Pith. sign in

REVIEW 3 major objections 5 minor 61 references

Learning Intrinsic Alignments from Local Galaxy Environments

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A deep learning model recovers galaxy intrinsic alignments from noisy, lensing-contaminated shapes by conditioning only on local galaxy positions, and reconstructs the underlying pure tidal alignment field in mock catalogs.

desk verdict DELTA is a genuinely new ML architecture that convincingly recovers injected intrinsic alignments from noisy mock data, but the paper's headline claim of separating IAs from lensing is not actually tested because no lensing signal is injected. read the letter →

arxiv 2506.05155 v2 pith:NJX2ZPDF submitted 2025-06-05 astro-ph.CO

Matthew Craigie , Eric Huff , Yuan-Sen Ting , Rossana Ruggeri , Tamara M. Davis This is my paper
classification astro-ph.CO
keywords intrinsicalignmentsweaklensingequivariantgraphneuralnetworkstidalalignmentdeeplearningcosmologygalaxyshapesvonMisesdistributionobservational
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

This paper introduces DELTA, a deep learning model that aims to recover galaxy intrinsic alignments (the tendency of nearby galaxies to align due to tidal forces) directly from observed noisy shapes and three-dimensional galaxy positions, without assuming a parametric form or using simulations. The central claim is that because lensing distortions and shape noise average to zero when conditioned on the local galaxy neighborhood, the expected orientation predicted from local structure equals the tidal alignment component alone. If correct, this would let upcoming weak lensing surveys separate a pernicious systematic from the cosmological signal nonparametrically, and map tidal alignment patterns across the sky. The paper demonstrates the claim on mock catalogs with injected alignments of varying strength, reporting 8-11% improvement over random for central galaxies and 16-70% for satellites in recovering the pure signal.

What carries the argument

The load-bearing object is the conditional expectation identity $E[\epsilon|N] = E[\epsilon_T|N]$, which converts the unobservable tidal component into a quantity learnable from data. The model that realizes it is an Equivariant Graph Neural Network (EGNN) whose message passing respects global rotation and translation invariance and spin-2 symmetry in the $xy$-plane, followed by a probabilistic head that outputs the mean and concentration parameters of a von Mises distribution over orientation angles. The EGNN aggregates information from each galaxy's $k$ nearest neighbors ($k=10$) in 3D, and projects to 2D spin-2 displacements at the output, so the predicted mean angle is equivariant to the same rotations as galaxy shapes. The von Mises parameterization lets the model represent the large, noise-dominated spread while its mean carries the alignment signal.

What would settle it

Run the same DELTA training pipeline on a mock that adds a realistic line-of-sight lensing shear field and compare the recovered alignment maps to the known pure tidal field; if the predicted mean orientations shift toward the lensing pattern, the independence assumption fails. Complementarily, on real data one could split galaxies by redshift uncertainty and check whether predicted alignments vary with photometric redshift scatter, which would indicate lensing information leaking through the neighborhood.

Watch

Extended reading notes

Core claim

DELTA's discovery is that intrinsic alignments can be isolated from weak lensing contamination by a conditioning argument: writing observed ellipticity as $\epsilon = \epsilon_T + \epsilon_L + \epsilon_N$, the paper argues that $E[\epsilon|N] = E[\epsilon_T|N]$, because lensing and noise have zero conditional mean given the local neighborhood $N$. A graph neural network restricted to local positions and safe, lensing-invariant galaxy properties then learns the mapping from neighborhood to expected orientation, with orientation modeled as a von Mises distribution. When trained on mock galaxies whose observed orientations are dominated by noise (the tidal signal is roughly 1% of the ellipticity magnitude), the model's predicted mean orientations track the injected noise-free tidal alignment field, and the discrepancy is measurable as an improvement over random guessing. The paper also shows that interpretation tools recover known physical structure, such as satellites radially aligned toward halo centers and a density dependence of alignment.

Load-bearing premise

The load-bearing premise is that gravitational lensing and shape noise are completely uncorrelated with the local galaxy neighborhood, so their conditional mean vanishes and the model cannot learn them; the mocks never inject a lensing signal, so this premise is asserted rather than demonstrated.

Editorial extensions

If this is right

  • DELTA can replace parametric intrinsic alignment models in weak lensing analyses, removing reliance on TATT-like functional forms and their uncertain redshift scaling.
  • Joint photometric shape surveys and spectroscopic redshift surveys, such as Euclid with DESI, would supply the roughly 20 million position-shape pairs needed to train the model on real data.
  • Because the model outputs a per-galaxy predicted alignment and confidence, alignment maps and uncertainty estimates become directly visualizable without two-point statistics.
  • The permutation-test and latent-space interpretability analyses show that learned alignments can be checked against physical expectations, offering a validation route when no ground truth exists.
  • Nonzero recovery in the medium and high alignment scenarios suggests DELTA is sensitive to signals comparable to those of intermediate- and high-mass galaxies in hydrodynamical simulations.

