Fisher Score Matching for Simulation-Based Forecasting and Inference

Benjamin D. Wandelt; Ce Sui; Shivam Pandey

REVIEW 3 major objections 5 minor 30 references

Fisher Score Matching for Simulation-Based Forecasting and Inference

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper proposes Fisher score matching, a training objective that lets a neural network recover the gradient of the log-likelihood from simulator samples alone, and shows the learned score drives Fisher forecasts and Hamiltonian Monte…

desk verdict Clean and correct core trick, modestly novel; the headline claim about non-differentiable simulators is not actually tested. read the letter →

arxiv 2507.07833 v1 pith:PBP6CVNK submitted 2025-07-10 astro-ph.CO astro-ph.IM

Ce Sui , Shivam Pandey , Benjamin D. Wandelt This is my paper

classification astro-ph.COastro-ph.IM

keywords Fisherscorematchingsimulation-basedinferencelikelihood-freeHamiltonianMonteCarloinformationmatrixcosmologicalparameterestimationlatentvariablemodel

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

Simulation-based models usually hide their likelihood, so the gradient of the log-likelihood with respect to parameters—the Fisher score—is out of reach unless the simulator is differentiable. This paper shows that the Fisher score is a conditional expectation of a latent-model score, which means an ordinary regression network trained on simulated data can learn it. If correct, this gives a purely simulation-driven route to Fisher information matrices, maximum likelihood estimation, and gradient-based Bayesian sampling for non-differentiable simulators. The paper validates the idea on a linear Gaussian model and a two-parameter weak lensing example, where learned scores closely match analytically or autodifferentiably computed ground truth for plausible data-parameter pairs.

What carries the argument

The load-bearing identity is $\nabla_\theta \log P(x|\theta) = \mathbb{E}_{P(\theta^*|x,\theta)}[\nabla_\theta \log P(\theta^*|\theta)]$, derived from the Markov chain $\theta \to \theta^* \to x$; it converts the Fisher score into a Bayes least squares regression target, which is why the simple mean squared error loss of Equation (3) suffices. The second mechanism is the auxiliary-noise construction $\theta = \tilde\theta + w$ for indecomposable simulators, which replaces the original model by an extended one whose score is a convolution of the true score with $P(\theta|\tilde\theta,x)$; the method then relies on that kernel being sharply peaked so the smoothed score approximates the true one.

What would settle it

Take a non-differentiable simulator with a low-dimensional parameter space, estimate the true log-likelihood gradient by finite differences from a large number of simulations at fiducial points, train the Fisher score matching network with the paper's noise scale choice, and check whether the difference between the learned and finite-difference scores shrinks as the noise scale is reduced; if it does not shrink, the peakedness assumption fails.

Watch

Extended reading notes

Core claim

Under a latent-variable model with Markov chain $\theta \to \theta^* \to x$, the Fisher score satisfies $\nabla_\theta \log P(x|\theta) = \mathbb{E}_{P(\theta^*|x,\theta)}[\nabla_\theta \log P(\theta^*|\theta)]$, so the intractable likelihood gradient becomes the posterior mean of a tractable latent score. Since the minimizer of mean squared error over simulated triplets $(x,\theta,\theta^*)$ is exactly that posterior mean, training $s(x,\theta)$ to predict $\nabla_\theta \log P(\theta^*|\theta)$ makes $s$ converge to the true Fisher score, with no derivative of the simulator required. For models without a natural latent, the paper adds auxiliary parameter noise $\theta = \tilde\theta + w$ and learns the score of the extended model $\nabla_{\tilde\theta} \log P(x|\tilde\theta)$, which equals the true score convolved with the sharp kernel $P(\theta|\tilde\theta,x)$ and is accurate when that kernel is concentrated. The experiments show the learned score field recovers the analytic score in the linear Gaussian case and the autodifferentiated score in the weak lensing case, and that the score can be used to estimate Fisher matrices and to run Hamiltonian Monte Carlo posterior sampling.

Load-bearing premise

The method's accuracy for a non-decomposable simulator rests on the added parameter noise being so small that the artificially blurred model still has essentially the same score as the original simulator, and when the simulator is not differentiable there is no direct way to verify that equivalence.

