Cosmological parameter estimation from large-scale structure deep learning

Cristiano G. Sabiu; Haitao Miao; Jaime Forero-Romero; Miaoxin Liu; Shuyang Pan; Xiao-Dong Li; Zhigang Li

REVIEW 4 major objections 5 minor 90 references

Cosmological parameter estimation from large-scale structure deep learning

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A lightweight convolutional neural network trained on simulated dark-matter density cubes recovers $\Omega_m$ and $\sigma_8$ with uncertainties several times smaller than two-point clustering analysis, after a persistent bias is corrected.

desk verdict Solid incremental CNN-for-cosmology paper with an honest bias discussion, but the headline precision claims rest on a bias correction fitted to labels and on an apples-to-oranges comparison with full-box 2pcf errors. read the letter →

arxiv 1908.10590 v5 pith:JURP6GP4 submitted 2019-08-28 astro-ph.CO gr-qc

Shuyang Pan , Miaoxin Liu , Jaime Forero-Romero , Cristiano G. Sabiu , Zhigang Li , Haitao Miao , Xiao-Dong Li This is my paper · ORCID

classification astro-ph.COgr-qc

keywords convolutionalneuralnetworkcosmologicalparameterestimationlarge-scalestructuredarkmatterdensityfieldOmega_msigma_8biascorrectiontwo-pointcorrelationfunction

topics Dark Matter

open problems Dark Matter

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 argues that a deliberately small convolutional neural network can estimate two cosmological parameters — the matter density $\Omega_m$ and the amplitude of matter fluctuations $\sigma_8$ — directly from $32^3$-voxel chunks of simulated dark-matter density fields. Trained on 465 COLA simulations spanning a flat $\Lambda$CDM parameter grid, the network reaches statistical uncertainties of $\delta\Omega_m=0.0015$ and $\delta\sigma_8=0.0029$ after a bias correction. The paper reports that these constraints are 3.5/2.3 and 19/11 times more precise than those from two-point correlation function analysis over the clustering ranges $0$ to $130$ and $10$ to $130\,h^{-1}\,\mathrm{Mpc}$, respectively. It also argues that the network tolerates masking, random noise, rotation, reflection, and resolution changes, while smoothing and global density variations shift predictions noticeably. The point of the exercise is to show that deep learning can compete with, and possibly beat, conventional summary statistics on the nonlinear information in the cosmic web.

What carries the argument

The load-bearing object is the CNN architecture itself: three convolution layers with 32, 64, and 128 filters, batch normalization, pooling, three dense layers, and ReLU activations, trained with mean squared error loss and Adam optimization on normalized density fields. Each input cube of $32^3$ voxels covers $(64\,h^{-1}\,\mathrm{Mpc})^3$; the first convolution mixes information over $(6\,h^{-1}\,\mathrm{Mpc})^3$ and subsequent layers expand the receptive field to scales relevant for $\sigma_8$. The second piece of machinery is the bias-correction step: a third-order polynomial in $\Omega_m$ and $\sigma_8$, fitted to the prediction bias across the multi-cosmology grid, is subtracted from the raw outputs. That correction is what turns the network's biased raw predictions into the quoted unbiased-looking constraints.

What would settle it

Train the identical network on the same 465 boxes, freeze it, and apply the polynomial correction fitted to a grid that excludes a held-out cosmology; then feed 500 new boxes at that cosmology and compare corrected predictions with the truth. If the residual scatter or central offset exceeds the quoted errors, or if the $\sigma_8$ offset grows beyond $1\sigma$, the claimed precision depends on the correction rather than on the network's reading of the density field.

Watch

Extended reading notes

Core claim

The central discovery claimed is that a network with three convolutional layers and three dense layers, fed by $32^3$ voxel subcubes from $(256\,h^{-1}\,\mathrm{Mpc})^3$ dark matter boxes, can recover $\Omega_m$ and $\sigma_8$ with errors $\delta\Omega_m=0.0015$ and $\delta\sigma_8=0.0029$ on 500 single-cosmology test volumes, after subtracting a polynomial bias model fitted on the 465 multi-cosmology samples. The quoted results are corrected predictions, not raw network outputs; the raw network underestimates $\sigma_8$ by about 2.5%, a bias that does not disappear with more training. The corrected central values, $\Omega_m=0.3073$ and $\sigma_8=0.8178$, are consistent with the ground truth $(0.3071,0.8228)$ at the $1\sigma$ level for $\Omega_m$, while the paper notes a residual near-$1\sigma$ offset in $\sigma_8$. The paper positions this as the first demonstration that a light CNN can outperform two-point clustering emulators on these parameters.

Load-bearing premise

The quoted errors are for predictions after a polynomial bias correction that is fitted using the true values of the test cosmologies; if that correction does not generalize to a new cosmology, the residual scatter it hides, including the paper's own noted near-$1\sigma$ offset in $\sigma_8$, is not a valid estimate of parameter uncertainty.

