Cosmological reconstruction from galaxy light: neural network based light-matter connection
Open accessAbstract
We present a method to reconstruct the initial conditions of the universe using observed galaxy positions and luminosities under the assumption that the luminosities can be calibrated with weak lensing to give the mean halo mass. Our method relies on following the gradients of forward model and since the standard way to identify halos is non-differentiable and results in a discrete sample of objects, we propose a framework to model the halo position and mass field starting from the non-linear matter field using Neural Networks (NN), which are differentiable, yet can produce very pointlike maps. We evaluate the performance of our model with multiple metrics and find that our model is more than 95% correlated with the halo-mass fields up to k ∼ 0.7 h/Mpc, and significantly reduces the stochasticity over the Poisson shot noise. We develop a data likelihood model that takes our modeling error and intrinsic scatter in the halo mass-light relation into account and show that a displaced log-normal model is a good approximation to it. We optimize over the corresponding loss function to reconstruct the initial density field of the dark matter starting from the halo mass field. To speed up and improve the convergence, we develop an annealing procedure for several parameters in the loss function, such as smoothing the likelihood starting with large smoothing and gradually decreasing it. We apply the method to halo number densities of = 2.5 × 10 −4 − 10 −3 (h/ Mpc ) 3 , typical of current and future redshift surveys, and recover a Gaussian initial density field, mapping all the higher order information in the data into the power spectrum of the linear reconstructed field. We show that our reconstruction improves over the standard reconstruction. For baryonic acoustic oscillations (BAO) the gains are relatively modest because BAO is dominated by large scales where standard reconstruction suffices. We improve upon it by ∼ 15–20% in terms of error on BAO peak as estimated by Fisher analysis at z =0. We expect larger gains will be achieved when applying this method to the broadband linear power spectrum reconstruction on smaller scales.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Modi et al. (2018) model provided a differentiable mapping between dark matter and halos, and this mapping could be used in a reconstruction scheme to infer initial conditions from a halo field by gradient descent.
“Perhaps most interestingly, this model provided effectively a differentiable mapping between dark matter and halos, a differentiable halo finder, and the authors demonstrated that this mapping could be used in a reconstruction scheme to infer initial conditions from a halo field by gradient descent through the neural network and a differentiable N-body simulation as illustrated on Figure 43.”
From The Dawes Review 10: The impact of deep learning for the analysis of galaxy surveys (2023) · cites this paper as Modi et al. (2018)
The predicted halo mass field from Modi et al. (2018) exhibits over 95% correlation with the true field up to k = 0.7 h Mpc^-1.
“The authors find that the predicted halo mass field exhibits over a 95% correlation with the true field up to k = 0.7 h Mpc -1 .”
From The Dawes Review 10: The impact of deep learning for the analysis of galaxy surveys (2023) · cites this paper as Modi et al. (2018)
Cites
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946 · see citing works
Emilia Pardo-Bazan and the Americas
Hilton, R. (1952) · Americas · doi:10.2307/978656 · Cited by 68,896 · cited 1×
“[15], to model discrimination [16], creating virtual universe [17] and to detecting strong lenses [18].”
BackgroundDistinctive Image Features from Scale-Invariant Keypoints
Lowe, D. G. (2004) · International Journal of Computer Vision · doi:10.1023/b:visi.0000029664.99615.94 · Cited by 55,589 · cited 1× · see citing works
“In addition, we use Gaussian difference (GD) which is an approximate blob detection technique used in computer vision [27] to identify peaks in”
MethodHe, H., & Garcia, E. A. (2009) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2008.239 · Cited by 10,359 · cited 1× · see citing works
“We also do not use the traditional ways of re-weighting and oversampling [28] to address this imbalance since most, but not all, high density regions host halos and using both these methods skews the ratio of halo to non-halo positions in these regimes.”
MethodTensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., Corrado, G. S., Davis, A., Dean, J., Devin, M., et al. (2016) · arXiv · doi:10.48550/arxiv.1603.04467 · Cited by 9,408 · see citing works
Formation of Galaxies and Clusters of Galaxies by Self-Similar Gravitational Condensation
Press, W., & Schechter, P. (1974) · The Astrophysical Journal · doi:10.1086/152650 · Cited by 4,042 · see citing works
On the Infall of Matter into Clusters of Galaxies and Some Effects on Their Evolution
Gunn, J., & Richard, I. G. J. (1972) · The Astrophysical Journal · doi:10.1086/151605 · Cited by 3,013 · cited 1× · see citing works
“The simplest models such as Spherical collapse [19] simply predict that halos form where overdensity exceeds a”
BackgroundLarge scale bias and the peak background split
Sheth, R., & Tormen, G. (1999) · Monthly Notices of the Royal Astronomical Society · doi:10.1046/j.1365-8711.1999.02692.x · Cited by 2,326 · cited 1× · see citing works
“Other halo formation models such as Ellipsoidal collapse advocate the role of shear [22, 23] while assembly bias models suggest that other environmental features [24, 25] also play a role in halo formation.”
BackgroundExcursion set mass functions for hierarchical Gaussian fluctuations
Bond, J., Cole, S., Efstathiou, G., & Kaiser, N. (1991) · The Astrophysical Journal · doi:10.1086/170520 · Cited by 1,554 · cited 1× · see citing works
“certain threshold with the mass of the halos corresponding to the largest scale at which it does so [20, 21].”