Editorial extensions

If this is right

A trained score estimator yields the Fisher information matrix at any parameter value as the covariance of the score over data samples, enabling forecasts without an analytic likelihood.
The learned score provides a likelihood-free gradient for Hamiltonian Monte Carlo, so posterior sampling can use gradient-based exploration on simulators that are not differentiable.
Maximum-likelihood-style point estimates become available by finding where the learned score vanishes, with no need to specify a prior during score training.
The training objective is prior-independent: the same score network can be reused across different analysis choices of prior, since only the simulator and latent model enter the loss.
For decomposable models, any valid latent decomposition trains the network to the same true Fisher score, so the choice of decomposition does not bias the result.

Reading between the lines

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

A practical certification recipe suggested by the method: for a new non-differentiable simulator, train with a few noise scales and compare the resulting Fisher matrices against finite-difference estimates on a low-dimensional parameter subspace to detect when the peakedness assumption breaks.
The same regression framework could be extended to learn the score as a function of the noise scale, letting one check convergence by seeing whether the learned score stabilizes as $\sigma \to 0$.
Because the method only needs forward samples, it could be combined with active learning that concentrates simulation budget around the fiducial parameters, improving score accuracy exactly in the region forecasts and inference use.
If the score can be learned at field level for high-dimensional summaries, it would connect naturally to field-level Fisher analyses and simulation-based inference pipelines.

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.

Desk Editor's Note

read the letter

The core identity in Eq. 2 is right, and the regression loss in Eq. 3 is a sensible way to learn the Fisher score. But this specific identity and MSE regression already appear in Brehmer et al. 2020, so the novelty is narrower than the abstract suggests. What is new is the explicit framing for Fisher forecasting and HMC, plus the auxiliary-noise trick for non-decomposable simulators. That framing is useful and the paper is honest about the main limitation: Appendix B shows the auxiliary-noise estimator targets a smoothed version of the true score, Eq. 5, and the approximation depends on the noise being sharply peaked. The paper does not hide this, but it also does not quantify the bias it introduces.

The experiments are clean as proofs of concept. The toy linear Gaussian matches the analytic score, and the weak-lensing example shows the learned score, posterior, and Fisher matrix agree visually with autodiff ground truth. Code is available. That is real evidence the method works when the simulator is differentiable and the latent-noise model is well-behaved.

The soft spots are proportionate. First, no experiment uses a genuinely non-differentiable simulator, so the central promise of the abstract is untested. The weak-lensing example uses jax-cosmo, which is differentiable; it validates the regression, not the auxiliary-noise trick in the regime it was designed for. Second, the smoothing bias is unquantified. The stress-test note is correct that in a Gaussian location model the extended Fisher information is (Sigma + sigma^2 I)^-1, so forecasts are biased low, and the paper does not measure this contraction for the weak-lensing case. Third, the results are qualitative: no error bars, no quantitative score errors. The reported 40% underestimation for implausible parameters is hand-waved as irrelevant to high-likelihood regions, which is plausible but not demonstrated.

These are addressable gaps, not fatal flaws. The method may well work for many non-differentiable simulators, but the current evidence does not establish it. The paper would benefit from a test on a real black-box simulator and a characterization of the bias as a function of sigma.

Who is this for? People doing SBI or Fisher forecasting with expensive simulators will find the framing useful. It deserves a serious referee, but the referee should push for a non-differentiable validation and quantitative error metrics.

Referee Report

3 major / 5 minor

Summary. This manuscript proposes Fisher Score Matching (FSM), a simulation-based method to train a neural network to approximate the Fisher score s(x, θ) = ∇θ log p(x | θ). The central identity (Eq. 2) represents the score as a posterior expectation of a latent score ∇θ log p(θ* | θ), which can be learned by regression (Eq. 3). For models without a natural latent variable, the authors introduce an auxiliary variable θ~ with a noise model p(θ | θ~), and show that the learned estimator approximates the score of the extended model, which is a smoothed version of the true score (Eq. 5). The method is validated on a two-dimensional linear Gaussian model and on a weak lensing angular power spectrum example using jax-cosmo, with comparisons to analytical and autodiff ground truth for HMC posterior sampling and Fisher matrix estimation.