Reading between the lines

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

  • A strict test of the paper's core assumption would inject a realistic lensing shear field into the mock and check whether DELTA's recovered alignments remain unbiased; the current mocks deliberately omit lensing, so the claim that the model ignores lensing is not yet directly demonstrated.
  • The locality argument suggests DELTA's recovered field could be used as a nonparametric estimator of the tidal field itself, potentially connecting to constrained realizations or tidal-tensor reconstruction methods.
  • If the conditioning identity holds with spectroscopic redshifts, the same architecture could be extended to predict higher-order shape information (for example, the full ellipticity distribution) rather than just the mean orientation, enabling probabilistic shear-IA separation.
  • Redshift errors perturb the neighborhood $N$, so one testable prediction is that DELTA's performance degrades smoothly with line-of-sight position error; this could be measured in mocks with increasing redshift scatter.
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 / 5 minor

Summary. The manuscript introduces DELTA, a deep-learning framework that combines an equivariant graph neural network with a probabilistic von Mises output to predict galaxy orientation from local environmental information. The authors argue that because lensing and shape noise have zero conditional mean given the local neighborhood, a model trained on observed noisy orientations and local positions learns the pure tidal alignment signal. They validate the method on mock catalogs with injected intrinsic alignments and demonstrate improvement over a random baseline in recovering noise-free alignments. They also apply permutation and latent-space interpretability tools to show that the model uses physically meaningful properties. The paper claims the method isolates IAs from lensing distortions using only observational data and is suitable for upcoming surveys such as Euclid, Rubin, and DESI.

Significance. The proposed approach is timely and potentially valuable: if it works, DELTA would provide a nonparametric, data-driven alternative to parametric IA models like TATT, with the ability to capture nonlinear and environmental dependencies. The code is publicly available, and the mock experiments are clearly described and reproducible. The interpretability analyses are a useful addition and demonstrate consistency with the injected IA model. However, the paper's central claim—that DELTA isolates IAs from lensing distortions—is not directly tested because no lensing signal is injected into the mocks. The theoretical justification for the angle-expectation equality also contains a gap. These issues are load-bearing for the advertised contribution and need to be addressed before the claim can be accepted.

major comments (3)
  1. [II C 1] The mock experiments do not include a lensing signal; the paper states 'We do not explicitly inject a lensing signal' and treats lensing as 'randomly oriented and indistinguishable from the existing isotropic shape noise.' This is not a test of the claimed ability to isolate IAs from lensing distortions, because a real lensing field is spatially coherent and correlated with the large-scale structure that defines each galaxy's neighborhood. The authors should inject a realistic lensing shear field (including source-lens clustering and photometric redshift errors) and show that DELTA's predicted orientation field is unbiased, or quantify the bias introduced. Without such a test, the abstract's claim that DELTA 'isolates galaxy intrinsic alignments from weak lensing distortions' is not empirically supported.
  2. [II A, Eq. (7)] The step from E[epsilon|N] = E[epsilon_T|N] to E[phi|N] = E[phi_T|N] is asserted without proof. The orientation angle phi is a nonlinear function of the complex ellipticity, and zero conditional mean of the lensing and noise components does not imply that the conditional mean of the angle is unaffected by them. For example, if the noise is isotropic and dominant, the circular mean of the observed angle is not generally equal to the tidal alignment angle except in a small-signal or high-concentration limit. The authors should either derive Eq. (7) under explicit assumptions or reframe the model as learning the conditional mean of the observed noisy angle and discuss under what conditions this equals the tidal direction.
  3. [II D 2] The assumption P(epsilon_L|N)=P(epsilon_L) is acknowledged to be violated by redshift uncertainties, lensing deflection, and coherence of large-scale structure. The paper asserts that these effects are 'subdominant' without providing quantitative estimates. Since this assumption is central to the method's advertised capability of removing lensing contamination, and since the mocks do not constrain it, the authors should provide a quantitative assessment (e.g., using realistic mock lensing fields or analytical estimates of the conditional mean shear given local density) and specify the survey conditions under which the assumption is valid.
minor comments (5)
  1. [II B, Eq. (12)] Equation (12) appears to contain a typo: it reads kappa_i = f_mu(kappa_i), which should likely be kappa_i = f_kappa(alpha_i).
  2. [II C 4, Figure 3] The y-axis of Figure 3 spans only from 0.3658 to 0.3666, which visually exaggerates the inverse trend between confidence and prediction error; the absolute variation is extremely small, and the authors should report the statistical significance of this trend.
  3. [III A] The permutation test conclusions are based on visual inspection ('visually apparent by eye') rather than a formal statistical test; a quantitative significance measure would strengthen the interpretability claim.
  4. [Abstract and Introduction] The phrase 'without relying on simulations' is imprecise because the method is validated on mock catalogs; the authors should clarify that DELTA does not require simulations for training, while simulations are used for validation.
  5. [Table I] In the 'Full' alignment scenario, the noisy improvement IN is listed as '–'; a brief explanation of why this metric is undefined (because there is no noise) would improve readability.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the pure IA ground truth is externally injected and never used in training, and Equation (5) is an analytic conditional-mean identity rather than a fitted relation.