Editorial extensions

If this is right

If the quoted errors hold, a network trained on $32^3$-voxel subcubes gives $\Omega_m$ and $\sigma_8$ constraints respectively 3.5/2.3 and 19/11 times tighter than two-point correlation function emulators over the same simulated volumes.
Scaling the training or observed volume to $(512\,h^{-1}\,\mathrm{Mpc})^3$ or $(1\,h^{-1}\,\mathrm{Gpc})^3$ should cut the statistical errors by factors of roughly 3 and 8 respectively, per the paper's extrapolation.
The persistent, training-resistant bias means architecture choices, especially the capacity of the dense regression layers, set a floor on accuracy; simply training longer does not remove it.
For real observations, smoothing and global depth variations would have to be controlled or modeled; masking, random noise, rotation, and reflection do not by themselves degrade predictions.

Reading between the lines

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

The polynomial correction is fitted using the true parameter values of the test grid, so the quoted error bars are conditional on knowing the answer; a forward application to real data would need the correction calibrated from mocks with an assumed cosmology, and that calibration error is not included in the quoted uncertainties.
The residual near-$1\sigma$ offset in $\sigma_8$ suggests that the quoted 0.0029 uncertainty may underestimate the systematic floor; testing on a cosmology off the training grid would reveal whether the correction generalizes.
If the network really reads nonlinear scales down to $6\,h^{-1}\,\mathrm{Mpc}$, the same architecture could be pointed at other parameters that imprint on small-scale structure, such as the dark-energy equation of state or modified gravity, rather than only $\Omega_m$ and $\sigma_8$.
The tolerance to missing voxels hints that the network is insensitive to the loss of a few cells, but the sensitivity to smoothing means the network may key on sharp small-scale features; adversarial perturbations of those features would map what the network actually uses.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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

Quick take: this is a reasonable incremental step in the CNN-for-LSS program, not a breakthrough. The genuinely new pieces are the light 32^3-voxel architecture, the explicit bias-correction scheme, and the first systematic error-tolerance tests for this kind of network. They also deserve credit for reporting the persistent bias honestly and for noting the residual ~1 sigma offset on sigma_8 after correction. The convergence and architecture tests are useful calibration work.

The soft spots are real, and one is load-bearing. The 2pcf comparison in Table 1 compares CNN errors derived from 32^3 subcubes (64 Mpc on a side) with 2pcf errors derived from the full (256 Mpc)^3 box. The paper never says whether the 500 single-cosmology samples are full boxes, independent subcubes, or multiple subcubes per box. If the CNN numbers are per-subcube scatter, the factors 3.5/2.3 and 19/11 do not compare like with like unless you scale by volume. That doesn't necessarily kill the conclusion, but it needs to be stated and re-derived.

The bias correction is the second issue. The polynomial is fitted to the labeled multi-cosmology test set, which sits on the same grid as training, and its uncertainty is not propagated. The residual ~1 sigma bias on sigma_8 shows the correction is not perfect; the quoted +/-0.0029 could therefore be optimistic by an unknown systematic floor. They acknowledge the residual, but the final numbers are still presented as the paper's headline result.

Third, the abstract overstates the robustness tests. The text says 1% smoothing shifts estimates by ~2 sigma and 3% smoothing is "disastrous," yet the abstract lists smoothing among the effects the network is robust against. That inconsistency should be fixed before publication.

Finally, no code or data are released. The COLA simulations and network architecture are described well enough to reproduce, but shipping the trained weights or the catalog would raise the confidence level considerably.

Bottom line: the paper deserves a serious referee, but it needs a major revision to clarify the volume comparison, propagate or bound the bias-correction uncertainty, and align the abstract with the body. I would not cite it in its current form, but I would be interested in a revised version.

Recommendation: send to peer review with a request for a careful response on the volume mismatch and the bias-correction error budget.

Referee Report

4 major / 5 minor

Summary. This paper presents a convolutional neural network that takes 32^3-voxel (64 h^{-1} Mpc)^3 subcubes of 128^3-voxel COLA dark matter density fields and predicts Ω_m and σ_8. The training set consists of 465 realizations on a 31×15 grid in (Ω_m, 10^9 A_s), and two test sets are used: 500 single-cosmology samples at (Ω_m, σ_8) = (0.3072, 0.8228) and a multi-cosmology set on the same grid as the training set. A persistent bias in the raw predictions is corrected with a third-order polynomial fitted to the multi-cosmology test set. The headline result, Eq. (2), reports Ω_m = 0.3073 ± 0.0015 and σ_8 = 0.8178 ± 0.0029, and the paper claims these are 3.5/2.3 and 19/11 times more precise than 2pcf constraints from the full simulation boxes. The paper also presents architecture variants, learning curves, and preliminary error-tolerance tests.