Significance. The paper addresses a useful problem: obtaining Fisher scores and gradient-based Bayesian inference from simulators that are not differentiable. The derivation in Eq. (2) is correct, the least-squares regression argument is standard, and the authors provide code for reproducible experiments. The visual agreement with analytical and autodiff ground truth in both the toy and weak lensing examples is encouraging. However, the practical significance rests on controlling the bias introduced by auxiliary noise in non-decomposable models; the current evidence does not yet establish the headline claim for genuinely non-differentiable simulators.

major comments (3)

[Section 2.2, Appendix B, Eq. (5)] The auxiliary-noise procedure trains the network to estimate ∇θ~ log p(x | θ~), the score of the extended model, not the original score ∇θ log p(x | θ). As Appendix B shows, this is a convolution of the true score with P(θ | θ~, x), and the approximation is accurate only when that kernel is sharply peaked. The paper states this but provides no quantitative bound or diagnostic in terms of σ and the likelihood curvature. The weak lensing validation in §3.2 uses jax-cosmo, a differentiable simulator, with σ = 1e-3 fixed, and all comparisons are made against autodiff ground truth; no experiment uses a genuinely non-differentiable simulator. For a Gaussian location model, the extended Fisher information is (Σ + σ²I)⁻¹, so forecasts are generically biased low. To support the abstract's claim that the method extends to non-differentiable simulators, the authors should either demonstrate the method on a non-differentiable forward model or provide an explicit, testable bound or diagnostic for the smoothing bias.
[Section 3.2, Figures 2, 3, and 4] The validation is qualitative: the learned score fields, HMC posteriors, and Fisher matrix contours are compared visually, with no error bars, no quantitative score errors, and no numerical comparison of posterior or Fisher matrix elements. The paper itself reports in §3.2 that score magnitudes are underestimated by roughly 40% for implausible parameters, yet the downstream claims that the learned scores are 'suitable' for forecasting and inference rely on visual contour agreement. The authors should report metrics such as relative L2 error of the learned score in the high-likelihood region, parameter bias and coverage of the HMC posterior relative to the true-score posterior, and relative errors in Fisher matrix elements or eigenvalue ratios.
[Section 4 and Section 3.2] The claim that the Fisher score approximation is independent of the chosen training proposal is only true for the exact minimizer of the loss in Eq. (3). In finite-sample practice, the paper acknowledges that the training set contains few examples for parameters that are implausible given the data, leading to factor-of-two errors in score magnitude. Because HMC trajectories traverse regions of moderate posterior probability, the statement that this 'will not affect' inference needs quantitative support, not just the observation that the final posterior contours look similar. A quantitative comparison of the posterior moments or credible intervals would strengthen this load-bearing point.

minor comments (5)

[Footnote 1] There is a typo: 'workshop' should be 'workshop'.
[Algorithm 2] The output line writes s(x, θ~) ≈ ∇θ log p(x | θ~); the subscript on the gradient should be θ~ rather than θ.
[Section 2.2, Eq. (4)] The condition 'both covariance matrices are positive definite' should be stated as Σθ* and Σn − Σθ* being positive definite; the latter is an additional constraint on the decomposition.
[Figure 1] The axis labels in the figure are partially garbled in the manuscript rendering; please check the final PDF.
[References] The JAX-COSMO reference entry formats the arXiv identifier as a journal field; please standardize the bibliography style.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the estimator's minimizer is derived from an exact identity and validated against an independent autodiff ground truth; the auxiliary-noise caveat is an approximation gap, not a circular step.

full rationale

The derivation chain is self-contained and no fitted value is renamed as a prediction. Equation (2) proves the exact identity ∇θ log P(x|θ) = E_{P(θ*|x,θ)}[∇θ log P(θ*|θ)], and Appendix A shows that minimizing the MSE loss of Eq. (3) yields exactly that conditional expectation; hence the network target is the desired Fisher score by a theorem, not by an equivalent definition imported from the authors' prior work. The toy-model validation compares the learned field against the analytic score of the fixed full Gaussian likelihood, which is independent of the chosen latent decomposition. The weak-lensing validation compares the learned extended-model score against an autodiff ground truth from the external differentiable library jax-cosmo (Eq. 7), so the benchmark is not constructed from the network's fitted values. Appendix B explicitly states the one substantive caveat: for indecomposable models Eq. (5) shows the method estimates the smoothed score ∇θ~ log P(x|θ~), which equals the original true score only when P(θ|θ~,x) is sharply peaked. That is a stated approximation about the target model and an unvalidated claim for genuinely non-differentiable simulators, but it is a correctness/completeness concern rather than circularity. Self-citations such as Alsing & Wandelt (2018) are contextual and not load-bearing for the method's internal validity.