full rationale

The central recovery claim is not circular. The pure IA ground truth (phi_hat) is produced by an external mock-injection pipeline (Van Alfen et al. 2024) from halo shapes and halo-centric radii; the model is trained only on noisy observed orientations phi and local positions, and never on phi_hat. Equation (5) is an analytic identity E[epsilon|N]=E[epsilon_T|N] following from the stated independence assumptions P(epsilon_L|N)=P(epsilon_L) and P(epsilon_N|N)=P(epsilon_N) with zero means; it is not fitted and is not equivalent to the network output. The validation metric I_P compares the network's conditional mean to an independent noise-free label, so a successful recovery is an empirical consistency check rather than a construction. The mock does not inject lensing, so the lensing-isolation part of the headline is less directly validated, but that is a limitation of the test, not a circular definition. The only self-citations (the DELTA GitHub repository and Sheldon & Huff 2017 METACALIBRATION) are not load-bearing for the derivation. No step in the paper reduces by construction to its own input.

Assumptions & free parameters 6 free parameters · 7 assumptions · 0 invented entities

The central claim rests on the domain assumption that lensing and noise average to zero conditioned on the local neighborhood, on the locality of tidal alignments, and on the von Mises ansatz for orientation output. The mock injection pipeline from Van Alfen et al. provides the external ground truth. No new physical entities are introduced. Model hyperparameters such as k=10 nearest neighbors, 3 hops, 16 hidden units, training epochs, and mask fraction are hand-chosen and influence the reported performance.

free parameters (6)
  • number of nearest neighbors k = 10
    Controls the neighborhood scope in the graph neural network; chosen as a simple distance limit approximately uniform across the dataset (Section II B).
  • latent and hidden dimensionality = 16
    All fully connected layers and latent vectors are fixed to 16 nodes due to GPU memory constraints (Section II C 2), limiting model capacity and affecting recovery performance.
  • number of message passing hops = 3
    EGNN aggregation depth; the central claim depends on this architectural choice.
  • training epochs = 2000 pretrain + 2000 full
    Convergence schedule chosen by the authors, not derived from principles (Section II C 2).
  • random masking fraction = 0.5
    Half of galaxy contributions are masked during loss aggregation to reintroduce stochasticity in full-batch training (Section II C 2).
  • local density radius = 5 Mpc
    Used for the permutation test and latent space coloring; arbitrary but affects interpretability results (Section III A).
assumptions (7)
  • domain assumption Observed ellipticity decomposes additively into independent tidal, lensing, and noise components.
    Equation 1 in Section II A. Standard in weak lensing, but independence is assumed without direct validation.
  • domain assumption Lensing and noise are statistically isotropic and independent of the local neighborhood, so E[epsilon_L|N] = 0 and E[epsilon_N|N] = 0.
    Equations 2 to 4 and Section II D. This is the load-bearing premise that lets DELTA attribute learned orientation to the tidal component.
  • domain assumption Tidal alignment is local and vanishes beyond roughly 10 Mpc, so the local galaxy distribution encodes the tidal component.
    Section II A, citing Zjupa et al. and Codis et al. Underpins the graph neighborhood construction.
  • domain assumption Galaxy positions and the chosen properties trace the underlying matter and tidal field sufficiently for the model to learn the tidal component, including implicit galaxy bias.
    Section II D 1 states DELTA can learn an internal mapping from galaxy positions to tidal ellipticity contribution. No explicit bias model is provided.
  • ad hoc to paper The conditional orientation distribution is approximately von Mises, with the mean parameter representing the expected tidal orientation.
    Equation 9 in Section II B. A single well-parameterized distribution is chosen for tractability and is not derived from the physical process.
  • domain assumption The mock catalogs reproduce realistic intrinsic alignment signal and noise through the injection pipeline of Van Alfen et al.
    Section II C 1. Validation quality depends on mock realism, and no lensing signal is included in the mocks.
  • ad hoc to paper The angular expectation equality E[phi|N] = E[phi_T|N] follows from the complex ellipticity decomposition.
    Equation 7 in Section II A. The orientation angle of a sum of ellipticity components is not the sum of orientation angles, and the paper provides no proof of this step.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Learning Intrinsic Alignments from Local Galaxy Environments." pith.science (2026). https://pith.science/paper/NJX2ZPDF

