arXiv subjects

Francisco Villaescusa-Navarro

Publications and source records attributed to Francisco Villaescusa-Navarro.

At least 145 records · Page 8Linked to original sources

arXiv2021

The Quijote simulations

The Quijote simulations are a set of 44,100 full N-body simulations spanning more than 7,000 cosmological models in the $\{Ω_{\rm m}, Ω_{\rm b}, h, n_s, σ_8, M_ν, w \}$ hyperplane. At a single redshift the simulations contain more than 8.5 trillions of particles over a combined volume of 44,100 $(h^{-1}{\rm Gpc})^3$; each simulation follow the evolution of $256^3$, $512^3$ or $1024^3$ particles in a box of $1~h^{-1}{\rm Gpc}$ length. Billions of dark matter halos and cosmic voids have been identified in the simulations, whose runs required more than 35 million core hours. The Quijote simulations have been designed for two main purposes: 1) to quantify the information content on cosmological observables, and 2) to provide enough data to train machine learning algorithms. In this paper we describe the simulations and show a few of their applications. We also release the Petabyte of data generated, comprising hundreds of thousands of simulation snapshots at multiple redshifts, halo and void catalogs, together with millions of summary statistics such as power spectra, bispectra, correlation functions, marked power spectra, and estimated probability density functions.

astro-ph.CO↗

arXiv2021

Primordial non-Gaussianity without tails -- how to measure fNL with the bulk of the density PDF

We investigate the possibility to detect primordial non-Gaussianity by analysing the bulk of the probability distribution function (PDF) of late-time cosmic density fluctuations. For this purpose we devise a new method to predict the impact of general non-Gaussian initial conditions on the late-time density PDF. At redshift $z=1$ and for a smoothing scale of 30Mpc/$h$ our predictions agree with the high-resolution Quijote N-body simulations to $\sim 0.2\%$ precision. This is within cosmic variance of a $\sim 100(\mathrm{Gpc}/h)^3$ survey volume. When restricting to this 30Mpc/$h$ smoothing scale and to mildly non-linear densities ($δ[30\mathrm{Mpc}/h] \in [-0.3, 0.4]$) and also marginalizing over potential ignorance of the amplitude of the non-linear power spectrum an analysis of the PDF for such a survey volume can still measure the amplitude of different primordial bispectrum shapes to an accuracy of \smash{$Δf_{\mathrm{NL}}^{\mathrm{loc}}=\pm 7.4\ ,\ Δf_{\mathrm{NL}}^{\mathrm{equi}}=\pm 22.0\ ,\ Δf_{\mathrm{NL}}^{\mathrm{ortho}}=\pm 46.0$} . When pushing to smaller scales and assuming a joint analysis of the PDF with smoothing radii of 30Mpc/$h$ and 15Mpc/$h$ ($δ[15\mathrm{Mpc}/h] \in [-0.4, 0.5]$) this improves to \smash{$Δf_{\mathrm{NL}}^{\mathrm{loc}}=\pm 3.3\ ,\ Δf_{\mathrm{NL}}^{\mathrm{equi}}=\pm 11.0\ ,\ Δf_{\mathrm{NL}}^{\mathrm{ortho}}=\pm 17.0\ $} - even when marginalizing over the non-linear variances at both scales as two free parameters. Especially, such an analysis could simultaneously measure $f_{\mathrm{NL}}$ and the amplitude and slope of the non-linear power spectrum. However, at 15Mpc/$h$ our predictions are only accurate to $\lesssim 0.8\%$ for the considered density range. We discuss how this has to be improved in order to push to these small scales and make full use of upcoming surveys with a PDF-based analysis.

astro-ph.CO↗

arXiv2021

HInet: Generating neutral hydrogen from dark matter with neural networks

Upcoming 21cm surveys will map the spatial distribution of cosmic neutral hydrogen (HI) over very large cosmological volumes. In order to maximize the scientific return of these surveys, accurate theoretical predictions are needed. Hydrodynamic simulations currently are the most accurate tool to provide those predictions in the mildly to non-linear regime. Unfortunately, their computational cost is very high: tens of millions of CPU hours. We use convolutional neural networks to find the mapping between the spatial distribution of matter from N-body simulations and HI from the state-of-the-art hydrodynamic simulation IllustrisTNG. Our model performs better than the widely used theoretical model: Halo Occupation Distribution (HOD) for all statistical properties up to the non-linear scales $k\lesssim1$ h/Mpc. Our method allows the generation of 21cm mocks over very big cosmological volumes with similar properties as hydrodynamic simulations.