Assumptions & free parameters 1 free parameters · 3 assumptions · 1 invented entities

The central derivation rests on the standard Bayes least squares identity and a Markov chain assumption. For non-decomposable simulators the method introduces an auxiliary parameter noise sigma as a hand-chosen scale that controls the bias of the learned score; this is the main free parameter. No new physical entities are claimed, though the auxiliary latent variable is a modeling construct.

free parameters (1)

Auxiliary noise scale sigma = 1e-3
Chosen by hand for the weak lensing experiment; the accuracy of the indecomposable approximation depends on this being small, and no sensitivity analysis is provided.

assumptions (3)

domain assumption Markov chain theta -> theta* -> x holds
Eq 2 assumes P(x|theta*,theta)=P(x|theta*) and P(theta*|theta) has a tractable gradient; this must be satisfied by the chosen latent decomposition.
domain assumption Auxiliary noise model theta = theta~ + w makes the learned score approximate the true score when w is small
Section 2.2 and Appendix B; the approximation is accurate only when P(theta|x,theta~) is sharply peaked. The paper chooses sigma=1e-3 but does not test sensitivity.
domain assumption Gaussian likelihood with fixed covariance Sigma_fid for the weak lensing example
The validation assumes x|theta ~ N(mu(theta),Sigma_fid), enabling analytic ground-truth scores; real surveys may not satisfy this.

invented entities (1)

Auxiliary latent parameter theta~
purpose: Perturbs parameters with additive noise to create a tractable latent score for non-decomposable simulators
A statistical device, not a physical quantity; its validity is only checked indirectly through the differentiable weak lensing example, and it introduces a bias that scales with the noise.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Fisher Score Matching for Simulation-Based Forecasting and Inference." pith.science (2026). https://pith.science/paper/PBP6CVNK

@misc{pith2026250707833,
  author       = {Pith},
  title        = {Pith review of: Fisher Score Matching for Simulation-Based Forecasting and Inference},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PBP6CVNK}},
  note         = {Machine review of arXiv:2507.07833}
}

read the original abstract

We propose a method for estimating the Fisher score--the gradient of the log-likelihood with respect to model parameters--using score matching. By introducing a latent parameter model, we show that the Fisher score can be learned by training a neural network to predict latent scores via a mean squared error loss. We validate our approach on a toy linear Gaussian model and a cosmological example using a differentiable simulator. In both cases, the learned scores closely match ground truth for plausible data-parameter pairs. This method extends the ability to perform Fisher forecasts, and gradient-based Bayesian inference to simulation models, even when they are not differentiable; it therefore has broad potential for advancing cosmological analyses.

Figures

Figures reproduced from arXiv: 2507.07833 by the authors.

**Figure 1.** Comparison of the learned Fisher score (blue arrows) and the analytical Fisher score (red arrows) across the parameter space for a fixed observation. The black dot denotes the fiducial parameter value corresponding to the observation. Arrows originate at parameter locations and point in the direction of the Fisher score. The close alignment of the two sets of arrows indicates accurate recovery of the Fisher score. … view at source ↗

**Figure 3.** , the posterior contours from the estimated scores (blue) align closely with those from the true scores (red), confirming that the learned scores are suitable for Bayesian inference. As a second example, we can estimate the Fisher information matrix at any point in parameter space θ0 by computing the covariance of the score model I(θ0) = Ep(x|θ0) [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

**Figure 2.** Extrapolation of the learned Fisher score (blue arrows) vs the true Fisher score (red arrows) in the weak lensing example, for a fixed observation. The black cross indicates the fiducial parameters and the Fisher matrix contours are shown for scale. For implausible parameters far away from the fiducial, the score is underestimated by ∼ 40%, but the direction is still accurately captured. This does not affect samplin… view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Comparison of estimated Fisher matrices (blue contours) and true Fisher matrices (red contours). Each contour corresponds to a Fisher matrix estimated at its center (indicated by a cross). 4. Conclusion We propose a score-matching-based method to estimate the Fisher sc…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

30 extracted references · 1 canonical work pages

[1]

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 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...
[2]

and Wandelt , B

Alsing , J. and Wandelt , B. Generalized massive optimal data compression . , 476 0 (1): 0 L60--L64, May 2018. doi:10.1093/mnrasl/sly029

doi:10.1093/mnrasl/sly029 2018
[3]

Fast likelihood-free cosmology with neural density estimators and active learning

