Pith. sign in

REVIEW 3 major objections 6 minor 63 references

Neural Flow Samplers with Shortcut Models

T0 review · 3 major / 6 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read A flow-based neural sampler that estimates the partition-function derivative with sequential Monte Carlo and enforces shortcut consistency can sample in fewer steps without degrading quality.

desk verdict Solid empirical flow-sampler paper; the real weakness is the unmeasured on-policy particle gap in the plug-in estimator, not the same-batch mean itself, and the abstract's 'outperforms' oversells the table. read the letter →

arxiv 2502.07337 v2 pith:OCEW7NEC submitted 2025-02-11 cs.LG

Wuhao Chen , Zijing Ou , Yingzhen Li This is my paper · ORCID
classification cs.LG
keywords neuralsamplersflow-basedsamplingsequentialMonteCarlocontrolvariatesshortcutmodelscontinuityequationunnormalizeddensitiesmoleculardynamics
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

The paper introduces NFS2, a continuous flow-based neural sampler that learns a velocity field by minimizing a physics-informed residual derived from the continuity equation. Its two moves target the known weak points of such samplers: estimating the time derivative of the log partition function accurately, and reducing the number of ODE integration steps needed at sampling time. The first is handled by a velocity-driven Sequential Monte Carlo estimator augmented with a Stein control variate; the second by a shortcut consistency model that enforces agreement between one large step and two smaller steps. The paper claims this makes training stable on GMM-40, MW-32, DW-4, and LJ-13, and lets sample quality degrade slowly as the sampling budget drops from 128 to 32 or even 8 steps.

What carries the argument

The load-bearing identity is $\partial_t \log Z_t = \arg\min_{c_t} \mathbb{E}_{p_t}[(\xi_t(x;v_t) - c_t)^2]$ with $\xi_t(x;v_t) = \partial_t \log \tilde p_t(x) + \nabla_x \cdot v_t(x) + v_t(x)\cdot \nabla_x \log p_t(x)$; it follows from Stein's identity, which makes the velocity-dependent terms a zero-mean control variate. The mechanism that estimates this quantity is the velocity-driven SMC: particles are moved by the current velocity field and refined with Hamiltonian Monte Carlo, weighted by the annealing path ratio, and resampled when the effective sample size drops. The second mechanism is the shortcut model $s_t(x,d;\theta)$, which predicts the average velocity over an interval of duration $d$, trained with a consistency loss that compares one step of size $d$ with two steps of sizes $\alpha d$ and $(1-\alpha)d$ for a random split $\alpha$.

What would settle it

Take a target where $\partial_t \log Z_t$ is analytically available, such as an annealed Gaussian mixture with known mixture weights and covariances; run NFS2 with its plug-in SMC estimate and with the exact value. If the plug-in is biased or sample quality materially improves when the exact value is used, the central estimation mechanism is not performing as claimed.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central claim is that a flow-based sampler can be trained to sample from an unnormalized density without learning the evolving partition function separately: the time derivative $\partial_t \log Z_t$ is estimated on the fly by a Sequential Monte Carlo procedure whose proposals are driven by the velocity being trained, and the estimate is made low-variance by exploiting Stein's identity, which turns the velocity divergence plus drift term into a zero-mean control variate. Because the estimator reuses the same particles that train the velocity, the method is amortized and stable. A second claim is that the same network can be trained to jump over finite time intervals: a shortcut model predicts the average velocity over a duration $d$, and a generalized consistency loss (one step of length $d$ should match two steps of lengths $\alpha d$ and $(1-\alpha)d$) makes these predictions coherent. Empirically, NFS2 matches or beats existing flow samplers on the four benchmarks, and the shortcut consistency is what keeps sample quality high when integration steps are reduced.

Load-bearing premise

The paper assumes that the weighted average of the integrand computed from the same SMC particles that the current model generated is a faithful enough estimate of the true time derivative of the log partition function; if this plug-in is biased, minimizing the squared difference can train the velocity toward the wrong distribution.

Editorial extensions

If this is right

  • The learned sampler no longer needs a separate network to output $\partial_t \log Z_t$; the derivative is read off the same SMC particles used for training, removing a known source of instability.
  • One trained NFS2 model can be evaluated at 128, 64, 32, or 8 integration steps by just changing the step size in sampling, with only gradual degradation down to about 32 steps and competitive quality even at 8-16 steps on some targets.
  • Stronger shortcut regularisation (e.g., $\lambda = 10$) can cut the sampling budget by more than an order of magnitude, matching a no-shortcut model at 64 steps with only 4 steps on GMM-40.
  • The generalized random-split consistency loss dominates the midpoint-only variant across all tested step counts.
  • Accurate estimation of $\partial_t \log Z_t$ is pivotal: the PINN baseline that learns this term by gradient descent diverges or fails in high-dimensional settings, whereas NFS2's SMC estimate stays close to the long-MCMC reference.