@misc{pith2026250605155,
  author       = {Pith},
  title        = {Pith review of: Learning Intrinsic Alignments from Local Galaxy Environments},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NJX2ZPDF}},
  note         = {Machine review of arXiv:2506.05155}
}
read the original abstract

We present DELTA (Data-Empiric Learned Tidal Alignments), a deep learning model that isolates galaxy intrinsic alignments (IAs) from weak lensing distortions using only observational data. The model uses an Equivariant Graph Neural Network backbone suitable for capturing information from the local galaxy environment, in conjunction with a probabilistic orientation output. Unlike parametric models, DELTA flexibly learns the relationship between galaxy shapes and their local environments, without assuming an explicit IA form or relying on simulations. When applied to mock catalogs with realistic noisy IAs injected, it accurately reconstructs the noise-free, pure IA signal. Mapping these alignments provides a direct visualization of IA patterns in the mock catalogs. Combining DELTA with deep learning interpretation techniques provides further insights into the physics driving tidal relationships between galaxies. This new approach to understanding and controlling IAs is suitable for application to joint photometric and spectroscopic surveys such as the combination of upcoming Euclid, Rubin, and DESI datasets.

Figures

Figures reproduced from arXiv: 2506.05155 by the authors.

Figure 1
Figure 1. FIG. 1. An overview schematic of the Data-Empiric Learned Tidal Alignments (DELTA) model. DELTA uses an equivariant [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Mapped intrinsic alignments for a region of the validation dataset, demonstrating DELTA’s ability to recover intrinsic [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Mean prediction confidence, binned by prediction er [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: FIG. 4. A permutation test for several galaxy properties, [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 6
Figure 6. Figure 6: FIG. 6. An example galaxy cluster colored by the second [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

61 extracted references · 42 canonical work pages

  1. [1]

    Leveraging Deep Learning Directly on Data DELTA combines the flexibility of deep learning with a fully data-driven approach, avoiding domain shift errors associated with models that train on simulated large- scale structure and apply to observational data. Deep learning’s high representational capacity can lead to over- fitting to features that are unique...

    work page

  2. [2]

    In our main analysis, we do not include any galaxy observables, and instead set ⃗ pi = (1,) for all galaxies

    Model Setup and Training The training dataset consists of galaxies indexed byi, each described by node positions and properties (⃗ ri, ⃗ pi), along with a target orientationϕ i. In our main analysis, we do not include any galaxy observables, and instead set ⃗ pi = (1,) for all galaxies. This ensures that the model extracts alignment information solely fro...

    work page 2000

  3. [3]

    Recovering their mean isolates the pure, noise-free, tidally induced ori- entation signal

    Model Evaluation Metrics DELTA’s training procedure differs from standard ap- proaches in that the training objective is to match the distribution of noisy orientations in the data, while the scientific goal is to recover their mean. Recovering their mean isolates the pure, noise-free, tidally induced ori- entation signal. In observational data, no ground...

    work page

  4. [4]

    Table I reports the improvement percentages over a random baseline for both the pure and noisy alignment metrics

    Results DELTA successfully recovers tidal alignments from noisy galaxy orientation data in mock datasets for all alignment scenarios. Table I reports the improvement percentages over a random baseline for both the pure and noisy alignment metrics. In all three alignment sce- narios, the model successfully captures a portion of the underlying satellite and...

    work page

  5. [5]

    Under the assumption of a statistically isotropic universe, the lens- ing contribution averages to zero, and the model cannot predict any lensing-induced ellipticity

    Avoiding Contamination from Lensing To isolate the intrinsic alignment contribution, we as- sume that the lensing signal is independent of the local neighborhood, such thatP(ϵ L|N) =P(ϵ L). Under the assumption of a statistically isotropic universe, the lens- ing contribution averages to zero, and the model cannot predict any lensing-induced ellipticity. ...

    work page

  6. [6]

    Lee and U.-L

    J. Lee and U.-L. Pen, Cosmic shear from galaxy spins, The Astrophysical Journal532, L5–L8 (2000)

    work page 2000

  7. [7]

    Mellier, Abdurro’uf, J

    Euclid Collaboration, Y. Mellier, Abdurro’uf, J. A. Acevedo Barroso, A. Ach´ ucarro, J. Adamek, R. Adam, G. E. Addison, N. Aghanim, M. Aguena, V. Ajani, Y. Akrami, A. Al-Bahlawan, A. Alavi, I. S. Albuquerque, G. Alestas, G. Alguero, A. Allaoui, S. W. Allen, V. Alle- vato, A. V. Alonso-Tetilla, B. Altieri, A. Alvarez-Candal, S. Alvi, A. Amara, L. Amendola,...