Alsing , J., Charnock , T., Feeney , S., and Wandelt , B. Fast likelihood-free cosmology with neural density estimators and active learning . , 488 0 (3): 0 4440--4458, September 2019. doi:10.1093/mnras/stz1960

doi:10.1093/mnras/stz1960 2019
[4]

N., Ivanov , M

Beyond-2pt Collaboration , : , Krause , E., Kobayashi , Y., Salcedo , A. N., Ivanov , M. M., Abel , T., Akitsu , K., Angulo , R. E., Cabass , G., Contarini , S., Cuesta-Lazaro , C., Hahn , C., Hamaus , N., Jeong , D., Modi , C., Nguyen , N.-M., Nishimichi , T., Paillas , E., Pellejero Iba \ n ez , M., Philcox , O. H. E., Pisani , A., Schmidt , F., Tanaka ...

arXiv doi:10.48550/arxiv.2405.02252 2024
[5]

K., Herold, M., Ziaeemehr, A., Pals, M., Gruner, T., Bischoff, S., Krouglova, N., Gao, R., Lappalainen, J

Boelts, J., Deistler, M., Gloeckler, M., Álvaro Tejero-Cantero, Lueckmann, J.-M., Moss, G., Steinbach, P., Moreau, T., Muratore, F., Linhart, J., Durkan, C., Vetter, J., Miller, B. K., Herold, M., Ziaeemehr, A., Pals, M., Gruner, T., Bischoff, S., Krouglova, N., Gao, R., Lappalainen, J. K., Mucsányi, B., Pei, F., Schulz, A., Stefanidi, Z., Rodrigues, P., ...

2025
[6]

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
[7]

Mining gold from implicit models to improve likelihood-free inference

Brehmer, J., Louppe, G., Pavez, J., and Cranmer, K. Mining gold from implicit models to improve likelihood-free inference. Proceedings of the National Academy of Sciences, 117 0 (10): 0 5242–5249, February 2020. ISSN 1091-6490. doi:10.1073/pnas.1915980117. URL http://dx.doi.org/10.1073/pnas.1915980117

doi:10.1073/pnas.1915980117 2020
[8]

and Lesgourgues, J

Brinckmann, T. and Lesgourgues, J. MontePython 3: boosted MCMC sampler and other features . 2018

work page 2018

Show all 30 references

[9]

JAX-COSMO: An End-to-End Differentiable and GPU Accelerated Cosmology Library

Campagne , J.-E., Lanusse , F., Zuntz , J., Boucaud , A., Casas , S., Karamanis , M., Kirkby , D., Lanzieri , D., Peel , A., and Li , Y. JAX-COSMO: An End-to-End Differentiable and GPU Accelerated Cosmology Library . The Open Journal of Astrophysics, 6: 0 15, April 2023. doi:1...

2023 arXiv
[10]

Charnock , T., Lavaux , G., and Wandelt , B. D. Automatic physical inference with information maximizing neural networks . , 97 0 (8): 0 083004, April 2018. doi:10.1103/PhysRevD.97.083004

2018 doi
[11]

A new approach to observational cosmology using the scattering transform

Cheng , S., Ting , Y.-S., M \'e nard , B., and Bruna , J. A new approach to observational cosmology using the scattering transform . , 499 0 (4): 0 5902--5914, December 2020. doi:10.1093/mnras/staa3165

2020 doi
[12]

The frontier of simulation-based inference

Cranmer , K., Brehmer , J., and Louppe , G. The frontier of simulation-based inference . Proceedings of the National Academy of Science, 117 0 (48): 0 30055--30062, December 2020. doi:10.1073/pnas.1912789117

2020 doi
[13]

and Mishra-Sharma , S

Cuesta-Lazaro , C. and Mishra-Sharma , S. Point cloud approach to generative modeling for galaxy surveys at the field level . , 109 0 (12): 0 123531, June 2024. doi:10.1103/PhysRevD.109.123531

2024 doi
[14]

DES Collaboration , Abbott , T. M. C., Aguena , M., Alarcon , A., Anbajagane , D., Andrade-Oliveira , F., Avila , S., Bacon , D., Becker , M. R., Bhargava , S., Blazek , J., Bocquet , S., Brooks , D., Carnero Rosell , A., Carretero , J., Castander , F. J., Chang , C., Choi , A...