Reading between the lines

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

  • The paper does not test the SMC plug-in estimator outside the PINN loss, but the same construction could provide normalizer derivatives for other simulation-free objectives such as flow matching or score-based samplers.
  • The velocity-driven proposal could also serve as a principled initialization for annealed importance sampling at inference time, separate from its role in training; the paper does not evaluate this use.
  • If the shortcut consistency generalizes as shown, it may push toward few-step or one-step sampling, though the paper's own LJ-13 results show that at 4 steps the sampler loses the fine structure of the interatomic distance distribution, so the regime of very few steps remains limited.
  • The divergence computation in the residual remains the scaling bottleneck; combining the method with a stochastic divergence estimator is the natural next step hinted at by the paper's limitations section.
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

3 major / 6 minor

Summary. The paper proposes Neural Flow Shortcut Sampler (NFS2), a continuous-time flow-based neural sampler trained by minimizing a residual derived from the continuity equation. The main contributions are (i) a velocity-driven Sequential Monte Carlo estimator for the time derivative of the log-partition function, augmented with a Stein-identity control variate, and (ii) a shortcut consistency loss that allows the trained model to generate samples with few integration steps. The method is evaluated on synthetic targets (GMM-40, MW-32) and n-body systems (DW-4, LJ-13), where it is reported to outperform or match flow-based baselines such as FAB, iDEM, LFIS, LIBD, and a PINN-trained baseline, while degrading gracefully when the number of sampling steps is reduced.

Significance. If the claims hold, NFS2 is a meaningful practical advance: it provides a lower-variance estimator for the intractable partition-function derivative during training, and the shortcut consistency mechanism gives a dynamically adjustable sampling budget. The paper contains a correct proof of the Stein-identity minimization in Appendix B.1, explicit training and sampling algorithms (Algorithms 3 and 4), and extensive ablations (including shortcut strength, architecture choice, and step-count sensitivity). The chief risk is that the training-time behavior of the key estimator is not characterized, and the shortcut consistency target is defined self-referentially, so the mechanism behind the reported gains is not fully established. The empirical study is broad and metrics are reasonable, but the lack of code or a training-time bias analysis weakens the confidence one can place in the main attribution claim.