    arXiv 2024

  8. [8]

    Spergel, N

    D. Spergel, N. Gehrels, C. Baltay, D. Bennett, J. Breckin- ridge, M. Donahue, A. Dressler, B. S. Gaudi, T. Greene, O. Guyon, C. Hirata, J. Kalirai, N. J. Kasdin, B. Macintosh, W. Moos, S. Perlmutter, M. Post- man, B. Rauscher, J. Rhodes, Y. Wang, D. Wein- berg, D. Benford, M. Hudson, W. S. Jeong, Y. Mellier, W. Traub, T. Yamada, P. Capak, J. Colbert, D. M...

    arXiv 2015

Show all 61 references
  1. [9]

    Ivezi´ c, S

    ˇZ. Ivezi´ c, S. M. Kahn, J. A. Tyson, B. Abel, E. Acosta, R. Allsman, D. Alonso, Y. AlSayyad, S. F. Anderson, J. Andrew, J. R. P. Angel, G. Z. Angeli, R. Ansari, P. Antilogus, C. Araujo, R. Armstrong, K. T. Arndt, P. Astier, ´E. Aubourg, N. Auza, T. S. Axelrod, D. J. Bard, J....

    2019 arXiv

  2. [10]

    R. A. C. Croft and C. A. Metzler, Weak-Lensing Sur- veys and the Intrinsic Correlation of Galaxy Ellipticities, Astrophys. J.545, 561 (2000), arXiv:astro-ph/0005384 [astro-ph]

    2000 arXiv

  3. [11]

    Heavens, A

    A. Heavens, A. Refregier, and C. Heymans, Intrinsic cor- relation of galaxy shapes: implications for weak lensing measurements, Monthly Notices of the Royal Astronom- ical Society319, 649–656 (2000)

    2000

  4. [12]

    Krause, T

    E. Krause, T. Eifler, and J. Blazek, The impact of intrin- sic alignment on current and future cosmic shear surveys, Monthly Notices of the Royal Astronomical Society456, 207–222 (2015)

    2015

  5. [13]

    Lee and U

    J. Lee and U. Pen, Galaxy spin statistics and spin-density correlation, The Astrophysical Journal555, 106–124 (2001)

    2001

  6. [14]

    Catelan, M

    P. Catelan, M. Kamionkowski, and R. D. Blandford, In- trinsic and extrinsic galaxy alignment, Monthly Notices of the Royal Astronomical Society320, L7–L13 (2001)

    2001

  7. [15]

    Joachimi and S

    B. Joachimi and S. L. Bridle, Simultaneous measurement of cosmology and intrinsic alignments using joint cosmic shear and galaxy number density correlations, Astron- omy & Astrophysics523, A1 (2010)

    2010

  8. [16]

    Joachimi, R

    B. Joachimi, R. Mandelbaum, F. B. Abdalla, and S. L. Bridle, Constraints on intrinsic alignment contamination of weak lensing surveys using the megaz-lrg sample, As- tronomy & Astrophysics527, A26 (2011)

    2011

  9. [17]

    D. Kirk, A. Rassat, O. Host, and S. Bridle, The cos- mological impact of intrinsic alignment model choice for cosmic shear: Impact of intrinsic alignment model choice, Monthly Notices of the Royal Astronomical Society424, 1647–1657 (2012)

    2012

  10. [18]

    Johnston, C

    H. Johnston, C. Georgiou, B. Joachimi, H. Hoekstra, N. E. Chisari, D. Farrow, M. C. Fortuna, C. Heymans, S. Joudaki, K. Kuijken, and A. Wright, Kids+gama: In- trinsic alignment model constraints for current and future weak lensing cosmology, Astronomy & Astrophysics624, A30 (2019)

    2019

  11. [19]

    J. A. Blazek, N. MacCrann, M. Troxel, and X. Fang, Be- yond linear galaxy alignments, Physical Review D100, 10.1103/physrevd.100.103506 (2019)

    2019 doi

  12. [20]

    Secco, S

    L. Secco, S. Samuroff, E. Krause, B. Jain, J. Blazek, M. Raveri, A. Campos, A. Amon, A. Chen, C. Doux, A. Choi, D. Gruen, G. Bernstein, C. Chang, J. DeRose, J. Myles, A. Fert´ e, P. Lemos, D. Huterer, J. Prat, M. Troxel, N. MacCrann, A. Liddle, T. Kacprzak, X. Fang, C. S´ anch...