BackgroundSheth, R., Mo, H.-J., & Tormen, G. (1999) · Monthly Notices of the Royal Astronomical Society · doi:10.1046/j.1365-8711.2001.04006.x · Cited by 1,467 · cited 1× · see citing works
“Other halo formation models such as Ellipsoidal collapse advocate the role of shear [22, 23] while assembly bias models suggest that other environmental features [24, 25] also play a role in halo formation.”
BackgroundBerlind, A., & Weinberg, D. (2001) · The Astrophysical Journal · doi:10.1086/341469 · Cited by 847 · see citing works
Mandelbaum, R., Seljak, U., Kauffmann, G., Hirata, C. M., & Brinkmann, J. (2006) · Monthly Notices of the Royal Astronomical Society · doi:10.1111/j.1365-2966.2006.10156.x · Cited by 728 · see citing works
Self-citationImproving Cosmological Distance Measurements by Reconstruction of the Baryon Acoustic Peak
Eisenstein, D., Seo, H., Sirko, E., & Spergel, D. (2006) · The Astrophysical Journal · doi:10.1086/518712 · Cited by 456 · see citing works
Measuring Cosmological Parameters with Galaxy Surveys
Tegmark, M. (1997) · Physical Review Letters · doi:10.1103/physrevlett.79.3806 · Cited by 353 · cited 1× · see citing works
“Briefly, assuming Gaussian likelihood for band powers of power spectrum, the Fisher matrix is approximately [41],”
Backgroundnbodykit: An Open-source, Massively Parallel Toolkit for Large-scale Structure
Hand, N., Feng, Y., Beutler, F., Li, Y., Modi, C., Seljak, U., & Slepian, Z. (2018) · The Astronomical Journal · doi:10.3847/1538-3881/aadae0 · Cited by 282 · see citing works
Self-citationImproved Forecasts for the Baryon Acoustic Oscillations and Cosmological Distance Scale
Seo, H., & Eisenstein, D. (2007) · The Astrophysical Journal · doi:10.1086/519549 · Cited by 278 · see citing works
Bayesian physical reconstruction of initial conditions from large-scale structure surveys
Jasche, J., & Wandelt, B. D. (2012) · Monthly Notices of the Royal Astronomical Society · doi:10.1093/mnras/stt449 · Cited by 255 · cited 1× · see citing works
“Linear bias model is also used in [4] to relate matter to the halo field, but they treat galaxies as a Poisson sampling of this biased field and develop the likelihood function to take this sampling into account.”
MethodCMU DeepLens: deep learning for automatic image-based galaxy–galaxy strong lens finding
Lanusse, F., Ma, Q., Li, N., Collett, T. E., Li, C.-L., Ravanbakhsh, S., Mandelbaum, R., & Póczos, B. (2018) · Monthly Notices of the Royal Astronomical Society · doi:10.1093/mnras/stx1665 · Cited by 191 · cited 1× · see citing works
“Over last decade, sophisticated versions of these neural networks have come to be widely used in cosmology for diverse purposes, from parameter estimation [15], to model discrimination [16], creating virtual universe [17] and to detecting strong lenses [18].”
BackgroundFastPM: a new scheme for fast simulations of dark matter and haloes
Feng, Y., Chu, M., Seljak, U., & McDonald, P. (2016) · Monthly Notices of the Royal Astronomical Society · doi:10.1093/mnras/stw2123 · Cited by 177 · cited 12× · see citing works
“J C A P 1 0 ( 2 0 1 8 ) 0 2 8 to g et h al o m as s & p os it io n fi el d N eu ra l N et w or k Fo rw ar d M od el Ba ck pr op ag at io n Optimization cycle for reconstruction Li ke lih oo d of th e da ta Initial Matter Density FastPM sim ulation Final Matter Density Neural Network Model Halo Mass Field True Halo Mas…”
BackgroundMethodInfluentialSelf-citationRodríguez-Torres, S. A., Chuang, C., Prada, F., Guo, H., Klypin, A., Behroozi, P., Hahn, C., Comparat, J., Yepes, G., Montero-Dorta, A., et al. (2015) · Monthly Notices of the Royal Astronomical Society · doi:10.1093/mnras/stw1014 · Cited by 160 · see citing works
CosmoGAN: creating high-fidelity weak lensing convergence maps using Generative Adversarial Networks
Mustafa, M., Bard, D., Bhimji, W., Lukić, Z., Al-Rfou, R., & Kratochvil, J. M. (2019) · Computational Astrophysics and Cosmology · doi:10.1186/s40668-019-0029-9 · Cited by 127 · see citing works
HIGH-PRECISION PREDICTIONS FOR THE ACOUSTIC SCALE IN THE NONLINEAR REGIME
Seo, H., Eckel, J., Eisenstein, D., Mehta, K., Metchnik, M., Padmanabhan, N., Pinto, P., Takahashi, R., White, M., & Xu, X.-Y. (2009) · The Astrophysical Journal · doi:10.1088/0004-637x/720/2/1650 · Cited by 117 · see citing works
Towards optimal extraction of cosmological information from nonlinear data
Seljak, U., Aslanyan, G., Feng, Y., & Modi, C. (2017) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2017/12/009 · Cited by 113 · see citing works
Self-citationWang, H.-Y., Mo, H.-J., Yang, X.-H., Jing, Y., & Lin, W. (2014) · The Astrophysical Journal · doi:10.1088/0004-637x/794/1/94 · Cited by 113 · see citing works
The Initial Conditions of the Universe from Constrained Simulations
Kitaura, F. (2012) · arXiv.org · doi:10.1093/mnrasl/sls029 · Cited by 93 · see citing works
Beyond assembly bias: exploring secondary halo biases for cluster-size haloes
Mao, Y.-Y., Zentner, A., & Wechsler, R. (2017) · Monthly Notices of the Royal Astronomical Society · doi:10.1093/mnras/stx3111 · Cited by 87 · cited 1× · see citing works
“Other halo formation models such as Ellipsoidal collapse advocate the role of shear [22, 23] while assembly bias models suggest that other environmental features [24, 25] also play a role in halo formation.”