major comments (3)
  1. [Algorithm 3, steps 4-5; Algorithm 2; Figures 1, 2, 9; Section 5.3] The training loss replaces the true PINN residual (ξ_t - ∂t log Z_t)^2 with (ξ_t - c_t)^2, where c_t is the same-batch SMC-weighted mean of ξ_t. If the SMC particles were exactly p_t-distributed with consistent weights, this would be a valid proxy because E_{p_t}[ξ_t] = ∂t log Z_t by Stein's identity and E[(ξ_t - c_t)^2] = (1 - 1/K) Var_{p_t}(ξ_t); the simple 'plug-in is biased' objection is therefore incomplete. However, during training the SMC particles are only approximately p_t-distributed: Algorithm 2 (line 14) seeds HMC with the velocity field that is being trained, and only 3-10 HMC steps are taken, so in early-training or low-ESS regimes E_{q_t}[ξ_t] can differ from ∂t log Z_t. This gap is never measured, and the estimator comparisons in Figures 1, 2, and 9 are all performed post-hoc with a well-trained velocity field. Since Section 5.3 attributes the method's success to accurate ∂t log Z_t estimation, the paper should provide either a training-time diagnostic (e.g., ESS and an estimate of E_{q_t}[ξ_t] - ∂t log Z_t on a tractable target where the truth is computable) or an experiment that computes c_t from an independent, more accurate particle set. Without this, the central attribution claim is unsupported.
  2. [Section 3.2 and Appendix B.2, Eq. (27)] The estimator is described as a Stein control variate, but the algorithm uses a fixed coefficient of 1 for the Stein term rather than the optimal coefficient β* derived in Appendix B.2. With β=1, the variance of the proposed estimator is Var_{p_t}(f) + Var_{p_t}(g) + 2Cov_{p_t}(f,g) where f = ∂t log p̃_t and g = ∇·v_t + v_t·∇ log p_t; this is lower than the variance of the simple Monte Carlo estimator only if 2Cov(f,g) < -Var(g). The paper does not establish this condition. The ideal-limit argument (g = -f + c when the velocity is optimal) gives β*=1, but outside that limit the fixed-coefficient estimator can increase variance. I recommend either estimating β* from samples, with a demonstration that variance is reduced, or providing explicit conditions and experiments under which the fixed-coefficient choice is variance-reducing.
  3. [Section 3.3, Eq. (11); Table 2; Figure 10] The shortcut consistency target s_target is computed from the model's own stop-gradient predictions, so the loss enforces self-consistency between one-step and two-step predictions but is not anchored to the true average velocity of the continuity-equation flow. If the instantaneous velocity is imperfect, the consistency loss can amplify the model's systematic error. The ablations show that stronger consistency improves few-step sampling, but they do not establish that the consistency target itself is correct. Please add a comparison with an alternative consistency target computed by numerically integrating the learned velocity with a fine solver, or at least discuss and test under what conditions the self-generated target is sufficient for accurate few-step sampling.
minor comments (6)
  1. [General] The paper does not report wall-clock training time or the computational overhead of the SMC and divergence computation, which matter for practical usability.
  2. [Figure 2 and Figure 1 captions] The figures present 'standard deviation' and 'MSE' of the ∂t log Z_t estimator, but do not specify the experimental target, the number of particles, the number of SMC time steps, or the exact estimator formulas used; please add these details.
  3. [Table 1] The table has irregular formatting: some entries are marked with '*' and a footnote about disjoint supports, but the meaning of missing or blank entries is not always clear; please unify the notation and clarify whether a '*' means the metric is undefined or simply not reported.
  4. [Appendix E.3] The training details state '10^4 epochs (500 steps/epoch)' for all tasks, but it is not explained how early stopping is determined or which validation metric is used; please clarify.
  5. [Section 6 and Appendix G] The limitation that 'the estimation of ∂t log Z_t remains intractable and could be problematic in large-scale particle systems' appears to conflict with the abstract's strong claim of an 'improved estimator'; please reconcile the language and state more precisely the regime in which the estimator is expected to be reliable.
  6. [Supplementary material] The text mentions a Jupyter notebook in the supplementary information, but no URL or repository is provided; please include a link to the code for reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the derivation chain is self-contained.

full rationale

The core derivation is self-contained. Eq. (6) defines the PINN residual, and Eqs. (7)-(10) establish, by Stein's identity, that the minimizer in Eq. (9) is exactly the true time-derivative of the log-partition function, E_{p_t}[xi_t] = ∂_t log Z_t. Algorithm 3's c_t is a Monte Carlo/SMC plug-in estimate of this expectation using the current particles, not a fitted parameter, so the training loss is not defined in terms of the model's own outputs in any circular way. In the ideal limit where SMC particles are p_t-distributed, E[(ξ−ĉ)^2] = (1−1/K)E[δ^2], so the surrogate loss is a scaled version of the stated residual rather than a quantity whose minimizer is forced by construction. The shortcut consistency term in Eq. (11) is a self-distillatory regularizer whose target uses stop-gradient predictions, but sampler correctness is anchored by the first residual term; consistency alone is not claimed to define the target distribution. The only self-citation (He et al., 2024) appears in background/related work and is not load-bearing. The paper's admitted limitation that ∂_t log Z_t estimation 'could be problematic in large-scale particle systems' is a robustness caveat, not evidence of circularity. No uniqueness theorem, fitted-then-predicted quantity, or renamed known result is used as a load-bearing step.

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

The central claim relies on a handful of assumptions: a well-chosen annealing path, boundary conditions for Stein's identity, and two unproven training heuristics (the same-batch control variate surrogate and the self-distilled shortcut target). There are no new physical entities or mediators.