    2022

  13. [21]

    Campos, S

    A. Campos, S. Samuroff, and R. Mandelbaum, An empirical approach to model selection: weak lensing and intrinsic alignments, Monthly No- tices of the Royal Astronomical Society525, 1885 (2023), https://academic.oup.com/mnras/article- pdf/525/2/1885/51236893/stad2213.pdf

    2023

  14. [22]

    C. M. Hirata, R. Mandelbaum, M. Ishak, U. Seljak, R. Nichol, K. A. Pimbblet, N. P. Ross, and D. Wake, In- trinsic galaxy alignments from the 2slaq and sdss surveys: luminosity and redshift scalings and implications for weak lensing surveys: Intrinsic alignments, Monthly No- ti...

    2007

  15. [23]

    Singh, R

    S. Singh, R. Mandelbaum, and S. More, Intrinsic align- ments of sdss-iii boss lowz sample galaxies, Monthly No- tices of the Royal Astronomical Society450, 2195–2216 (2015)

    2015

  16. [24]

    Zhang, X

    Y. Zhang, X. Yang, H. Wang, L. Wang, W. Luo, H. J. Mo, and F. C. van den Bosch, Spin alignments of spiral galaxies within the large-scale structure from sdss dr7, The Astrophysical Journal798, 17 (2014)

    2014

  17. [25]

    Mandelbaum, A

    R. Mandelbaum, A. Slosar, T. Baldauf, U. Seljak, C. M. Hirata, R. Nakajima, R. Reyes, and R. E. Smith, Cosmological parameter constraints from galaxy–galaxy lensing and galaxy clustering with the sdss dr7, Monthly Notices of the Royal Astronomical Society432, 1544 (2013), http...

    2013

  18. [26]

    Blazek, R

    J. Blazek, R. Mandelbaum, U. Seljak, and R. Nakajima, Separating intrinsic alignment and galaxy-galaxy lensing, Journal of Cosmology and Astroparticle Physics2012 (05), 041–041

  19. [27]

    Mandelbaum, C

    R. Mandelbaum, C. M. Hirata, M. Ishak, U. Sel- jak, and J. Brinkmann, Detection of large-scale intrin- sic ellipticity–density correlation from the sloan digital sky survey and implications for weak lensing surveys, 19 Monthly Notices of the Royal Astronomical Society367, 611–...

    2006

  20. [28]

    Blazek, M

    J. Blazek, M. McQuinn, and U. Seljak, Testing the tidal alignment model of galaxy intrinsic alignment, Jour- nal of Cosmology and Astroparticle Physics2011(05), 010–010

  21. [29]

    Tsaprazi, N.-M

    E. Tsaprazi, N.-M. Nguyen, J. Jasche, F. Schmidt, and G. Lavaux, Field-level inference of galaxy intrinsic align- ment from the sdss-iii boss survey, Journal of Cosmology and Astroparticle Physics2022(08), 003

  22. [30]

    Georgiou, H

    C. Georgiou, H. Johnston, H. Hoekstra, M. Viola, K. Kui- jken, B. Joachimi, N. E. Chisari, D. J. Farrow, H. Hilde- brandt, B. W. Holwerda, and A. Kannawadi, The depen- dence of intrinsic alignment of galaxies on wavelength using kids and gama, Astronomy & Astrophysics622, A90 (2019)

    2019

  23. [31]

    Pahwa, N

    I. Pahwa, N. I. Libeskind, E. Tempel, Y. Hoffman, R. B. Tully, H. M. Courtois, S. Gottl¨ ober, M. Steinmetz, and J. G. Sorce, The alignment of galaxy spin with the shear field in observations, Monthly Notices of the Royal As- tronomical Society457, 695–703 (2016)

    2016

  24. [32]

    Abbott, M

    T. Abbott, M. Aguena, A. Alarcon, S. Allam, O. Alves, A. Amon, F. Andrade-Oliveira, J. Annis, S. Avila, D. Bacon, E. Baxter, K. Bechtol, M. Becker, G. Bern- stein, S. Bhargava, S. Birrer, J. Blazek, A. Brandao- Souza, S. Bridle, D. Brooks, E. Buckley-Geer, D. Burke, H. Camacho...

    2022

  25. [33]

    Samuroff, R

    S. Samuroff, R. Mandelbaum, and J. Blazek, Ad- vances in constraining intrinsic alignment mod- els with hydrodynamic simulations, Monthly No- tices of the Royal Astronomical Society508, 637 (2021), https://academic.oup.com/mnras/article- pdf/508/1/637/40453160/stab2520.pdf

    2021

  26. [34]

    A. M. Delgado, B. Hadzhiyska, S. Bose, V. Springel, L. Hernquist, M. Barrera, R. Pakmor, F. Ferlito, R. Kannan, C. Hern´ andez-Aguayo, S. D. M. White, and C. Frenk, The millenniumtng project: intrinsic align- ments of galaxies and haloes, Monthly Notices of the Royal Astronomi...