BackgroundThe DTFE public software - The Delaunay Tessellation Field Estimator code
Cautun, M., & van de Weygaert, R. (2011) · doi:10.48550/arxiv.1105.0370 · Cited by 84
Reconstructing baryon oscillations
Noh, Y., White, M., & Padmanabhan, N. (2009) · Physical review. D. Particles, fields, gravitation, and cosmology/Physical review. D. Particles and fields · doi:10.1103/physrevd.80.123501 · Cited by 83 · cited 1× · see citing works
“BAO peak is damped due to the Silk damping and non-linear evolution and the information lost in the latter damping is modeled well with a Gaussian damping model ([42]) in theory and the propagator defined above in practice.”
Iterative initial condition reconstruction
Schmittfull, M., Baldauf, T., & Zaldarriaga, M. (2017) · Physical review. D/Physical review. D. · doi:10.1103/physrevd.96.023505 · Cited by 75 · cited 1× · see citing works
“Note that unlike figure 4 of [3], we do not rescale the reconstructed fields here to have the correct transfer function.”
BackgroundFast cosmological parameter estimation using neural networks
Auld, T., Bridges, M., Hobson, M., & Gull, S. (2006) · Monthly Notices of the Royal Astronomical Society Letters · doi:10.1111/j.1745-3933.2006.00276.x · Cited by 71 · cited 1× · see citing works
“Over last decade, sophisticated versions of these neural networks have come to be widely used in cosmology for diverse purposes, from parameter estimation [15], to model discrimination [16], creating virtual universe [17] and to detecting strong lenses [18].”
BackgroundReconstructing the cosmic density field with the distribution of dark matter haloes
Wang, H.-Y., Mo, H.-J., Jing, Y.-P., Guo, Y.-C., van den Bosch, F. C., & Yang, X.-H. (2008) · doi:10.1111/j.1365-2966.2008.14301.x · Cited by 69 · cited 1×
“For example, [9] has proposed a method to assign matter profiles to halos which have been constructed from”
MethodExtending the modeling of the anisotropic galaxy power spectrum to k = 0.4 hMpc−1
Hand, N., Seljak, U., Beutler, F., & Vlah, Z. (2017) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2017/10/009 · Cited by 68 · see citing works
Self-citationModeling the reconstructed BAO in Fourier space
Seo, H., Beutler, F., Ross, A., & Saito, S. (2015) · Monthly Notices of the Royal Astronomical Society · doi:10.1093/mnras/stw1138 · Cited by 67 · cited 2× · see citing works
“Modeling the sensitivity of the power spectrum to sound horizon in this way leads to the Fisher error given by (for full derivation, see [39, 40])”
MethodRECONSTRUCTING THE INITIAL DENSITY FIELD OF THE LOCAL UNIVERSE: METHODS AND TESTS WITH MOCK CATALOGS
Wang, H.-Y., Mo, H.-J., Yang, X.-H., & van den Bosch, F. C. (2013) · The Astrophysical Journal · doi:10.1088/0004-637x/772/1/63 · Cited by 64 · see citing works
Halo bias in Lagrangian Space: Estimators and theoretical predictions
Modi, C., Castorina, E., & Seljak, U. (2016) · doi:10.1093/mnras/stx2148 · Cited by 60 · cited 1×
“It is a significantly higher reduction as compared to simplistic bias models often used to model halo fields [36].”
ResultSelf-citationMachine learning cosmological structure formation
Lucie-Smith, L., Peiris, H., Pontzen, A., & Lochner, M. (2018) · Monthly notices of the Royal Astronomical Society · doi:10.1093/mnras/sty1719 · Cited by 56 · see citing works
Han, J.-X., Li, Y., Jing, Y., Nishimichi, T., Wang, W., & Jiang, C.-Y. (2018) · Monthly notices of the Royal Astronomical Society · doi:10.1093/mnras/sty2822 · Cited by 40
Yoo, J., & Seljak, U. (2012) · Physical review. D. Particles, fields, gravitation, and cosmology/Physical review. D. Particles and fields · doi:10.1103/physrevd.86.083504 · Cited by 29 · see citing works
Self-citationExploring the posterior surface of the large scale structure reconstruction
Feng, Y., Seljak, U., & Zaldarriaga, M. (2018) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2018/07/043 · Cited by 27 · see citing works
Self-citationYu, Y., Zhu, H.-M., & Pen, U. (2017) · The Astrophysical Journal · doi:10.3847/1538-4357/aa89e7 · Cited by 24 · see citing works
Zhu, H.-M., Yu, Y., Pen, U., Chen, X.-L., & Yu, H.-R. (2016) · Physical review. D/Physical review. D. · doi:10.1103/physrevd.96.123502 · Cited by 22 · see citing works
A fast algorithm for identifying Friends-of-Friends halos
Feng, Y., & Modi, C. (2016) · Astronomy and Computing · doi:10.1016/j.ascom.2017.05.004 · Cited by 12 · cited 1×
“presented in [34] using a linking length of 0.”
MethodSelf-citation
Flagged first, then by citations
6 of the 48 references couldn’t be matched to a known paper.