free parameters (1)
  • Shortcut consistency weight lambda = 0.01 to 10, task-dependent
    The weight of the consistency loss in Eq. (11) is chosen by hand and varied in experiments; it controls the trade-off between few-step sampling quality and full-budget accuracy (Figure 10).
assumptions (5)
  • domain assumption The annealing interpolation defines a valid, integrable probability path for all t.
    Used in Eq. (5); requires target and base to have common support and sufficient integrability.
  • standard math Boundary conditions for Stein's identity hold: the target density vanishes fast enough at infinity.
    Lemma 1 in Appendix B.1 uses the divergence theorem; valid for Gaussian tails and the harmonically confined molecular systems considered.
  • domain assumption The HMC kernels within SMC mix well enough that the weighted particles approximate the expectation of the integrand.
    Algorithm 1 and 2 rely on SMC estimates of the log-partition derivative; if HMC mixes poorly, the estimates are biased and training can diverge, as acknowledged in Section 5.3.
  • ad hoc to paper The same-batch control variate c_t is an unbiased or sufficiently accurate surrogate for the true expectation.
    Algorithm 3 step 5 computes c_t from the same particles used in the residual loss; no proof is given that the squared deviation is a valid estimate of the PINN residual.
  • ad hoc to paper The self-generated shortcut target starget enforces the correct finite-step flow.
    Eq. (11) trains the shortcut model against the model's own stop-gradient predictions; without the residual term this would be degenerate, and no guarantee is given that the optimum is the true flow.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Neural Flow Samplers with Shortcut Models." pith.science (2026). https://pith.science/paper/OCEW7NEC

@misc{pith2026250207337,
  author       = {Pith},
  title        = {Pith review of: Neural Flow Samplers with Shortcut Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OCEW7NEC}},
  note         = {Machine review of arXiv:2502.07337}
}
read the original abstract

Sampling from unnormalized densities presents a fundamental challenge with wide-ranging applications, from posterior inference to molecular dynamics simulations. Continuous flow-based neural samplers offer a promising approach, learning a velocity field that satisfies key principles of marginal density evolution (e.g., the continuity equation) to generate samples. However, this learning procedure requires accurate estimation of intractable terms linked to the computationally challenging partition function, for which existing estimators often suffer from high variance or low accuracy. To overcome this, we introduce an improved estimator for these challenging quantities, employing a velocity-driven Sequential Monte Carlo method enhanced with control variates. Furthermore, we introduce a shortcut consistency model to boost the runtime efficiency of the flow-based neural sampler by minimizing its required sampling steps. Our proposed Neural Flow Shortcut Sampler empirically outperforms existing flow-based neural samplers on both synthetic datasets and complex n-body system targets.

Figures

Figures reproduced from arXiv: 2502.07337 by the authors.

Figure 1
Figure 1. MSE of the estimated ∂t log Zt. To demonstrate the efficacy of the proposed velocity￾driven SMC sampler, we measure the Mean Squared Error (MSE) of the ∂t log Zt estimation against the ref￾erence (obtained from very long MCMC chains) for varying numbers of MCMC steps within each SMC tran￾sition (see [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Variance reduction as shown by the standard deviations of the estimators. Empirically, we observe that Eq. (9) achieves lower variance compared to a simple Monte Carlo estima￾tor ∂t log ZtEpt ∂t log ˜pt(x) ≈ PK k=1 w¯ (k) t log ˜pt(x (k) t ), and it sometimes leads to better optimisation of the training objective Eq. (6). We visualize the compar￾ison of the standard deviation of the two estimation methods in [PITH_… view at source ↗
Figure 3
Figure 3. Samples of GMM-40, with contour lines representing the ground truth distribution. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (14 more)
Figure 4
Figure 4. Figure 4: 2D marginal samples from the 1st and 3rd dimensions of MW-32. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Histogram of interatomic distance and sample energy on DW-4. As shown in Table [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 8
Figure 8. Figure 8: Visual comparison of shortcut regularisation [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 6
Figure 6. Figure 6: Histogram of interatomic distance and sample energy on LJ-13. On the more complex LJ-13 system, NFS2 demonstrates good performance, as detailed in Table [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Histogram of interatomic distance and sam [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 9
Figure 9. Figure 9: Comparison of ∂t log Zt estimates. Shortcut regularisation enables faster sampling. To reduce the sampling cost, NFS2 is trained using ad￾ditional trajectory shortcut consistency loss (Eq. (11)). We compared a model trained with the regularisation enabled against one t…
Figure 10
Figure 10. Figure 10: Comparison of X -TV across dif￾ferent shortcut regularisation strength λ. Better shortcut consistency leads to fewer steps required for high quality samples. Empirically in GMM-40 experiments, stronger shortcut consistency regularisation substantially reduces the samp…
Figure 11
Figure 11. Figure 11: Training loss of using different estimators of ∂t log Zt. Remark. Equation (9) provides an alternative approach to estimate ∂t log Zt. As illustrated in [PITH_FULL_IMAGE:figures/full_fig_p018_11.png]
Figure 12
Figure 12. Figure 12: Samples on MW-32. First row: 2D marginal samples from the 1st and 4th dimensions; [PITH_FULL_IMAGE:figures/full_fig_p024_12.png]
Figure 13
Figure 13. Figure 13: Histogram of sample energy on MW￾32. This section presents additional visualizations of generated samples on MW-32. As shown in [PITH_FULL_IMAGE:figures/full_fig_p024_13.png]
Figure 14
Figure 14. Figure 14: Comparison of (a) interatomic distance distributions and (b) energy distributions for the 2 [PITH_FULL_IMAGE:figures/full_fig_p025_14.png]
Figure 15
Figure 15. Figure 15: Illustration of the generated samples using different sampling steps on GMM-40. [PITH_FULL_IMAGE:figures/full_fig_p026_15.png]
Figure 16
Figure 16. Figure 16: Illustration of the generated samples using different sampling steps on MW-32. [PITH_FULL_IMAGE:figures/full_fig_p027_16.png]
Figure 17
Figure 17. Figure 17: Illustration of the generated samples using models trained with different shortcut regulari [PITH_FULL_IMAGE:figures/full_fig_p028_17.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

