi, Arian Jamasb, Ilia Igashov, Weitao Du, Carla Gomes, Tom L. Blundell, Pietro Lio, Max Welling, Michael Bronstein, and Bruno Correia. Structure-based drug design with equivariant diffusion models. Nature Computational Scien...

    2024

  40. [48]

    Bespoke solvers for generative flow models

    N Shaul, J Perez, RTQ Chen, A Thabet, A Pumarola, and Y Lipman. Bespoke solvers for generative flow models. In 12th International Conference on Learning Representations , 2024

    2024

  41. [49]

    Renormalization group flow, optimal transport and diffusion-based generative model.arXiv preprint arXiv:2402.17090, 2024

    Artan Sheshmani, Yi-Zhuang You, Baturalp Buyukates, Amir Ziashahabi, and Salman Avestimehr. Renormalization group flow, optimal transport and diffusion-based generative model.arXiv preprint arXiv:2402.17090, 2024

    2024 arXiv

  42. [50]

    Deep unsu- pervised learning using nonequilibrium thermodynamics

    Jascha Sohl-Dickstein, Eric A Weiss, Niru Maheswaranathan, and Surya Ganguli. Deep unsu- pervised learning using nonequilibrium thermodynamics. In Proceedings of the 32nd International Conference on Machine Learning , pages 2256–2265, 2015. SCALE-ADAPTIVE GENERATIVE FLOWS FOR ...

    2015

  43. [51]

    Generative modeling by estimating gradients of the data distribu- tion

    Yang Song and Stefano Ermon. Generative modeling by estimating gradients of the data distribu- tion. Advances in Neural Information Processing Systems , 32, 2019

    2019

  44. [52]

    Improved techniques for training score-based generative models

    Yang Song and Stefano Ermon. Improved techniques for training score-based generative models. Advances in Neural Information Processing Systems , 33:12438–12448, 2020

    2020

  45. [53]

    Score-based generative modeling through stochastic differential equations

    Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole. Score-based generative modeling through stochastic differential equations. arXiv preprint arXiv:2011.13456, 2020

    2011 arXiv

  46. [54]

    Interpolation of Spatial Data: Some Theory for Kriging

    Michael L Stein. Interpolation of Spatial Data: Some Theory for Kriging . Springer Science & Business Media, 1999

    1999

  47. [55]

    Inverse problems: A Bayesian perspective

    Andrew M Stuart. Inverse problems: A Bayesian perspective. Acta Numerica, 19:451–559, 2010

    2010

  48. [56]

    FunDiff: Diffusion models over function spaces for physics-informed generative modeling

    Sifan Wang, Zehao Dou, Tong-Rui Liu, and Lu Lu. FunDiff: Diffusion models over function spaces for physics-informed generative modeling. arXiv preprint arXiv:2506.07902 , 2025

    2025

  49. [57]

    On stationary processes in the plane

    Peter Whittle. On stationary processes in the plane. Biometrika, pages 434–449, 1954

    1954

  50. [58]

    Diffusion models: A comprehensive survey of methods and appli- cations

    Ling Yang, Zhilong Zhang, Yang Song, Shenda Hong, Runsheng Xu, Yue Zhao, Wentao Zhang, Bin Cui, and Ming-Hsuan Yang. Diffusion models: A comprehensive survey of methods and appli- cations. ACM Computing Surveys , 56(4):1–39, 2023

    2023

  51. [59]

    Wavelet flow: Fast training of high resolution normalizing flows

    Jason J Yu, Konstantinos G Derpanis, and Marcus A Brubaker. Wavelet flow: Fast training of high resolution normalizing flows. Advances in Neural Information Processing Systems , 33:6184–6196, 2020

    2020

  52. [60]

    A generative model for inorganic materials design

    Claudio Zeni, Robert Pinsler, Daniel Z¨ ugner, Andrew Fowler, Matthew Horton, Xiang Fu, Zilong Wang, Aliaksandra Shysheya, Jonathan Crabb´ e, Shoko Ueda, Roberto Sordillo, Lixin Sun, Jake Smith, Bichlien Nguyen, Hannes Schulz, Sarah Lewis, Chin-Wei Huang, Ziheng Lu, Yichi Zhou...

    2025

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.