Cited by
Sorted by citations
Cosmological parameters and neutrino masses from the final Planck and full-shape BOSS data
Ivanov, M., Simonović, M., & Zaldarriaga, M. (2019) · Physical Review D · doi:10.1103/physrevd.101.083504 · Cited by 108 · see citing works
The Dawes Review 10: The impact of deep learning for the analysis of galaxy surveys
Huertas-Company, M., & Lanusse, F. (2023) · Publications of the Astronomical Society of Australia · doi:10.1017/pasa.2022.55 · Cited by 102 · see citing works
“One of the first approaches to learn this connection was proposed in Modi et al. (2018), and assumed a shallow neural network mapping between the local 3D dark matter density and "halo mask" and "halo mass" fields.”
BackgroundModeling biased tracers at the field level
Schmittfull, M., Simonović, M., Assassi, V., & Zaldarriaga, M. (2018) · Physical Review D · doi:10.1103/physrevd.100.043514 · Cited by 95 · see citing works
Nearest neighbour distributions: New statistical measures for cosmological clustering
Banerjee, A., & Abel, T. (2020) · Monthly Notices of the Royal Astronomical Society · doi:10.1093/mnras/staa3604 · Cited by 85 · see citing works
Reconstructing large-scale structure with neutral hydrogen surveys
Modi, C., White, M., Slosar, A., & Castorina, E. (2019) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2019/11/023 · Cited by 76 · see citing works
“of the foreground wedge make reconstruction much less eective [18], though some of the lost signal can be recovered with other surveys [19]. Our approach provides another route to BAO reconstruction [24], so it is interesting to ask how the performance of the algorithm compares to standard reconstruction and how it is…”
BackgroundMethodInfluentialSelf-citationA rigorous EFT-based forward model for large-scale structure
Schmidt, F., Elsner, F., Jasche, J., Nguyen, N.-M., & Lavaux, G. (2018) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2019/01/042 · Cited by 76
Villaescusa-Navarro, F., Genel, S., Anglés-Alcázar, D., Thiele, L., Davé, R., Narayanan, D., Nicola, A., Li, Y., Villanueva-Domingo, P., Wandelt, B. D., et al. (2021) · Astrophysical Journal Supplement Series · doi:10.3847/1538-4365/ac5ab0 · Cited by 69 · see citing works
“…up astrophysics (Makinen et al. 2021; Villanueva-Domingo & Villaescusa-Navarro 2021), painting stars and gas properties on the dark matter field (Modi et al. 2018; Jo & Kim 2019; Tröster et al. 2019; Yip et al. 2019; Zhang et al. 2019; Kasmanoff et al. 2020; Thiele et al. 2020; Harrington et…”
BackgroundAI-assisted superresolution cosmological simulations
Li, Y., Ni, Y., Croft, R. A. C., Matteo, T. D., Bird, S., & Yu, F. (2021) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2022038118 · Cited by 99 · see citing works
“NN models have also been trained to predict galaxies (10, 11) and 21 cm emission from neutral hydrogen (12), from simulations that only contain dark matter.”
BackgroundSelf-citationFlowPM: Distributed TensorFlow implementation of the FastPM cosmological N-body solver
Modi, C., Lanusse, F., & Seljak, U. (2020) · Astronomy and Computing · doi:10.1016/j.ascom.2021.100505 · Cited by 51 · see citing works
“astPM, vmad and abopt (hence11 forth referred to as FastPM reconstruction). Both the reconstructions follow the same adiabatic optimization schedule to promote converge on large scales as described inModi et al. (2018);Feng et al.(2018). FastPMreconstruction uses gradient descent algorithm with single step line-search…”
BackgroundMethodSelf-citationForecasts on primordial non-Gaussianity from 21 cm intensity mapping experiments
Karagiannis, D., Slosar, A., & Liguori, M. (2019) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2020/11/052 · Cited by 51 · see citing works
A volumetric deep Convolutional Neural Network for simulation of mock dark matter halo catalogues
Berger, P., & Stein, G. (2019) · Monthly Notices of the Royal Astronomical Society · doi:10.1093/mnras/sty2949 · Cited by 50 · see citing works
“…Gillet et al. 2018), simulating twodimensional slices of the non-linear density field (Rodriguez et al. 2018), initial conditions reconstruction (Modi et al. 2018), as well as classifying evolved structures in N-body simulations (Aragon-Calvo 2018) (who uses a similar CNN architecture to that…”
MethodTranslation and rotation equivariant normalizing flow (TRENF) for optimal cosmological analysis
Dai, B., & Seljak, U. (2022) · Monthly Notices of the Royal Astronomical Society · doi:10.1093/mnras/stac2010 · Cited by 48 · see citing works
Self-citationBayer, A., Banerjee, A., & Feng, Y. (2020) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2021/01/016 · Cited by 39 · see citing works
“experimental data. Furthermore, after applying FastPM’s inbuilt halo-nder, one can reconstruct the initial conditions of the universe, for example BAO, from galaxy positions [74,75] and luminosities [76]. Because massive neutrinos modify the information content of the CDM and total matter elds, this work will enable t…”