63 extracted references · 29 canonical work pages

  1. [1]

    Albergo, M. S. and Vanden-Eijnden, E. Nets: A non-equilibrium transport sampler, 2025. URL https://arxiv.org/abs/2410.02711

    arXiv 2025

  2. [2]

    S., Kanwar, G., and Shanahan, P

    Albergo, M. S., Kanwar, G., and Shanahan, P. E. Flow-based generative models for markov chain monte carlo in lattice field theory. Physical Review D, 100 0 (3): 0 034515, 2019

    2019

  3. [3]

    P., Simm, G

    Batatia, I., Kov \'a cs, D. P., Simm, G. N. C., Ortner, C., and Cs \'a nyi, G. Mace: Higher order equivariant message passing neural networks for fast and accurate force fields, 2022

    work page 2022

  4. [4]

    An optimal control perspective on diffusion-based generative modeling

    Berner, J., Richter, L., and Ullrich, K. An optimal control perspective on diffusion-based generative modeling. arXiv preprint arXiv:2211.01364, 2022

    arXiv 2022

  5. [5]

    J., Leary, C., Maclaurin, D., Necula, G., Paszke, A., Vander P las, J., Wanderman- M ilne, S., and Zhang, Q

    Bradbury, J., Frostig, R., Hawkins, P., Johnson, M. J., Leary, C., Maclaurin, D., Necula, G., Paszke, A., Vander P las, J., Wanderman- M ilne, S., and Zhang, Q. JAX : composable transformations of P ython+ N um P y programs, 2018. URL http://github.com/jax-ml/jax

    2018

  6. [6]

    Handbook of Markov Chain Monte Carlo

    Brooks, S., Gelman, A., Jones, G., and Meng, X.-L. Handbook of Markov Chain Monte Carlo. CRC press, 2011

    work page 2011

  7. [7]

    Blackjax: Composable B ayesian inference in JAX , 2024

    Cabezas, A., Corenflos, A., Lao, J., and Louf, R. Blackjax: Composable B ayesian inference in JAX , 2024

    work page 2024

  8. [8]

    T., and Pavon, M

    Chen, Y., Georgiou, T. T., and Pavon, M. On the relation between optimal transport and schr \"o dinger bridges: A stochastic control viewpoint. Journal of Optimization Theory and Applications, 169: 0 671--691, 2016

    work page 2016

Show all 63 references
  1. [9]

    Sequential monte carlo samplers

    Del Moral, P., Doucet, A., and Jasra, A. Sequential monte carlo samplers. Journal of the Royal Statistical Society Series B: Statistical Methodology, 68 0 (3): 0 411--436, 2006

    2006

  2. [10]

    An introduction to sequential monte carlo methods

    Doucet, A., De Freitas, N., and Gordon, N. An introduction to sequential monte carlo methods. Sequential Monte Carlo methods in practice, pp.\ 3--14, 2001

    2001

  3. [11]

    D., Pendleton, B

    Duane, S., Kennedy, A. D., Pendleton, B. J., and Roweth, D. Hybrid monte carlo. Physics Letters B, 195 0 (2): 0 216--222, 1987. ISSN 0370-2693. doi:10.1016/0370-2693(87)91197-X