    2023

  27. [35]

    Zjupa, B

    J. Zjupa, B. M. Sch¨ afer, and O. Hahn, Intrinsic align- ments in illustristng and their implications for weak lens- ing: Tidal shearing and tidal torquing mechanisms put to the test (2020), arXiv:2010.07951 [astro-ph.GA]

    2020 arXiv

  28. [36]

    Craigie, DELTA: Data-Empiric Learned Tidal Alignments,https://github.com/mattcraigie/DELTA (2025), accessed: 2025-06-07

    M. Craigie, DELTA: Data-Empiric Learned Tidal Alignments,https://github.com/mattcraigie/DELTA (2025), accessed: 2025-06-07

    2025

  29. [37]

    S. Chen, J. DeRose, R. Zhou, M. White, S. Ferraro, C. Blake, J. U. Lange, R. H. Wechsler, J. Aguilar, S. Ahlen, D. Brooks, T. Claybaugh, K. Dawson, A. de la Macorra, P. Doel, A. Font-Ribera, E. Gazta˜ naga, S. G. A. Gontcho, G. Gutierrez, K. Honscheid, C. Howlett, R. Ke- hoe, ...

    2024 arXiv

  30. [38]

    Aric` o, R

    G. Aric` o, R. E. Angulo, M. Zennaro, S. Contreras, A. Chen, and C. Hern´ andez-Monteagudo, Des y3 cos- mic shear down to small scales: Constraints on cosmol- ogy and baryons, Astronomy & Astrophysics678, A109 (2023)

    2023

  31. [39]

    Gordon, B

    J. Gordon, B. F. de Aguiar, J. Rebou¸ cas, G. Brando, F. Falciano, V. Miranda, K. Koyama, and H. A. Winther, Modeling nonlinear scales with cola: preparing for lsst-y1 (2024), arXiv:2404.12344 [astro-ph.CO]

    2024 arXiv

  32. [40]

    Zjupa, B

    J. Zjupa, B. M. Sch¨ afer, and O. Hahn, Intrin- sic alignments in illustristng and their implica- tions for weak lensing: Tidal shearing and tidal torquing mechanisms put to the test, Monthly Notices of the Royal Astronomical Society514, 2049 (2022), https://academic.oup.com/m...

    2022

  33. [41]

    Codis, R

    S. Codis, R. Gavazzi, Y. Dubois, C. Pichon, K. Ben- abed, V. Desjacques, D. Pogosyan, J. Devriendt, and A. Slyz, Intrinsic alignment of simulated galaxies in the cosmic web: implications for weak lensing surveys, Monthly Notices of the Royal Astronomical Society448, 3391 (2015...

    2015

  34. [42]

    V. G. Satorras, E. Hoogeboom, and M. Welling, E(n) equivariant graph neural networks (2022), arXiv:2102.09844 [cs.LG]

    2022 arXiv

  35. [43]

    A. J. Larson, J. F. Wu, and C. Jones, Predicting dark matter halo masses from simulated galaxy images and environments (2024), arXiv:2407.13735 [astro-ph.CO]

    2024 arXiv

  36. [44]

    N. S. M. de Santi, H. Shao, F. Villaescusa-Navarro, L. R. Abramo, R. Teyssier, P. Villanueva-Domingo, Y. Ni, D. Angl´ es-Alc´ azar, S. Genel, E. Hern´ andez-Mart ´ ınez, 20 U. P. Steinwandel, C. C. Lovell, K. Dolag, T. Castro, and M. Vogelsberger, Robust field-level likelihood...

    2023

  37. [45]

    Lehman, S

    K. Lehman, S. Krippendorf, J. Weller, and K. Dolag, Learning optimal and interpretable summary statistics of galaxy catalogs with sbi (2024), arXiv:2411.08957 [astro- ph.CO]

    2024 arXiv

  38. [46]

    Roncoli, A

    A. Roncoli, A. ´Ciprijanovi´ c, M. Voetberg, F. Villaescusa- Navarro, and B. Nord, Domain adaptive graph neu- ral networks for constraining cosmological parameters across multiple data sets (2024), arXiv:2311.01588 [astro- ph.CO]

    2024 arXiv

  39. [47]

    N. S. de Santi, F. Villaescusa-Navarro, L. Raul Abramo, H. Shao, L. A. Perez, T. Castro, Y. Ni, C. C. Lovell, E. Hern´ andez-Mart ´ ınez, F. Marinacci, D. N. Spergel, K. Dolag, L. Hernquist, and M. Vogelsberger, Field-level simulation-based inference with galaxy catalogs: the ...