Significance. If the headline precision were established, the paper would make a useful contribution: it demonstrates that a lightweight CNN on small subvolumes can yield tight constraints, and its systematic robustness tests are a useful first look. The paper is transparent about the persistent bias, and its limitation statements in Section 5 acknowledge that the bias correction is not fully satisfactory. The convergence tests and architecture ablations are strengths, since they provide evidence that training is stable and that the architecture choices are not finely tuned. However, the central claim that the CNN outperforms 2pcf by the quoted factors is not currently supported because the comparison appears to mix subcube-level CNN errors with full-box 2pcf errors, and the calibration procedure contributes unquantified uncertainty.

major comments (4)

[Sections 2, 3, and 4.4; Table 1] The network input is a 32^3-voxel subcube, i.e. a (64 h^{-1} Mpc)^3 volume, but Section 2 does not state whether the 500 single-cosmology test 'samples' are 500 independent full boxes, 500 independent subcubes, or subcubes drawn from a smaller number of parent boxes. The scatter reported in Eq. (2) is therefore, on the face of the paper, a per-subcube scatter. Table 1 and the surrounding text compare this scatter with 2pcf constraints derived from full (256 h^{-1} Mpc)^3 boxes, so the quoted factors 3.5/2.3 and 19/11 compare different volumes. Please state exactly how the 500 test predictions were constructed and either perform the comparison at matched volume (for example, by averaging subcube predictions within each full box before computing the scatter, or by measuring 2pcf on the same subcubes) or rescale the errors with the appropriate volume factor.
[Sections 4.3 and 4.4] The bias-correction polynomial is fitted to the multi-cosmology test set, whose parameter values lie on the same 31×15 grid as the training set, using the known true parameter values as inputs. The uncertainty of this polynomial fit is not propagated into the uncertainties quoted in Eq. (2), and the paper itself notes a residual ≈1σ bias on σ_8. The quoted error bars are therefore conditional on the calibration being exact, and the method as presented is not a forward prediction applicable to real data. Please quantify the calibration uncertainty and state how the residual bias affects the central values and error budget.
[Table 1 vs. Eq. (2)] Table 1 lists the CNN relative error on σ_8 as 0.0053, but Eq. (2) gives σ_8 = 0.8178 ± 0.0029, i.e. a relative error of 0.0035. The comparison factors 3.5/2.3 and 19/11 quoted in Section 4.4 appear to be computed from Table 1 rather than from Eq. (2). Please reconcile these numbers and state the exact definition of 'relative error' used in Table 1, since the footnote says it includes both statistical error and bias, which is not what Eq. (2) reports.
[Section 4.4 and Table 1] The 2pcf comparison is under-specified. The text says that 'the 2pcf constraints on parameters are derived by measuring the shape and amplitude of the 2pcfs using samples in the many cosmologies, to build an emulator,' but it does not describe the number of samples used for the emulator, the covariance matrix, the fitting procedure, or whether the quoted 2pcf errors correspond to the same volume as the CNN predictions. Without these details the factors 3.5/2.3 and 19/11 cannot be assessed or reproduced. Please provide the full methodology or remove the quantitative comparison.

minor comments (5)

[Section 3.3 and Figure 3] The text says the fully connected layers have 1024, 256, and 2 neurons, while the Figure 3 caption says 1028, 24, and 2 neurons; please correct the inconsistency.
[Introduction] The introduction contains placeholder citations and garbled author names, including '?Lucie-Smith et al. 2018', 'Trster et al. 2019', and 'Mnchmeyer & Smith 2019'; these should be fixed.
[Figure 10] Figure 10 has heavily garbled axis labels and legend text, such as '0m', 'Maski%g', 'Grou%d trut', and 'Simulatin res&luti&n'; the figure needs to be regenerated with clean labels.
[Section 4.5] The error-tolerance tests use 64 subcubes split from a single 128^3 box, so the 64 parameter estimates are not independent; statements such as 'errors unchanged' should be qualified accordingly.
[Throughout] There are numerous typographical errors, including 'convultion', 'volxel', 'Origianl', and 'relfection'; a careful proofreading pass is needed.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: bias correction is held-out calibration, and the final scatter is measured on independent samples.

full rationale

The derivation chain is not circular. The CNN is trained on simulated volumes with known parameters, and the single-cosmology test set is a separate held-out sample generated with different initial conditions. The only step with a calibration flavor is the Section 4.3 polynomial bias correction. The paper explicitly avoids self-correction by using the multi-cosmology samples for the regression: 'To avoid self-correction, the regression is derived using the multi-cosmology samples, which have no overlapping from the single-cosmology samples.' The final constraints in Eq. (2) are then evaluated on the 500 single-cosmology samples. Because those 500 samples share the same true cosmology, the fitted polynomial contributes only a constant shift when applied to them; it cannot manufacture the quoted scatter, so the precision claim is not an artifact of the fit. The fact that the correction requires ground-truth values is a practical limitation (the method is not yet a forward estimator for real data), and the subcube-versus-full-box volume question is a comparison-validity concern, but neither reduces the claimed result to its own input. Self-citations in the introduction are contextual and not load-bearing.