    1987 doi

  4. [12]

    One step diffusion via shortcut models

    Frans, K., Hafner, D., Levine, S., and Abbeel, P. One step diffusion via shortcut models. arXiv preprint arXiv:2410.12557, 2024

    2024 arXiv

  5. [13]

    and Smit, B

    Frenkel, D. and Smit, B. Understanding molecular simulation: from algorithms to applications. Elsevier, 2023

    2023

  6. [14]

    and Domke, J

    Geffner, T. and Domke, J. Using large ensembles of control variates for variational inference. Advances in Neural Information Processing Systems, 31, 2018

    2018

  7. [15]

    and Domke, J

    Geffner, T. and Domke, J. Mcmc variational inference via uncorrected hamiltonian annealing. Advances in Neural Information Processing Systems, 34: 0 639--651, 2021

    2021

  8. [16]

    Gerdes, M., de Haan, P., Rainone, C., Bondesan, R., and Cheng, M. C. Learning lattice quantum field theories with equivariant continuous flows. SciPost Physics, 15 0 (6): 0 238, 2023

    2023

  9. [17]

    Novel approach to nonlinear/non-gaussian bayesian state estimation

    Gordon, N., Salmond, D., and Smith, A. Novel approach to nonlinear/non-gaussian bayesian state estimation. IEE Proceedings F (Radar and Signal Processing), 140 0 (2): 0 107--113, apr 1993. doi:10.1049/ip-f-2.1993.0015

    1993

  10. [18]

    Grathwohl, W., Chen, R. T. Q., Bettencourt, J., Sutskever, I., and Duvenaud, D. FFJORD: free-form continuous dynamics for scalable reversible generative models. CoRR, abs/1810.01367, 2018. URL http://arxiv.org/abs/1810.01367

    2018 arXiv

  11. [19]

    K., Yan, B., Domingo-Enrich, C., Sriram, A., Wood, B., Levine, D., Hu, B., Amos, B., Karrer, B., Fu, X., Liu, G.-H., and Chen, R

    Havens, A., Miller, B. K., Yan, B., Domingo-Enrich, C., Sriram, A., Wood, B., Levine, D., Hu, B., Amos, B., Karrer, B., Fu, X., Liu, G.-H., and Chen, R. T. Q. Adjoint sampling: Highly scalable diffusion samplers via adjoint matching, 2025. URL https://arxiv.org/abs/2504.11713

    2025 arXiv

  12. [20]

    He, J., Chen, W., Zhang, M., Barber, D., and Hern \'a ndez-Lobato, J. M. Training neural samplers with reverse diffusive kl divergence. arXiv preprint arXiv:2410.12456, 2024

    2024 arXiv

  13. [21]

    Denoising diffusion probabilistic models

    Ho, J., Jain, A., and Abbeel, P. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33: 0 6840--6851, 2020

    2020

  14. [22]

    Hutchinson, M. F. A stochastic estimator of the trace of the influence matrix for laplacian smoothing splines. Communications in Statistics-Simulation and Computation, 18 0 (3): 0 1059--1076, 1989

    1989

  15. [23]

    On the determination of molecular fields

    Jones, J. On the determination of molecular fields. i. from the variation of the viscosity of a gas with temperature. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 106 0 (738): 0 441--462, 1924. doi:10.1098/rspa.1924.0081

    1924

  16. [24]

    Random sampling (monte carlo) techniques in neutron attenuation problems

    Kahn, H. Random sampling (monte carlo) techniques in neutron attenuation problems. i. Nucleonics (US) Ceased publication, 6 0 (See also NSA 3-990), 1950

    1950

  17. [25]

    O n N eural D ifferential E quations

    Kidger, P. O n N eural D ifferential E quations . PhD thesis, University of Oxford, 2021

    2021

  18. [26]

    and Garcia, C

    Kidger, P. and Garcia, C. E quinox: neural networks in JAX via callable P y T rees and filtered transformations. Differentiable Programming workshop at Neural Information Processing Systems 2021, 2021

    2021

  19. [27]

    Equivariant flow matching

    Klein, L., Kr \"a mer, A., and No \'e , F. Equivariant flow matching. Advances in Neural Information Processing Systems, 36, 2024

    2024

  20. [28]

    E., Platen, E., Kloeden, P

    Kloeden, P. E., Platen, E., Kloeden, P. E., and Platen, E. Stochastic differential equations. Springer, 1992

    1992

  21. [29]

    Equivariant flows: Exact likelihood generative learning for symmetric densities

    K \"o hler, J., Klein, L., and Noe, F. Equivariant flows: Exact likelihood generative learning for symmetric densities. In III, H. D. and Singh, A. (eds.), Proceedings of the 37th International Conference on Machine Learning, volume 119 of Proceedings of Machine Learning Resea...