2025 doi arXiv
[15]

a nen , E., Kermiche , S., Kiessling , A., Kilbinger , M., Kubik , B., Kuijken , K., K \

Euclid Collaboration , Congedo , G., Miller , L., Taylor , A. N., Cross , N., Duncan , C. A. J., Kitching , T., Martinet , N., Matthew , S., Schrabback , T., Tewes , M., Welikala , N., Aghanim , N., Amara , A., Andreon , S., Auricchio , N., Baldi , M., Bardelli , S., Bender , ...

2024
[16]

F., Sellentin , E., de Mijolla , D., and Vianello , A

Heavens , A. F., Sellentin , E., de Mijolla , D., and Vianello , A. Massive data compression for parameter-dependent covariance matrices . , 472 0 (4): 0 4244--4250, December 2017. doi:10.1093/mnras/stx2326

2017 doi
[17]

J., Chartier , N., Cuesta-Lazaro , C., Ding , S., Lapel , A., Lemos , P., Lovell , C

Ho , M., Bartlett , D. J., Chartier , N., Cuesta-Lazaro , C., Ding , S., Lapel , A., Lemos , P., Lovell , C. C., Makinen , T. L., Modi , C., Pandya , V., Pandey , S., Perez , L. A., Wandelt , B., and Bryan , G. L. LtU-ILI: An All-in-One Framework for Implicit Inference in Astr...

2024 doi
[18]

and Seljak , U

Jain , B. and Seljak , U. Cosmological Model Predictions for Weak Lensing: Linear and Nonlinear Regimes . , 484 0 (2): 0 560--573, July 1997. doi:10.1086/304372

1997 doi
[19]

H., Hahn , C., Ho , S., Eickenberg , M., Hou , J., Massara , E., Modi , C., Moradinezhad Dizgah , A., R \'e galdo-Saint Blancard , B., and Spergel , D

Lemos , P., Parker , L. H., Hahn , C., Ho , S., Eickenberg , M., Hou , J., Massara , E., Modi , C., Moradinezhad Dizgah , A., R \'e galdo-Saint Blancard , B., and Spergel , D. SimBIG: Field-level Simulation-based Inference of Large-scale Structure . In Machine Learning for Ast...

2023 doi arXiv
[20]

pmwd: A Differentiable Cosmological Particle-Mesh N -body Library

Li , Y., Lu , L., Modi , C., Jamieson , D., Zhang , Y., Feng , Y., Zhou , W., Pok Kwan , N., Lanusse , F., and Greengard , L. pmwd: A Differentiable Cosmological Particle-Mesh N -body Library . arXiv e-prints, art. arXiv:2211.09958, November 2022. doi:10.48550/arXiv.2211.09958

2022 doi arXiv
[21]

L., Sui , C., Wandelt , B

Makinen , T. L., Sui , C., Wandelt , B. D., Porqueres , N., and Heavens , A. Hybrid Summary Statistics . arXiv e-prints, art. arXiv:2410.07548, October 2024. doi:10.48550/arXiv.2410.07548

2024 doi
[22]

Mudur , N., Cuesta-Lazaro , C., and Finkbeiner , D. P. Diffusion-HMC: Parameter Inference with Diffusion-model-driven Hamiltonian Monte Carlo . , 978 0 (1): 0 64, January 2025. doi:10.3847/1538-4357/ad8bc3

2025 doi
[23]

J., Barreiro , R

Planck Collaboration , Aghanim , N., Akrami , Y., Ashdown , M., Aumont , J., Baccigalupi , C., Ballardini , M., Banday , A. J., Barreiro , R. B., Bartolo , N., Basak , S., Battye , R., Benabed , K., Bernard , J. P., Bersanelli , M., Bielewicz , P., Bock , J. J., Bond , J. R., ...

2018
[24]

Sequential Neural Score Estimation: Likelihood-Free Inference with Conditional Score Based Diffusion Models

Sharrock , L., Simons , J., Liu , S., and Beaumont , M. Sequential Neural Score Estimation: Likelihood-Free Inference with Conditional Score Based Diffusion Models . arXiv e-prints, art. arXiv:2210.04872, October 2022. doi:10.48550/arXiv.2210.04872

2022 doi arXiv
[25]

and Ermon , S

Song , Y. and Ermon , S. Generative Modeling by Estimating Gradients of the Data Distribution . arXiv e-prints, art. arXiv:1907.05600, July 2019. doi:10.48550/arXiv.1907.05600