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

The central claim rests on the COLA simulation suite, the fixed background parameters, the uniform grid prior, the chosen CNN architecture, and especially the bias-correction polynomial fitted to labeled test data. No new physical entities are introduced.

free parameters (2)

Bias-correction polynomial coefficients = Third-order polynomial in Omega_m and sigma_8; e.g., Delta_Omega_m ≈ -0.53 Omega_m - 0.48 Omega_m^2 + 0.30 Omega_m^3…
Fitted to the multi-cosmology test samples using their known ground-truth parameters, then subtracted to produce the final reported constraints (Eq. 2). Its uncertainty is not propagated into the quoted errors.
Network hyperparameters = 32/64/128 conv filters; 1024/256/2 dense neurons; 20% dropout; sgd vs Adam; max vs average pooling; training epoch…
Chosen by hand during architecture search; the choices affect the bias level and the reported precision, and the paper does not provide a sensitivity analysis over all of them.

assumptions (5)

domain assumption COLA simulations with 40 timesteps and 2LPT initial conditions at zi=39 accurately represent nonlinear dark matter clustering on the scales used.
The training and test data are all COLA outputs (Section 2); if COLA is inaccurate, the learned mapping and the quoted errors do not transfer to real or N-body density fields.
domain assumption The fixed cosmological parameters Omega_b=0.048206, h=0.6777, and ns=0.96 are correct and unchanged.
Taken from MultiDark Planck simulations (Section 2); the network is never trained for variations in these parameters, so its estimates assume them.
domain assumption A uniform prior over the 31x15 grid in Omega_m and As is adequate, and the resulting Omega_m-sigma_8 degeneracy does not bias training.
Section 2 notes the prior 'may influence the performance... and this influence is unchecked in this work.'
ad hoc to paper The bias-correction polynomial fitted on the multi-cosmology grid is valid at the single-cosmology test point and elsewhere in the parameter space.
Section 4.4 applies the correction from Figure 8 to the 500 single-cosmology samples; the paper acknowledges the sigma_8 prediction still carries a ~1 sigma residual bias, showing this assumption is not fully satisfied.
domain assumption A 32^3-voxel subcube of the density field contains sufficient information to infer Omega_m and sigma_8.
Section 3 justifies the input size by focusing on scales below about 50 h^-1 Mpc; if the relevant cosmological information lives on larger scales, the network cannot access it.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Cosmological parameter estimation from large-scale structure deep learning." pith.science (2026). https://pith.science/paper/JURP6GP4

@misc{pith2026190810590,
  author       = {Pith},
  title        = {Pith review of: Cosmological parameter estimation from large-scale structure deep learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JURP6GP4}},
  note         = {Machine review of arXiv:1908.10590}
}

abstract

We propose a light-weight deep convolutional neural network (CNN) to estimate the cosmological parameters from simulated 3-dimensional dark matter distributions with high accuracy. The training set is based on 465 realizations of a cubic box with a side length of $256\ h^{-1}\ \rm Mpc$, sampled with $128^3$ particles interpolated over a cubic grid of $128^3$ voxels. These volumes have cosmological parameters varying within the flat $\Lambda$CDM parameter space of $0.16 \leq \Omega_m \leq 0.46$ and $2.0 \leq 10^9 A_s \leq 2.3$. The neural network takes as an input cubes with $32^3$ voxels and has three convolution layers, three dense layers, together with some batch normalization and pooling layers. In the final predictions from the network we find a $2.5\%$ bias on the primordial amplitude $\sigma_8$ that can not easily be resolved by continued training. We correct this bias to obtain unprecedented accuracy in the cosmological parameter estimation with statistical uncertainties of $\delta \Omega_m$=0.0015 and $\delta \sigma_8$=0.0029, which are several times better than the results of previous CNN works. Compared with a 2-point analysis method using clustering region of 0-130 and 10-130 $h^{-1}$ Mpc, the CNN constraints are several times and an order of magnitude more precise, respectively. Finally, we conduct preliminary checks of the error-tolerance abilities of the neural network, and find that it exhibits robustness against smoothing, masking, random noise, global variation, rotation, reflection, and simulation resolution. Those effects are well understood in typical clustering analysis, but had not been tested before for the CNN approach. Our work shows that CNN can be more promising than people expected in deriving tight cosmological constraints from the cosmic large scale structure.

Figures

Figures reproduced from arXiv: 1908.10590 by the authors.

**Figure 1.** The density field (left) and particle distribution (right) in three cosmologies (Ωm, As, σ8) = (0.16, 2, 0.43), (0.26, 2.16, 0.72),(0.36, 2.0, 0.89), selected from the training sample. We plot the 2D distribution, with the third dimension restricted to a thin slice 0h −1Mpc < z < 2h −1Mpc. The clustering strength is enhanced when increasing Ωm or As, making the structures more “compact”. We train neural networks to … view at source ↗

**Figure 2.** Ωm and σ8 values for the 465 training samples and single-cosmology test samples. The multi-cosmology test samples have exactly the same values of Ωm and σ8 as those in training samples. In Figures 4,5, we show how the CNN works. Step by step, the features are extracted by the three layers, and become more and more condensed. Different filters identify different features. With a large number of filters we are able to… view at source ↗