    2020

  22. [30]

    Md simulation of nanometric cutting of single crystal aluminum--effect of crystal orientation and direction of cutting

    Komanduri, R., Chandrasekaran, N., and Raff, L. Md simulation of nanometric cutting of single crystal aluminum--effect of crystal orientation and direction of cutting. Wear, 242 0 (1-2): 0 60--88, 2000

    2000

  23. [31]

    T., Ben-Hamu, H., Nickel, M., and Le, M

    Lipman, Y., Chen, R. T., Ben-Hamu, H., Nickel, M., and Le, M. Flow matching for generative modeling. arXiv preprint arXiv:2210.02747, 2022

    2022 arXiv

  24. [32]

    Action-depedent control variates for policy optimization via stein's identity

    Liu, H., Feng, Y., Mao, Y., Zhou, D., Peng, J., and Liu, Q. Action-depedent control variates for policy optimization via stein's identity. arXiv preprint arXiv:1710.11198, 2017

    2017 arXiv

  25. [33]

    Liu, J. S. and Chen, R. Sequential monte carlo methods for dynamic systems. Journal of the American Statistical Association, 93 0 (443): 0 1032--1044, 1998. doi:10.1080/01621459.1998.10473765

    1998

  26. [35]

    J., and Doucet, A

    Matthews, A., Arbel, M., Rezende, D. J., and Doucet, A. Continual repeated annealed flow transport monte carlo. In International Conference on Machine Learning, pp.\ 15196--15219. PMLR, 2022

    2022

  27. [36]

    I., Stimper, V., Simm, G

    Midgley, L. I., Stimper, V., Simm, G. N., Sch \"o lkopf, B., and Hern \'a ndez-Lobato, J. M. Flow annealed importance sampling bootstrap. International Conference on Learning Representations (ICLR), 2023

    2023

  28. [37]

    and Fleuret, F

    Máté, B. and Fleuret, F. Learning interpolations between boltzmann densities, 2023. URL https://arxiv.org/abs/2301.07388

    2023 arXiv

  29. [38]

    Neal, R. M. Probabilistic inference using markov chain monte carlo methods. 1993

    1993

  30. [39]

    Neal, R. M. Annealed importance sampling. Statistics and computing, 11: 0 125--139, 2001

    2001

  31. [40]

    Transport meets variational inference: Controlled monte carlo diffusions

    Nusken, N., Vargas, F., Padhy, S., and Blessing, D. Transport meets variational inference: Controlled monte carlo diffusions. In The Twelfth International Conference on Learning Representations: ICLR 2024, 2024

    2024

  32. [41]

    OuYang, R., Qiang, B., and Hern \'a ndez-Lobato, J. M. Bnem: A boltzmann sampler based on bootstrapped noised energy matching. arXiv preprint arXiv:2409.09787, 2024

    2024

  33. [42]

    Stochastic control and nonequilibrium thermodynamical systems

    Pavon, M. Stochastic control and nonequilibrium thermodynamical systems. Applied Mathematics and Optimization, 19: 0 187--202, 1989

    1989

  34. [43]

    and Xie, S

    Peebles, W. and Xie, S. Scalable diffusion models with transformers. arXiv preprint arXiv:2212.09748, 2022

    2022 arXiv

  35. [44]

    Black box variational inference

    Ranganath, R., Gerrish, S., and Blei, D. Black box variational inference. In Artificial intelligence and statistics, pp.\ 814--822. PMLR, 2014

    2014

  36. [45]

    and Mohamed, S

    Rezende, D. and Mohamed, S. Variational inference with normalizing flows. In International conference on machine learning, pp.\ 1530--1538. PMLR, 2015

    2015

  37. [46]

    Roberts, G. O. and Rosenthal, J. S. Optimal scaling for various metropolis-hastings algorithms. Statistical science, 16 0 (4): 0 351--367, 2001

    2001

  38. [47]

    A., Rector-Brooks, J., Bose, A

    Sadegh, T. A., Rector-Brooks, J., Bose, A. J., Mittal, S., Lemos, P., Liu, C.-H., Sendera, M., Ravanbakhsh, S., Gidel, G., Bengio, Y., Malkin, N., and Tong, A. Iterated denoising energy matching for sampling from boltzmann densities. ArXiv, abs/2402.06121, 2024. URL https://ap...