MethodSelf-citationDai, B., & Seljak, U. (2020) · Proceedings of the National Academy of Sciences of the United States of America · doi:10.1073/pnas.2020324118 · Cited by 37 · see citing works
“Previous work in this area covers a wide array of tasks, such as identifying halos (protohalos) [13, 14, 15, 16], producing 3D galaxy distribution [17], generating tSZ signals [18], predicting dark matter annihilation feedback [19], learning neutrino e ff ects [20], emulating high resolution…”
BackgroundSelf-citationHorowitz, B., Lee, K., White, M., Krolewski, A., & Ata, M. (2019) · Astrophysical Journal · doi:10.3847/1538-4357/ab4d4c · Cited by 36 · see citing works
Cosmological parameter estimation from large-scale structure deep learning
Pan, S., Liu, M.-X., Forero-Romero, J., Sabiu, C., Li, Z., Miao, H., & Li, X.-D. (2019) · Science China Physics Mechanics and Astronomy · doi:10.1007/s11433-020-1586-3 · Cited by 35 · see citing works
“…Rodriguez et al. 2018; Perraudin et al. 2019; Mnchmeyer & Smith 2019; Mishra et al. 2019), cosmic large scale structure (Lucie-Smith et al. 2018; Modi et al. 2018; Berger & Stein 2019; He et al. 2019; Lucie-Smith et al. 2019; Pfeffer et al. 2019; Ramanah et al. 2019a; Trster et al. 2019; Zhang…”
BackgroundConsistency tests of field level inference with the EFT likelihood
Kostić, A., Nguyen, N.-M., Schmidt, F., & Reinecke, M. (2022) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2023/07/063 · Cited by 33 · see citing works
HInet: Generating Neutral Hydrogen from Dark Matter with Neural Networks
Wadekar, D., Villaescusa-Navarro, F., Ho, S., & Perreault-Levasseur, L. (2021) · The Astrophysical Journal · doi:10.3847/1538-4357/ac033a · Cited by 32 · see citing works
Cosmology inference at the field level from biased tracers in redshift-space
Stadler, J., Schmidt, F., & Reinecke, M. (2023) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2023/10/069 · Cited by 30 · see citing works
“The recovery of initial conditions from simulated mock galaxies at fixed cosmology has been explored with various forward models, such as Lagrangian Perturbation Theory (LPT) [15–17], full particle-mesh simulations [18] or neural networks [19–25].”
Painting halos from cosmic density fields of dark matter with physically motivated neural networks
Ramanah, D. K., Charnock, T., & Lavaux, G. (2019) · Physical Review D · doi:10.1103/physrevd.100.043515 · Cited by 30 · see citing works
“Modi et al. (2018) proposed a framework, with the end product being the converse, to reconstruct initial conditions from the halo fields using the multilayer perceptron, i.e. fully connected neural networks.”
MethodNonlinear 3D cosmic web simulation with heavy-tailed generative adversarial networks
Feder, R. M., Berger, P., & Stein, G. (2020) · Physical Review D · doi:10.1103/physrevd.102.103504 · Cited by 30 · see citing works
Baryon acoustic oscillations reconstruction using convolutional neural networks
Mao, T.-X., Wang, J., Li, B., Cai, Y.-C., Falck, B., Neyrinck, M., & Szalay, A. (2020) · Monthly Notices of the Royal Astronomical Society · doi:10.1093/mnras/staa3741 · Cited by 29 · see citing works
ADDGALS: Simulated Sky Catalogs for Wide Field Galaxy Surveys
Wechsler, R., DeRose, J., Busha, M., Becker, M., Rykoff, E., & Evrard, A. (2021) · Astrophysical Journal · doi:10.3847/1538-4357/ac5b0a · Cited by 28 · see citing works
“This approach is similar in spirit to other recent works that employ statistical learning techniques to connect the dark matter distribution to the distribution of biased tracers (Modi et al. 2018; Berger & Stein 2019; Ramanah et al. 2019; Tröster et al. 2019; Zhang et al. 2019; Dai & Seljak 2021).”
ResultNguyen, N.-M., Schmidt, F., Lavaux, G., & Jasche, J. (2020) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2021/03/058 · Cited by 26 · see citing works
“Here, we are interested in the aforementioned more direct route: forward modeling the matter and galaxy clustering directly at the field-level [9–24].”
BackgroundQin, F., Parkinson, D., Hong, S. E., & Sabiu, C. (2023) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2023/06/062 · Cited by 25 · see citing works
Bayer, A., Modi, C., & Ferraro, S. (2022) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2023/06/046 · Cited by 24 · see citing works
“Note that one could also consider a neural network forward model for the halo overdensity field [6] to model individual discrete objects, but we do not pursue that here.”
BackgroundMethodInfluentialSelf-citationNi, Y.-Y., Li, Y., Lachance, P., Croft, R., di Matteo, T., Bird, S., & Feng, Y. (2021) · Monthly notices of the Royal Astronomical Society · doi:10.1093/mnras/stab2113 · Cited by 24 · see citing works
Self-citationKaragiannis, D., Fonseca, J., Maartens, R., & Camera, S. (2020) · Physics of the Dark Universe · doi:10.1016/j.dark.2021.100821 · Cited by 23
Weighing neutrinos with the halo environment
Banerjee, A., Castorina, E., Villaescusa-Navarro, F., Court, T., & Viel, M. (2019) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2020/06/032 · Cited by 23
“[78]) in order for the environmental scale-dependent bias effect to be clearly detected.”