**Figure 3.** The architecture of our neural network. A cube having 323 voxels is fed to the network. The three convolution layers have 32, 64, 128 filters, respectively. Beside each convolution layer, a batch normalization layer is added before it to normalize the distribution (so that to enhance the stability), and a pooling layer is placed after it to decrease the size of the output. After that, we got 128 × 2 3 voxels contain… view at source ↗

Figures from the paper (7 more)

**Figure 4.** Figure 4: Layer-by-layer outputs of the CNN when fed by a sample with cosmology parameters (Ωm, As, σ8) = (0.26, 2.16, 0.72).The many filters, determined by the 896/55,360/221,312 trainable parameters in the three convolutions layers, can capture various types of features. The f…

**Figure 5.** Figure 5: Same as [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

**Figure 6.** Figure 6: Learning curve using different architectures. First panel: two runs using the default options reaches convergence after 160 epochs. Second panel: decreasing the number of CNN filters or dense neurons by 50%, no significant change in the performance. Third panel: among …

**Figure 7.** Figure 7: Test of a CNN architecture (sgd) on a multi-cosmology grid. There is a strong degeneracy between Ωm and σ8. Left panel: Ground truth and CNN predictions of Ωm and σ8, in the 2-d parameter space. The black lines show the difference between them. The bias is larger at th…

**Figure 8.** Figure 8: Distribution of the systematic bias in the CNN predicted Ωm and σ8 (denoted as ∆Ωm and ∆σ8). Very roughly, in the parameter space we studied, there is |∆Ωm| . 0.03 and |∆σ8| . 0.05, with mean value of ¯|∆Ωm| = 0.01 and ¯|∆σ8| = 0.018. In practice one can calibrate the …

**Figure 9.** Figure 9: Test of a CNN architecture (sgd) on the single-cosmology samples. Left panel: Ground truth (red star) and CNN predictions (blue dots) of Ωm and σ8, in the 2-d parameter space. The CNN well predicts the values of Ωm, but has a bias in estimating σ8. Middle and Right pan…

**Figure 10.** Figure 10: Error-tolerance tests. A 3% smoothing or 10% global variation leads to considerable change in the predicted results (∼ 2σ shift in central values, ∼ 100% enlarged errors). 1% smoothing, 5% global variation, and 10% change in the simulation’s resolution mildly affect t…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

90 extracted references · 47 canonical work pages

[1]

P., Abbott, R., Abbott, T

Abbott, B. P., Abbott, R., Abbott, T. D., et al. 2017, Nature, 551, 85

2017
[2]

A., Aghanim, N., Arnaud, M., et al

Ade, P. A., Aghanim, N., Arnaud, M., et al. 2016, Astronomy & Astrophysics, 594, A13

2016
[3]

2017, Monthly Notices of the Royal Astronomical Society, 470, 2617

Alam, S., Ata, M., Bailey, S., et al. 2017, Monthly Notices of the Royal Astronomical Society, 470, 2617

2017
[4]

1979, Nature, 281, 358

Alcock, C., & Paczy´ nski, B. 1979, Nature, 281, 358

1979
[5]

2012, Monthly Notices of the Royal Astronomical Society, 427, 3435

Anderson, L., Aubourg, E., Bailey, S., et al. 2012, Monthly Notices of the Royal Astronomical Society, 427, 3435

2012
[6]

2014, Monthly Notices of the Royal Astronomical Society, 441, 24

Anderson, L., Aubourg, ´E., Bailey, S., et al. 2014, Monthly Notices of the Royal Astronomical Society, 441, 24

2014
[7]

1996, Monthly Notices of the Royal Astronomical Society, 282, 877

Ballinger, W., Peacock, J., & Heavens, A. 1996, Monthly Notices of the Royal Astronomical Society, 282, 877

1996
[8]

M., Bond, J

Bardeen, J. M., Bond, J. R., Kaiser, N., & Szalay, A. S. 1986, ApJ, 304, 15

1986

Show all 90 references

[9]

2019, Mon

Berger, P., & Stein, G. 2019, Mon. Not. Roy. Astron. Soc., 482, 2861

2019
[10]

2011, Monthly Notices of the Royal Astronomical Society, 416, 3017

Beutler, F., Blake, C., Colless, M., et al. 2011, Monthly Notices of the Royal Astronomical Society, 416, 3017

2011
[11]

2016, Monthly Notices of the Royal Astronomical Society, 466, 2242

Beutler, F., Seo, H.-J., Saito, S., et al. 2016, Monthly Notices of the Royal Astronomical Society, 466, 2242

2016
[12]

2003, ApJ, 594, 665

Blake, C., & Glazebrook, K. 2003, ApJ, 594, 665

2003
[13]

Caldeira, J., Wu, W. L. K., Nord, B., et al. 2018, arXiv:1810.01483

2018 arXiv
[14]

2019, arXiv:1903.10563

Carleo, G., Cirac, I., Cranmer, K., et al. 2019, arXiv:1903.10563

2019 arXiv
[15]