    2024 arXiv

  39. [48]

    G., Hoogeboom, E., and Welling, M

    Satorras, V. G., Hoogeboom, E., and Welling, M. E(n) equivariant graph neural networks. In Meila, M. and Zhang, T. (eds.), Proceedings of the 38th International Conference on Machine Learning, ICML 2021, 18-24 July 2021, Virtual Event , volume 139 of Proceedings of Machine Lea...

    2021

  40. [49]

    Use of exchangeable pairs in the analysis of simulations

    Stein, C., Diaconis, P., Holmes, S., and Reinert, G. Use of exchangeable pairs in the analysis of simulations. Lecture Notes-Monograph Series, pp.\ 1--26, 2004

    2004

  41. [50]

    Stein, C. M. Estimation of the mean of a multivariate normal distribution. The Annals of Statistics, 9 0 (6): 0 1135--1151, November 1981. ISSN 0090-5364. doi:10.1214/aos/1176345632

    1981

  42. [51]

    Thi \'e baux, H. J. and Zwiers, F. W. The interpretation and estimation of effective sample size. Journal of Applied Meteorology and Climatology, 23 0 (5): 0 800--811, 1984

    1984

  43. [52]

    Tian, Y., Panda, N., and Lin, Y. T. Liouville flow importance sampler. arXiv preprint arXiv:2405.06672, 2024

    2024 arXiv

  44. [53]

    and Raginsky, M

    Tzen, B. and Raginsky, M. Theoretical guarantees for sampling and inference in generative models with latent diffusions. In Conference on Learning Theory, pp.\ 3084--3114. PMLR, 2019

    2019

  45. [54]

    The unscented particle filter

    Van Der Merwe, R., Doucet, A., De Freitas, N., and Wan, E. The unscented particle filter. Advances in neural information processing systems, 13, 2000

    2000

  46. [55]

    Denoising diffusion samplers

    Vargas, F., Grathwohl, W., and Doucet, A. Denoising diffusion samplers. arXiv preprint arXiv:2302.13834, 2023

    2023 arXiv

  47. [56]

    N., Kaiser, L

    Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L. u., and Polosukhin, I. Attention is all you need. In Guyon, I., Luxburg, U. V., Bengio, S., Wallach, H., Fergus, R., Vishwanathan, S., and Garnett, R. (eds.), Advances in Neural Informatio...

    2017

  48. [57]

    Villani, C. et al. Optimal transport: old and new, volume 338. Springer, 2009

    2009

  49. [58]

    and Ahn, S

    Woo, D. and Ahn, S. Iterated energy-based flow matching for sampling from boltzmann densities. arXiv preprint arXiv:2408.16249, 2024

    2024 arXiv

  50. [59]

    A comprehensive study of non-adaptive and residual-based adaptive sampling for physics-informed neural networks

    Wu, C., Zhu, M., Tan, Q., Kartha, Y., and Lu, L. A comprehensive study of non-adaptive and residual-based adaptive sampling for physics-informed neural networks. Computer Methods in Applied Mechanics and Engineering, 403: 0 115671, January 2023. ISSN 0045-7825. doi:10.1016/j.c...

    2023

  51. [60]

    Stochastic normalizing flows

    Wu, H., K \"o hler, J., and No \'e , F. Stochastic normalizing flows. Advances in Neural Information Processing Systems, 33: 0 5933--5944, 2020

    2020

  52. [61]

    Mars: Markov molecular sampling for multi-objective drug discovery

    Xie, Y., Shi, C., Zhou, H., Yang, Y., Zhang, W., Yu, Y., and Li, L. Mars: Markov molecular sampling for multi-objective drug discovery. arXiv preprint arXiv:2103.10432, 2021

    2021 arXiv

  53. [62]

    Zhang, G., Hsu, K., Li, J., Finn, C., and Grosse, R. B. Differentiable annealed importance sampling and the perils of gradient noise. Advances in Neural Information Processing Systems, 34: 0 19398--19410, 2021

    2021

  54. [64]

    and Chen, Y

    Zhang, Q. and Chen, Y. Path integral sampler: a stochastic control approach for sampling, 2022. URL https://arxiv.org/abs/2111.15141

    2022 arXiv

  55. [65]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.