astro-ph.CO↗

arXiv2021

The GIGANTES dataset: precision cosmology from voids in the machine learning era

We present GIGANTES, the most extensive and realistic void catalog suite ever released -- containing over 1 billion cosmic voids covering a volume larger than the observable Universe, more than 20 TB of data, and created by running the void finder VIDE on QUIJOTE's halo simulations. The expansive and detailed GIGANTES suite, spanning thousands of cosmological models, opens up the study of voids, answering compelling questions: Do voids carry unique cosmological information? How is this information correlated with galaxy information? Leveraging the large number of voids in the GIGANTES suite, our Fisher constraints demonstrate voids contain additional information, critically tightening constraints on cosmological parameters. We use traditional void summary statistics (void size function, void density profile) and the void auto-correlation function, which independently yields an error of $0.13\,\mathrm{eV}$ on $\sum\,m_ν$ for a 1 $h^{-3}\mathrm{Gpc}^3$ simulation, without CMB priors. Combining halos and voids we forecast an error of $0.09\,\mathrm{eV}$ from the same volume. Extrapolating to next generation multi-Gpc$^3$ surveys such as DESI, Euclid, SPHEREx, and the Roman Space Telescope, we expect voids should yield an independent determination of neutrino mass. Crucially, GIGANTES is the first void catalog suite expressly built for intensive machine learning exploration. We illustrate this by training a neural network to perform likelihood-free inference on the void size function. Cosmology problems provide an impetus to develop novel deep learning techniques, leveraging the symmetries embedded throughout the universe from physical laws, interpreting models, and accurately predicting errors. With GIGANTES, machine learning gains an impressive dataset, offering unique problems that will stimulate new techniques.

astro-ph.CO↗

arXiv2021

deep21: a Deep Learning Method for 21cm Foreground Removal

We seek to remove foreground contaminants from 21cm intensity mapping observations. We demonstrate that a deep convolutional neural network (CNN) with a UNet architecture and three-dimensional convolutions, trained on simulated observations, can effectively separate frequency and spatial patterns of the cosmic neutral hydrogen (HI) signal from foregrounds in the presence of noise. Cleaned maps recover cosmological clustering statistics within 10% at all relevant angular scales and frequencies. This amounts to a reduction in prediction variance of over an order of magnitude on small angular scales ($\ell > 300$), and improved accuracy for small radial scales ($k_{\parallel} > 0.17\ \rm h\ Mpc^{-1})$ compared to standard Principal Component Analysis (PCA) methods. We estimate posterior confidence intervals for the network's prediction by training an ensemble of UNets. Our approach demonstrates the feasibility of analyzing 21cm intensity maps, as opposed to derived summary statistics, for upcoming radio experiments, as long as the simulated foreground model is sufficiently realistic. We provide the code used for this analysis on Github https://github.com/tlmakinen/deep21 as well as a browser-based tutorial for the experiment and UNet model via the accompanying http://bit.ly/deep21-colab Colab notebook.

astro-ph.CO↗

arXiv2021

CARPool: fast, accurate computation of large-scale structure statistics by pairing costly and cheap cosmological simulations

To exploit the power of next-generation large-scale structure surveys, ensembles of numerical simulations are necessary to give accurate theoretical predictions of the statistics of observables. High-fidelity simulations come at a towering computational cost. Therefore, approximate but fast simulations, surrogates, are widely used to gain speed at the price of introducing model error. We propose a general method that exploits the correlation between simulations and surrogates to compute fast, reduced-variance statistics of large-scale structure observables without model error at the cost of only a few simulations. We call this approach Convergence Acceleration by Regression and Pooling (CARPool). In numerical experiments with intentionally minimal tuning, we apply CARPool to a handful of GADGET-III $N$-body simulations paired with surrogates computed using COmoving Lagrangian Acceleration (COLA). We find $\sim 100$-fold variance reduction even in the non-linear regime, up to $k_\mathrm{max} \approx 1.2$ $h {\rm Mpc^{-1}}$ for the matter power spectrum. CARPool realises similar improvements for the matter bispectrum. In the nearly linear regime CARPool attains far larger sample variance reductions. By comparing to the 15,000 simulations from the Quijote suite, we verify that the CARPool estimates are unbiased, as guaranteed by construction, even though the surrogate misses the simulation truth by up to $60\%$ at high $k$. Furthermore, even with a fully configuration-space statistic like the non-linear matter density probability density function, CARPool achieves unbiased variance reduction factors of up to $\sim 10$, without any further tuning. Conversely, CARPool can be used to remove model error from ensembles of fast surrogates by combining them with a few high-accuracy simulations.

astro-ph.CO↗

arXiv2021

The impact of massive neutrinos on halo assembly bias

Using the publicly available Quijote simulations, we present the first measurements of the assembly bias of dark matter halos in N-body simulations which include massive neutrinos. We focus on the dependence of the linear bias $b_1$ on three halo properties: 1) concentration $c$, 2) spin $λ$, and 3) ellipticity $s$. Although these simulations cover a large volume, superior to any future surveys, we do not detect any effect of neutrinos on the relations $b_1(c)$, $b_1(λ)$ and $b_1(s)$ at fixed halo mass. We further study the dependence of halo properties and environment on neutrinos, finding these quantities to be impacted by neutrino masses at the same level as assembly bias. We find that the effect of neutrinos on spin and shape can be largely attributed to the change in the cold dark matter $σ_8$ in neutrinos simulations, which is not the case for concentration.