  40. [48]

    Gatti, G

    M. Gatti, G. Campailla, N. Jeffrey, L. Whiteway, A. Porredon, J. Prat, J. Williamson, M. Raveri, B. Jain, V. Ajani, G. Giannini, M. Yamamoto, C. Zhou, J. Blazek, D. Anbajagane, S. Samuroff, T. Kacprzak, A. Alar- con, A. Amon, K. Bechtol, M. Becker, G. Bernstein, A. Campos, C. ...

    2025

  41. [49]

    Bishop,Mixture density networks, Working Paper (Aston University, 1994)

    C. Bishop,Mixture density networks, Working Paper (Aston University, 1994)

    1994

  42. [50]

    Van Alfen, D

    N. Van Alfen, D. Campbell, J. Blazek, C. D. Leonard, F. Lanusse, A. Hearin, and R. Mandelbaum, An empiri- cal model for intrinsic alignments: Insights from cosmo- logical simulations, The Open Journal of Astrophysics7, 10.33232/001c.118783 (2024)

    2024 doi

  43. [51]

    Prada, A

    F. Prada, A. A. Klypin, A. J. Cuesta, J. E. Betancort-Rijo, and J. Primack, Halo concentrations in the standardλcold dark matter cosmology, Monthly Notices of the Royal Astronomical Society423, 3018 (2012), https://academic.oup.com/mnras/article- pdf/423/4/3018/4872728/mnras04...

    2012

  44. [52]

    receive” node and subscriptjindexes the “send

    with shape measurements from Euclid. These sur- veys have a∼9000 square degree overlap [53], yielding approximately 20 million galaxy shape and 3D position pairs, which is an order-of-magnitude increase over the mock catalogs used in this work. As an alternative to the high-re...

  45. [53]

    P. S. Behroozi, R. H. Wechsler, and H.-Y. Wu, The ROCKSTAR Phase-space Temporal Halo Finder and the Velocity Offsets of Cluster Cores, Astrophys. J.762, 109 (2013), arXiv:1110.4372 [astro-ph.CO]

    2013 arXiv

  46. [54]

    R´ egaldo-Saint Blancard, C

    B. R´ egaldo-Saint Blancard, C. Hahn, S. Ho, J. Hou, P. Lemos, E. Massara, C. Modi, A. M. Dizgah, L. Parker, Y. Yao, and M. Eickenberg, Galaxy clustering analysis with simbig and the wavelet scattering transform, Physi- cal Review D109, 10.1103/physrevd.109.083535 (2024)

    2024 doi

  47. [55]

    Breiman, Random forests, Machine Learning45, 5 (2001)

    L. Breiman, Random forests, Machine Learning45, 5 (2001)

    2001

  48. [56]

    Lundberg and S.-I

    S. Lundberg and S.-I. Lee, A unified approach to in- terpreting model predictions (2017), arXiv:1705.07874 [cs.AI]

    2017 arXiv

  49. [57]

    McInnes, J

    L. McInnes, J. Healy, and J. Melville, Umap: Uniform manifold approximation and projection for dimension re- duction (2020), arXiv:1802.03426 [stat.ML]

    2020 arXiv

  50. [58]

    Abareshi, J

    DESI Collaboration, B. Abareshi, J. Aguilar, S. Ahlen, S. Alam, D. M. Alexander, R. Alfarsy, L. Allen, C. Allende Prieto, O. Alves, J. Ameel, E. Armen- gaud, J. Asorey, A. Aviles, S. Bailey, A. Balaguera- Antol ´ ınez, O. Ballester, C. Baltay, A. Bault, S. F. Bel- tran, B. Ben...

    2022 arXiv

  51. [59]

    Naidoo, H

    K. Naidoo, H. Johnston, B. Joachimi, J. L. van den Busch, H. Hildebrandt, O. Ilbert, O. Lahav, N. Aghanim, B. Altieri, A. Amara, M. Baldi, R. Bender, C. Bo- dendorf, E. Branchini, M. Brescia, J. Brinchmann, S. Camera, V. Capobianco, C. Carbone, J. Carretero, F. J. Castander, M...

    2023

  52. [60]

    E. S. Sheldon and E. M. Huff, Practical weak-lensing shear measurement with metacalibration, The Astro- physical Journal841, 24 (2017)

    2017

  53. [61]

    L. Wu, P. Cui, J. Pei, and L. Zhao,Graph Neu- ral Networks: Foundations, Frontiers, and Applications (Springer Singapore, Singapore, 2022) p. 725

    2022

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.