BackgroundCosmological constraints from the power spectrum and bispectrum of 21cm intensity maps
Karagiannis, D., Maartens, R., & Randrianjanahary, L. (2022) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2022/11/003 · Cited by 22 · see citing works
Revealing the Local Cosmic Web from Galaxies by Deep Learning
Hong, S. E., Jeong, D., Hwang, H. S., & Kim, J. (2021) · The Astrophysical Journal · doi:10.3847/1538-4357/abf040 · Cited by 28 · see citing works
“6 Note that Modi et al. (2018) performed a similar study to reconstruct the (initial) density perturbation from the dark matter halo distributions by DL, but they focused more on large scales such as baryon acoustic oscillation (BAO) rather than relatively small scales such as…”
BackgroundMethodWadekar, D., Thiele, L., Villaescusa-Navarro, F., Hill, J., Cranmer, M., Spergel, D., Battaglia, N., Anglés-Alcázar, D., Hernquist, L., & Ho, S. (2022) · Proceedings of the National Academy of Sciences of the United States of America · doi:10.1073/pnas.2202074120 · Cited by 20 · see citing works
Effective cosmic density field reconstruction with convolutional neural network
Chen, X., Zhu, F., Gaines, S., & Padmanabhan, N. (2023) · Monthly Notices of the Royal Astronomical Society · doi:10.1093/mnras/stad1868 · Cited by 24 · see citing works
“Modi et al. (2018) propose a forward modeling based approach for reconstruction and a fully-connected neural network is applied to predict halo mass and position field from matter density field.”
MethodModi, C., Lanusse, F., Seljak, U., Spergel, D., & Perreault-Levasseur, L. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2104.12864 · Cited by 19 · see citing works
“Over the last several years, different works have built physically motivated annealing schemes relying on the dynamics of the forward models to solve this problem (Seljak et al. (2017); Modi et al. (2018; 2019)).”
MethodSelf-citationCosmological information in perturbative forward modeling
Cabass, G., Simonovi'c, M., & Zaldarriaga, M. (2023) · Physical Review D · doi:10.1103/physrevd.109.043526 · Cited by 17 · see citing works
“One alternative to using n -point functions is forward modeling [28–33].”
MethodModeling assembly bias with machine learning and symbolic regressionPreprint
Wadekar, D., Villaescusa-Navarro, F., Ho, S., & Perreault-Levasseur, L. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.00111 · Cited by 17 · see citing works
“There is also a problem of data sparsity when using DNNs directly at the field level: most of the cosmological information comes from halos, which are found rarely in the input matter field (51, 52, 54, 57).”
BackgroundThe HalfDome multi-survey cosmological simulations: N-body simulations
Bayer, A., Zhong, Y., Li, Z., DeRose, J., Feng, Y., & Liu, J. (2024) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2025/05/016 · Cited by 14
“Its combination of speed and accuracy has facilitated hierarchical Bayesian reconstruction of the initial cosmic field [37, 38] and field-level inference [39] using GPU adaptations of the code [40, 41].”
Self-citationReconstructing the universe with variational self-boosted sampling
Modi, C., Li, Y., & Blei, D. (2022) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2023/03/059 · Cited by 14 · see citing works
“The simplest inference is to reconstruct a maximum-a-posterior (MAP) estimate by maximizing this posterior with traditional gradient based optimization algorithms [4, 8, 9] or by using learnt optimization [10].”
Self-citationLINNA: Likelihood Inference Neural Network Accelerator
To, C., Rozo, E., Krause, E., Wu, H.-Y., Wechsler, R., & Salcedo, A. (2022) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2023/01/016 · Cited by 14 · see citing works
Horowitz, B., Zhang, B. J., Lee, K., & Kooistra, R. (2020) · The Astrophysical Journal · doi:10.3847/1538-4357/abca35 · Cited by 14 · see citing works
hyphy: Deep Generative Conditional Posterior Mapping of Hydrodynamical Physics
Horowitz, B., Dornfest, M. A. A., Lukic, Z., & Harrington, P. (2021) · Astrophysical Journal · doi:10.3847/1538-4357/ac9ea7 · Cited by 13 · see citing works
“This is similar in spirit to work done in Modi et al. (2018), where a neural network was used to paint halo fields onto forward-modeling dark matter density.”
Cosmic Velocity Field Reconstruction Using AI
Wu, Z., Zhang, Z., Pan, S., Miao, H., Luo, X., Wang, X., Sabiu, C. G., Forero-Romero, J., Wang, Y., & Li, X.-D. (2021) · The Astrophysical Journal · doi:10.3847/1538-4357/abf3bb · Cited by 22 · see citing works
“…Perraudin et al. 2019; Münchmeyer & Smith 2019; Mishra et al. 2019), the large-scale structure (Ravanbakhsh et al. 2017; Lucie-Smith et al. 2018; Modi et al. 2018; Berger & Stein 2019; He et al. 2019; Lucie-Smith et al. 2019; Pfeffer et al. 2019; Ramanah et al. 2019; Tröster et al. 2019; Zhang…”
Background(DarkAI) Mapping the large-scale density field of dark matter using artificial intelligence
Wang, Z.-T., Shi, F., Yang, X.-H., Li, Q.-Y., Liu, Y.-M., & Li, X.-P. (2023) · Science China Physics Mechanics and Astronomy · doi:10.1007/s11433-023-2192-9 · Cited by 12 · see citing works
“…structure formation (Lucie-Smith et al. 2018; He et al. 2019), populating the halo position and mass field from the nonlinear matter field (Modi et al. 2018; Lucie-Smith et al. 2019; Berger & Stein 2019; Kodi Ramanah et al. 2019), classifying the LSS of the universe (Aragon-Calvo 2019),…”
BackgroundOn the connection between field-level inference and n-point correlation functions
Schmidt, F. (2025) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2025/09/056 · Cited by 11 · see citing works
“(3.29) is precisely the Fisher information contained in the linear power spectrum up to the scale Λ (see also the discussion in [51] and [22]).”