2019, arXiv:1905.06958

Chardin, J., Uhlrich, G., Aubert, D., et al. 2019, arXiv:1905.06958

2019 arXiv
[16]

2017, Monthly Notices of the Royal Astronomical Society, 471, 2370

Chuang, C.-H., Pellejero-Ibanez, M., Rodr´ ıguez-Torres, S., et al. 2017, Monthly Notices of the Royal Astronomical Society, 471, 2370

2017
[17]

2003, arXiv:astro-ph/0306581 16 de Lapparent, V., Geller, M

Colless, M., et al. 2003, arXiv:astro-ph/0306581 16 de Lapparent, V., Geller, M. J., & Huchra, J. P. 1986, ApJL, 302, L1

2003 arXiv
[18]

2019, Phys

Prix, R. 2019, Phys. Rev., D100, 044009

2019
[19]

1980, ApJ, 236, 351

Dressler, A. 1980, ApJ, 236, 351

1980
[20]

J., Hu, W., & Tegmark, M

Eisenstein, D. J., Hu, W., & Tegmark, M. 1998, Astrophys. J., 504, L57

1998
[21]

J., et al

Eisenstein, D. J., et al. 2005, Astrophys. J., 633, 560

2005
[22]

2019, Mon

Fang, F., Forero-Romero, J., Rossi, G., Li, X.-D., & Feng, L.-L. 2019, Mon. Not. Roy. Astron. Soc., 485, 5276

2019
[23]

2019, arXiv:1906.03156

Fluri, J., Kacprzak, T., Lucchi, A., et al. 2019, arXiv:1906.03156

2019 arXiv
[24]

D., Kilbertus, N., Harry, I., & Schlkopf, B

Gebhard, T. D., Kilbertus, N., Harry, I., & Schlkopf, B. 2019

2019
[25]

2019, Mon

Gillet, N., Mesinger, A., Greig, B., Liu, A., & Ucci, G. 2019, Mon. Not. Roy. Astron. Soc., 484, 282

2019
[26]

Gupta, A., Matilla, J. M. Z., Hsu, D., & Haiman, Z. 2018, Phys. Rev., D97, 103515

2018
[27]

2014, A&A, 566, A108

Guzzo, L., Scodeggio, M., Garilli, B., et al. 2014, A&A, 566, A108

2014
[28]

2019a, arXiv:1907.07787

Hassan, S., Andrianomena, S., & Doughty, C. 2019a, arXiv:1907.07787

1907 arXiv
[29]

2019, Proc

He, S., Li, Y., Feng, Y., et al. 2019, Proc. Nat. Acad. Sci., 116, 13825

2019
[30]

P., Macri, L

Huchra, J. P., Macri, L. M., Masters, K. L., et al. 2012, ApJS, 199, 26

2012
[31]

Ishida, E. E. O., et al. 2019, Mon. Not. Roy. Astron. Soc., 483, 2 Jeﬀrey, N., Lanusse, F., Lahav, O., & Starck, J.-L. 2019, arXiv:1908.00543

2019 arXiv
[32]

McEwen, J. D. 2019, Mon. Not. Roy. Astron. Soc., 483, 2907

2019
[33]

1987, Monthly Notices of the Royal Astronomical Society, 227, 1

Kaiser, N. 1987, Monthly Notices of the Royal Astronomical Society, 227, 1

1987
[34]

P., & Ba, J

Kingma, D. P., & Ba, J. 2014, arXiv e-prints, arXiv:1412.6980

2014 arXiv
[35]

2016, Monthly Notices of the Royal Astronomical Society, 457, 4340

Klypin, A., Yepes, G., Gottl¨ ober, S., Prada, F., & Hess, S. 2016, Monthly Notices of the Royal Astronomical Society, 457, 4340

2016
[36]

2016, Mon

Koda, J., Blake, C., Beutler, F., Kazin, E., & Marin, F. 2016, Mon. Not. Roy. Astron. Soc., 459, 2118 La Plante, P., & Ntampaka, M. 2018, Astrophys. J., 810, 110

2016
[37]

Lavaux, G., & Wandelt, B. D. 2012, The Astrophysical Journal, 754, 109

2012
[38]

2019a, arXiv:1907.00568

Li, S.-Y., Li, Y.-L., & Zhang, T.-J. 2019a, arXiv:1907.00568

1907 arXiv
[39]

E., & Kim, J

Li, X.-D., Park, C., Forero-Romero, J. E., & Kim, J. 2014, ApJ, 796, 137

2014
[40]

E., & Kim, J

Li, X.-D., Park, C., Forero-Romero, J. E., & Kim, J. 2014, The Astrophysical Journal, 796, 137

2014
[41]

G., & Kim, J

Li, X.-D., Park, C., Sabiu, C. G., & Kim, J. 2015, Monthly Notices of the Royal Astronomical Society, 450, 807

2015
[42]

G., et al

Li, X.-D., Park, C., Sabiu, C. G., et al. 2017, Astrophys. J., 844, 91

2017
[43]