1907 doi arXiv
[26]

P., Kumar , A., Ermon , S., and Poole , B

Song , Y., Sohl-Dickstein , J., Kingma , D. P., Kumar , A., Ermon , S., and Poole , B. Score-Based Generative Modeling through Stochastic Differential Equations . arXiv e-prints, art. arXiv:2011.13456, November 2020. doi:10.48550/arXiv.2011.13456

2011 doi arXiv
[27]

J., Battye , R

Square Kilometre Array Cosmology Science Working Group , Bacon , D. J., Battye , R. A., Bull , P., Camera , S., Ferreira , P. G., Harrison , I., Parkinson , D., Pourtsidou , A., Santos , M. G., Wolz , L., Abdalla , F., Akrami , Y., Alonso , D., Andrianomena , S., Ballardini , ...

2018
[28]

N., and Heavens , A

Tegmark , M., Taylor , A. N., and Heavens , A. F. Karhunen-Lo \`e ve Eigenvalue Problems in Cosmology: How Should We Tackle Large Data Sets? , 480 0 (1): 0 22--35, May 1997. doi:10.1086/303939

1997 doi
[29]

and Lewis , A

Torrado , J. and Lewis , A. Cobaya: code for Bayesian analysis of hierarchical physical models . , 2021 0 (5): 0 057, May 2021. doi:10.1088/1475-7516/2021/05/057

2021 doi
[30]

Zhao , X., Mao , Y., Cheng , C., and Wandelt , B. D. Simulation-based Inference of Reionization Parameters from 3D Tomographic 21 cm Light-cone Images . , 926 0 (2): 0 151, February 2022. doi:10.3847/1538-4357/ac457d

2022 doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

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 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

[2] [2]

and Wandelt , B

Alsing , J. and Wandelt , B. Generalized massive optimal data compression . , 476 0 (1): 0 L60--L64, May 2018. doi:10.1093/mnrasl/sly029

doi:10.1093/mnrasl/sly029 2018

[3] [3]

Fast likelihood-free cosmology with neural density estimators and active learning

Alsing , J., Charnock , T., Feeney , S., and Wandelt , B. Fast likelihood-free cosmology with neural density estimators and active learning . , 488 0 (3): 0 4440--4458, September 2019. doi:10.1093/mnras/stz1960

doi:10.1093/mnras/stz1960 2019

[4] [4]

N., Ivanov , M

Beyond-2pt Collaboration , : , Krause , E., Kobayashi , Y., Salcedo , A. N., Ivanov , M. M., Abel , T., Akitsu , K., Angulo , R. E., Cabass , G., Contarini , S., Cuesta-Lazaro , C., Hahn , C., Hamaus , N., Jeong , D., Modi , C., Nguyen , N.-M., Nishimichi , T., Paillas , E., Pellejero Iba \ n ez , M., Philcox , O. H. E., Pisani , A., Schmidt , F., Tanaka ...

arXiv doi:10.48550/arxiv.2405.02252 2024

[5] [5]

K., Herold, M., Ziaeemehr, A., Pals, M., Gruner, T., Bischoff, S., Krouglova, N., Gao, R., Lappalainen, J

Boelts, J., Deistler, M., Gloeckler, M., Álvaro Tejero-Cantero, Lueckmann, J.-M., Moss, G., Steinbach, P., Moreau, T., Muratore, F., Linhart, J., Durkan, C., Vetter, J., Miller, B. K., Herold, M., Ziaeemehr, A., Pals, M., Gruner, T., Bischoff, S., Krouglova, N., Gao, R., Lappalainen, J. K., Mucsányi, B., Pei, F., Schulz, A., Stefanidi, Z., Rodrigues, P., ...

2025

[6] [6]

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

[7] [7]

Mining gold from implicit models to improve likelihood-free inference

Brehmer, J., Louppe, G., Pavez, J., and Cranmer, K. Mining gold from implicit models to improve likelihood-free inference. Proceedings of the National Academy of Sciences, 117 0 (10): 0 5242–5249, February 2020. ISSN 1091-6490. doi:10.1073/pnas.1915980117. URL http://dx.doi.org/10.1073/pnas.1915980117

doi:10.1073/pnas.1915980117 2020

[8] [8]

and Lesgourgues, J

Brinckmann, T. and Lesgourgues, J. MontePython 3: boosted MCMC sampler and other features . 2018

work page 2018