InfluentialStadler, J., Schmidt, F., & Reinecke, M. (2024) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2025/04/089 · Cited by 11 · see citing works
Deep learning insights into cosmological structure formation
Lucie-Smith, L., Peiris, H., Pontzen, A., Nord, B., & Thiyagalingam, J. (2020) · Physical Review D · doi:10.1103/physrevd.109.063524 · Cited by 10 · see citing works
Reconstruction with velocities
Zhu, H.-M., White, M., Ferraro, S., & Schaan, E. (2019) · Monthly notices of the Royal Astronomical Society · doi:10.1093/mnras/staa1002 · Cited by 10
“Forward modeling methods have also been explored (Seljak et al. 2017; Modi et al. 2018; Schmidt et al. 2019; Elsner et al. 2019; Modi et al. 2019).”
MethodOn the primordial information available to galaxy redshift surveys
McQuinn, M. (2020) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2021/06/024 · Cited by 9 · see citing works
BaryonBridge: Stochastic Interpolant Model for Fast Hydrodynamical Simulations
Horowitz, B., Cuesta-Lázaro, C., & Yehia, O. (2025) · doi:10.48550/arxiv.2510.19224 · Cited by 8
“(Seljak et al., 2017; Modi et al., 2018; Horowitz et al., 2019; 2021) For our particle mesh simulations, we take the particle/velocity initial conditions given in the CAMELS simulation set and use them as ICs for our particle mesh code.”
Fast, accurate and perturbative forward modeling of galaxy clustering. Part II. Redshift space
Stadler, J., Schmidt, F., Reinecke, M., & Esposito, M. (2024) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2025/11/055 · Cited by 8 · see citing works
Mind the gap: the power of combining photometric surveys with intensity mapping
Modi, C., White, M., Castorina, E., & Slosar, A. (2021) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2021/10/056 · Cited by 8 · see citing works
“Previous work has suggested correcting for this using simulations [23, 24], but implementing such a correction is beyond the scope of this work.”
BackgroundMethodSelf-citationField-level reconstruction from foreground-contaminated 21-cm maps
Chen, S.-F., Chen, K.-F., & Dvorkin, C. (2025) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2025/11/082 · Cited by 7 · see citing works
Testing large-scale structure measurements against Fisher matrix predictions
Foroozan, S., Krolewski, A., & Percival, W. (2021) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2021/10/044 · Cited by 7
Biased Tracer Reconstruction with Halo Mass Information
Liu, Y., Yu, Y., & Li, B.-J. (2020) · Astrophysical Journal Supplement Series · doi:10.3847/1538-4365/abe868 · Cited by 7
“…tackle this problem, including forward modeling (e.g., Jasche & Wandelt 2013; Kitaura 2013; Wang et al. 2013; Seljak et al. 2017; Feng et al. 2018; Modi et al. 2018, 2019; Jasche & Lavaux 2019, etc.), backward reconstruction (hereafter reconstruction, e.g., Frisch et al. 2002; Brenier et al.…”
Nonlinear Reconstruction of the Velocity Field
Yu, Y., & Zhu, H.-M. (2019) · Astrophysical Journal · doi:10.3847/1538-4357/ab5580 · Cited by 7
Initial conditions from galaxies: machine-learning subgrid correction to standard reconstruction
Parker, L., Bayer, A., & Seljak, U. (2025) · arXiv.org · doi:10.1088/1475-7516/2025/09/039 · Cited by 5
“The BAO can then be measured in this reconstructed linear field, with [23] showing that the MAP approach significantly outperforms standard BAO reconstruction, and [34, 35] showing improvements of up to 50% on the BAO scale using Monte Carlo with an Effective Field Theory forward model.”
Self-citationRuchika, Mukherjee, P., & Favale, A. (2026) · Astrophysical Journal · doi:10.3847/1538-4357/ae650b · Cited by 4 · see citing works
“Notably, ML has demonstrated its ability to significantly reduce scatter in mass estimates of clusters compared to traditional methods ( M. Ntampaka et al. 2015; C. Modi et al. 2018 ) .”
Non-Gaussian galaxy stochasticity and the noise-field formulation
Rubira, H., & Schmidt, F. (2025) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2026/08/023 · Cited by 3
Krywonos, J., Kvasiuk, Y., Johnson, M. C., & Münchmeyer, M. (2025) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2026/03/037 · Cited by 3 · see citing works
“There are many models for δ m → δ g in the literature, working at different scales, some examples including [48, 49].”
AI-powered Reconstruction of Dark Matter Velocity Fields from Redshift-space Halo Distribution
Xiao, X., Ding, J., Luo, X., Lan, S.-K., Xiao, L., Liu, S., Wang, X., Zhang, L., & Li, X.-D. (2024) · Astrophysical Journal · doi:10.3847/1538-4357/ae10af · Cited by 3 · see citing works
“…the distribution of matter (Ravanbakhsh et al. 2017; Lucie-Smith et al. 2018; Pan et al. 2020; Lazanu 2021), identifying DM halos, and reconstructing the initial conditions of the universe using machine learning (Modi et al. 2018; Berger & Stein 2019; Lucie-Smith et al. 2019; Ramanah et al. 2019).”