G., Park, C., et al

Li, X.-D., Sabiu, C. G., Park, C., et al. 2018, Astrophys. J., 856, 88

2018
[44]

Winter, M. K. 2016, Astrophys. J. Suppl., 225, 31

2016
[45]

V., & Pontzen, A

Lucie-Smith, L., Peiris, H. V., & Pontzen, A. 2019, arXiv:1906.06339

2019 arXiv
[46]

V., Pontzen, A., & Lochner, M

Lucie-Smith, L., Peiris, H. V., Pontzen, A., & Lochner, M. 2018, Mon. Not. Roy. Astron. Soc., 479, 3405

2018
[47]

2010, Nature, 468, 539

Marinoni, C., & Buzzi, A. 2010, Nature, 468, 539

2010
[48]

2018, arXiv:1808.04728

Mathuriya, A., et al. 2018, arXiv:1808.04728

2018 arXiv
[49]

2019, Phys

Mehta, P., Bukov, M., Wang, C.-H., et al. 2019, Phys. Rept., 810, 1

2019
[50]

2019, Mon

Merten, J., Giocoli, C., Baldi, M., et al. 2019, Mon. Not. Roy. Astron. Soc., 487, 104

2019
[51]

2011, Communications in Theoretical Physics, 56, 525

Miao, L., Xiao-Dong, L., Shuang, W., & Yi, W. 2011, Communications in Theoretical Physics, 56, 525

2011
[52]

2019, arXiv:1908.04682

Mishra, A., Reddy, P., & Nigam, R. 2019, arXiv:1908.04682

2019 arXiv
[53]

2018, JCAP, 1810, 028

Modi, C., Feng, Y., & Seljak, U. 2018, JCAP, 1810, 028

2018
[54]

2018, arXiv:1810.06441

Moss, A. 2018, arXiv:1810.06441

2018 arXiv
[55]

2019, arXiv:1903.02557

Muthukrishna, D., Parkinson, D., & Tucker, B. 2019, arXiv:1903.02557

2019 arXiv
[56]

Mnchmeyer, M., & Smith, K. M. 2019, arXiv:1905.05846

2019 arXiv
[57]

2019, arXiv:1902.10159

Ntampaka, M., et al. 2019, arXiv:1902.10159

2019 arXiv
[58]

Peebles, P. J. E., & Ratra, B. 2003, Reviews of modern physics, 75, 559

2003
[59]

2019, Phys

Peel, A., Lalande, F., Starck, J.-L., et al. 2019, Phys. Rev., D100, 023508

2019
[60]

J., Cole, S., Eisenstein, D

Percival, W. J., Cole, S., Eisenstein, D. J., et al. 2007, Mon. Not. Roy. Astron. Soc., 381, 1053

2007
[61]

1999, The Astrophysical Journal, 517, 565

Perlmutter, S., Aldering, G., Goldhaber, G., et al. 1999, The Astrophysical Journal, 517, 565

1999
[62]

2019, Astron

Perraudin, N., Deﬀerrard, M., Kacprzak, T., & Sgier, R. 2019, Astron. Comput., 27, 130 Cosmology from LSS deep learning 17 Pfeﬀer, D. N., Breysse, P. C., & Stein, G. 2019, arXiv:1905.10376

2019 arXiv
[63]

2017, arXiv:1711.02033

Ravanbakhsh, S., Oliva, J., Fromenteau, S., et al. 2017, arXiv:1711.02033

2017 arXiv
[64]

G., Filippenko, A

Riess, A. G., Filippenko, A. V., Challis, P., et al. 1998, The Astronomical Journal, 116, 1009

1998
[65]

G., Macri, L

Riess, A. G., Macri, L. M., Hoﬀmann, S. L., et al. 2016, ApJ, 826, 56

2016
[66]

C., Kacprzak, T., Lucchi, A., et al

Rodriguez, A. C., Kacprzak, T., Lucchi, A., et al. 2018, Comput. Astrophys. Cosmol., 5, 4

2018
[67]

J., Samushia, L., Howlett, C., et al

Ross, A. J., Samushia, L., Howlett, C., et al. 2015, Monthly Notices of the Royal Astronomical Society, 449, 835

2015
[68]

1995, arXiv preprint astro-ph/9506028

Ryden, B. 1995, arXiv preprint astro-ph/9506028

1995 arXiv
[69]

G., Hoyle, B., Kim, J., & Li, X.-D

Sabiu, C. G., Hoyle, B., Kim, J., & Li, X.-D. 2019, ApJS, 242, 29

2019
[70]

G., Mota, D

Sabiu, C. G., Mota, D. F., Llinares, C., & Park, C. 2016, A&A, 592, A38

2016
[71]

A., White, M., et al

Samushia, L., Reid, B. A., White, M., et al. 2014, Monthly Notices of the Royal Astronomical Society, 439, 3504 S´ anchez, A. G., Sc´ occola, C., Ross, A., et al. 2012, Monthly Notices of the Royal Astronomical Society, 425, 415 S´ anchez, A. G., Kazin, E. A., Beutler, F., et ...