ELUCID. VIII. Simulating the Coma Galaxy Cluster to Calibrate Model and Understand Feedback
Luo, X., Wang, H.-Y., Cui, W., Mo, H.-J., Li, R.-J., Jing, Y.-P., Katz, N., Davé, R., Yang, X.-H., Chen, Y.-Y., et al. (2024) · Astrophysical Journal · doi:10.3847/1538-4357/ad392e · Cited by 3 · see citing works
Karagiannis, D., Ballardini, M., & Maartens, R. (2026) · doi:10.48550/arxiv.2606.28172 · Cited by 2
“Although reconstruction methods can recover part of the lost information [146, 155–164], here we adopt a foreground-avoidance approach and exclude all scales that satisfy k ∥ k ∥ , min , taking k ∥ , min = 0 .”
On the relation between field-level posteriors, correlators, and their likelihoods
Pietroni, M., & Schmidt, F. (2026) · Journal of Cosmology and Astroparticle Physics · doi:10.1088/1475-7516/2026/09/011 · Cited by 2 · see citing works
Effective bias expansion for circumventing 21 cm foregrounds
Qin, W., Chen, K.-F., Schutz, K., & Liu, A. (2025) · doi:10.48550/arxiv.2508.13268 · Cited by 2
Constraining Neutrino Cosmologies with Nonlinear Reconstruction
Zang, S.-H., & Zhu, H.-M. (2023) · Astrophysical Journal · doi:10.3847/1538-4357/ad0f97 · Cited by 2 · see citing works
“…on different principles, e.g., iterative methods (Hada & Eisenstein 2018, 2019), optimization techniques (Seljak et al. 2017; Feng et al. 2018; Modi et al. 2018), forwarding modeling (Jasche & Wan-delt 2013; Wang et al. 2014), effective field theory based likelihood (Schmidt et al. 2019;…”
MethodJagvaral, Y., Lanusse, F., Singh, S., Mandelbaum, R., Ravanbakhsh, S., & Campbell, D. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2212.05596 · Cited by 1 · see citing works
“In some cases (Agarwal et al., 2018; Modi et al., 2018; Zhang et al., 2019) the approach has been to ’paint’ galaxy properties onto N-body (dark matter – gravity only) simulations.”
Dipole distortions in the intergalactic medium
Inman, D., Pen, U., & Villaescusa-Navarro, F. (2018) · Monthly notices of the Royal Astronomical Society · doi:10.1093/mnras/stz1542 · Cited by 1
“However, there are a number of nonlinear reconstruction methods (e.g. Zhu et al. (2017); Schmittfull et al. (2017); Modi et al. (2018); Hada & Eisenstein (2018); Shi et al. (2018)) which use alternate methods to determine the displacement field.”
MethodInsights into cosmological structure formation with machine learning.
Lucie-Smith, L. (2020) · doi:10.2172/1771179 · Cited by 1 · see citing works
“2019), the non-linear density field and the halo distribution (Charnock et al. 2019; Kodi Ramanah et al. 2019; Modi et al. 2018) and the dark matter and galaxy distributions (Zhang et al.”
BackgroundSaleh, N., Mohd-Sahabuddin, M. F., Ghafar, N. A., Awaluddin, Z. L., & Abdrazak, B. (2026) · ALAM CIPTA International Journal Of Sustainable Tropical Design & Practice · doi:10.47836/ac.19.3.paper06
DeepVoid: A Deep Learning Void Detector
Kumagai, S., Vogeley, M. S., Aragon-Calvo, M., Douglass, K., Benzvi, S., & Neyrinck, M. (2025) · Astrophysical Journal · doi:10.3847/1538-4357/ae2c80
“…the cosmic web cheaply using generative networks (Rodr´ıguez et al. 2018), reconstructing initial conditions from galactic luminosities and positions (Modi et al. 2018), studying the formation of dark matter (DM) halos (Lucie-Smith et al. 2019), recovering the underlying DM distribution from…”
Bartlett, D., Ho, M., & Wandelt, B. D. (2024) · Astrophysical Journal Letters · doi:10.3847/2041-8213/ad97b9
- Authors
- Modi, C., Feng, Y., & Seljak, U.
- Affiliations
- University of California, Berkeley (US)
- Lawrence Berkeley National Laboratory (US)
- ORCID
- Modi, Chirag
- Venue
- Journal of Cosmology and Astroparticle Physics
- Publisher
- IOP Publishing
- Published
- 2018-10-17
- Volume
- 2018
- Issue
- 10
- Pages
- 028-028
- ISSN
- 1475-7508 (ISSN-L)
- 1475-7516
- OpenAlex
- W2799807151
- Cited by
- 86
- References
- 48
- Citations by year
- 2026: 5 · 2025: 12 · 2024: 9 · 2023: 14 · 2022: 7 · 2021: 16 · 2020: 11 · 2019: 10 · 2018: 2
- FWCI
- 5.23
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Licence
- Locations
- Journal of Cosmology and Astroparticle Physics (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- U.S. Department of Energy (-AC02-05CH11231, 05CH11231, DE AC02 05CH11231, DE-AC02-)
- National Aeronautics and Space Administration (NNX15AL17G, DE-AC02-05CH11231)
- Office of Science (DEAC02–05CH11231, AC02-05CH11231, DE-AC02, 05CH11231)
- Topics
- Physical Sciences › Physics and Astronomy › Astronomy and Astrophysics › Cosmology and Gravitation Theories
- Physical Sciences › Physics and Astronomy › Astronomy and Astrophysics › Galaxies: Formation, Evolution, Phenomena
- Physical Sciences › Physics and Astronomy › Nuclear and High Energy Physics › Dark Matter and Cosmic Phenomena
- Keywords
- cosmological reconstruction, artificial neural networks, shot noise, baryon acoustic oscillations, power spectrum reconstruction, dark matter, galaxy redshift surveys, annealing
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