2014
[72]

2017, arXiv:1707.05167

Schmelzle, J., Lucchi, A., Kacprzak, T., et al. 2017, arXiv:1707.05167

2017 arXiv
[73]

Seo, H.-J., & Eisenstein, D. J. 2003, ApJ, 598, 720

2003
[74]

J., Brownstein, J

Slepian, Z., Eisenstein, D. J., Brownstein, J. R., et al. 2017, MNRAS, 469, 1738

2017
[75]

P., & Faraoni, V

Sotiriou, T. P., & Faraoni, V. 2010, Reviews of Modern Physics, 82, 451

2010
[76]

M., Ofek, E

Springer, O. M., Ofek, E. O., Weiss, Y., & Merten, J. 2018, arXiv:1808.07491

2018 arXiv
[77]

2013, JCAP, 1306, 036

Tassev, S., Zaldarriaga, M., & Eisenstein, D. 2013, JCAP, 1306, 036

2013
[78]

R., Strauss, M

Tegmark, M., Blanton, M. R., Strauss, M. A., et al. 2004, ApJ, 606, 702

2004
[79]

2019, Astron

Tewes, M., Kuntzer, T., Nakajima, R., et al. 2019, Astron. Astrophys., 621, A36

2019
[80]

Trster, T., Ferguson, C., Harnois-Draps, J., & McCarthy, I. G. 2019, Mon. Not. Roy. Astron. Soc., 487, L24

2019
[81]

2011, Astrophysics and Space Science Library, Vol

Tsujikawa, S. 2011, Astrophysics and Space Science Library, Vol. 370, Dark Energy: Investigation and Modeling, ed. S. Matarrese, M. Colpi, V. Gorini, & U. Moschella, 331

2011
[82]

2017, ApJ, 849, 84

Wang, Y., Xu, L., & Zhao, G.-B. 2017, ApJ, 849, 84

2017
[83]

H., Mortonson, M

Weinberg, D. H., Mortonson, M. J., Eisenstein, D. J., et al. 2013, Physics reports, 530, 87

2013
[84]

1989, Reviews of modern physics, 61, 1

Weinberg, S. 1989, Reviews of modern physics, 61, 1

1989
[85]

White, S. D. M., & Rees, M. J. 1978, MNRAS, 183, 341

1978
[86]

G., Adelman, J., Anderson Jr, J

York, D. G., Adelman, J., Anderson Jr, J. E., et al. 2000, The Astronomical Journal, 120, 1579

2000
[87]

2019, Science China Physics, Mechanics, and Astronomy, 62, 110431

Zhang, X. 2019, Science China Physics, Mechanics, and Astronomy, 62, 110431

2019
[88]

2019a, arXiv:1902.05965

Zhang, X., Wang, Y., Zhang, W., et al. 2019a, arXiv:1902.05965

1902 arXiv
[89]

2018, Science China Physics, Mechanics, and Astronomy, 61, 39811

Zhao, Z., & Wang, S. 2018, Science China Physics, Mechanics, and Astronomy, 61, 39811

2018
[90]

2018, Science China Physics, Mechanics, and Astronomy, 61, 79531

Zheng, H., Wei, L.-F., Wen, H., & Li, F.-Y. 2018, Science China Physics, Mechanics, and Astronomy, 61, 79531

2018

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

P., Abbott, R., Abbott, T

Abbott, B. P., Abbott, R., Abbott, T. D., et al. 2017, Nature, 551, 85

2017

[2] [2]

A., Aghanim, N., Arnaud, M., et al

Ade, P. A., Aghanim, N., Arnaud, M., et al. 2016, Astronomy & Astrophysics, 594, A13

2016

[3] [3]

2017, Monthly Notices of the Royal Astronomical Society, 470, 2617

Alam, S., Ata, M., Bailey, S., et al. 2017, Monthly Notices of the Royal Astronomical Society, 470, 2617

2017

[4] [4]

1979, Nature, 281, 358

Alcock, C., & Paczy´ nski, B. 1979, Nature, 281, 358

1979

[5] [5]

2012, Monthly Notices of the Royal Astronomical Society, 427, 3435

Anderson, L., Aubourg, E., Bailey, S., et al. 2012, Monthly Notices of the Royal Astronomical Society, 427, 3435

2012

[6] [6]

2014, Monthly Notices of the Royal Astronomical Society, 441, 24

Anderson, L., Aubourg, ´E., Bailey, S., et al. 2014, Monthly Notices of the Royal Astronomical Society, 441, 24

2014

[7] [7]

1996, Monthly Notices of the Royal Astronomical Society, 282, 877

Ballinger, W., Peacock, J., & Heavens, A. 1996, Monthly Notices of the Royal Astronomical Society, 282, 877

1996

[8] [8]

M., Bond, J

Bardeen, J. M., Bond, J. R., Kaiser, N., & Szalay, A. S. 1986, ApJ, 304, 15

1986