astro-ph.CO↗

arXiv2021

Removing Astrophysics in 21 cm maps with Neural Networks

Measuring temperature fluctuations in the 21 cm signal from the Epoch of Reionization and the Cosmic Dawn is one of the most promising ways to study the Universe at high redshifts. Unfortunately, the 21 cm signal is affected by both cosmology and astrophysics processes in a non-trivial manner. We run a suite of 1,000 numerical simulations with different values of the main astrophysical parameters. From these simulations we produce tens of thousands of 21 cm maps at redshifts $10\leq z\leq 20$. We train a convolutional neural network to remove the effects of astrophysics from the 21 cm maps, and output maps of the underlying matter field. We show that our model is able to generate 2D matter fields that not only resemble the true ones visually, but whose statistical properties agree with the true ones within a few percent down to pretty small scales. We demonstrate that our neural network retains astrophysical information, that can be used to constrain the value of the astrophysical parameters. Finally, we use saliency maps to try to understand which features of the 21 cm maps is the network using in order to determine the value of the astrophysical parameters.

astro-ph.CO↗

arXiv2021

Information Content of Higher-Order Galaxy Correlation Functions

The shapes of galaxy N-point correlation functions can be used as standard rulers to constrain the distance-redshift relationship and thence the expansion rate of the Universe. The cosmological density fields traced by late-time galaxy formation are initially nearly Gaussian, and hence all the cosmological information can be extracted from their 2-Point Correlation Function (2PCF) or its Fourier-space analog the power spectrum. Subsequent nonlinear evolution under gravity, as well as halo and then galaxy formation, generate higher-order correlation functions. Since the mapping of the initial to the final density field is, on large scales, invertible, it is often claimed that the information content of the initial field's power spectrum is equal to that of all the higher-order functions of the final, nonlinear field. This claim implies that reconstruction of the initial density field from the nonlinear field renders analysis of higher-order correlation functions of the latter superfluous. We here show that this claim is false when the N-point functions are used as standard rulers. Constraints available from joint analysis of the galaxy power spectrum and bispectrum (Fourier-space analog of the 3-Point Correlation Function) can, in some cases, exceed those offered by the initial power spectrum even when the reconstruction is perfect. We provide a mathematical justification for this claim and also demonstrate it using a large suite of N-body simulations. In particular, we show that for the z = 0 real-space matter field in the limit of vanishing shot noise, taking modes up to k_max = 0.2 h/Mpc, using the bispectrum alone offers a factor of two reduction in the variance on the cosmic distance scale relative to that available from the power spectrum.

astro-ph.CO↗

arXiv2020

Learning the Evolution of the Universe in N-body Simulations

Understanding the physics of large cosmological surveys down to small (nonlinear) scales will significantly improve our knowledge of the Universe. Large N-body simulations have been built to obtain predictions in the non-linear regime. However, N-body simulations are computationally expensive and generate large amount of data, putting burdens on storage. These data are snapshots of the simulated Universe at different times, and fine sampling is necessary to accurately save its whole history. We employ a deep neural network model to predict the nonlinear N-body simulation at an intermediate time step given two widely separated snapshots. Our results outperform the cubic Hermite interpolation benchmark method in interpolating N-body simulations. This work can greatly reduce the storage requirement and allow us to reconstruct the cosmic history from far fewer snapshots of the universe.

astro-ph.CO↗

arXiv2020

Constraining $M_ν$ with the Bispectrum II: The Total Information Content of the Galaxy Bispectrum

Massive neutrinos suppress the growth of structure on small scales and leave an imprint on large-scale structure that can be measured to constrain their total mass, $M_ν$. With standard analyses of two-point clustering statistics, $M_ν$ constraints are severely limited by parameter degeneracies. Hahn et al.(2020) demonstrated that the bispectrum, the next higher-order statistic, can break these degeneracies and dramatically improve constraints on $M_ν$ and other cosmological parameters. In this paper, we present the constraining power of the redshift-space galaxy bispectrum, $B_0^g$. We construct the Molino suite of 75,000 mock galaxy catalogs from the Quijote $N$-body simulations using the halo occupation distribution (HOD) model, which provides a galaxy bias framework well-suited for simulation-based approaches. Using these mocks, we present Fisher matrix forecasts for $\{Ω_m,Ω_b,h,n_s,σ_8, M_ν\}$ and quantify, for the first time, the total information content of $B_0^g$ down to nonlinear scales. For $k_{\rm max}=0.5h/Mpc$, $B_0^g$ improves constraints on $Ω_m,Ω_b,h,n_s,σ_8$, and $M_ν$ by 2.8, 3.1, 3.8, 4.2, 4.2, and $4.6\times$ over the power spectrum, after marginalizing over HOD parameters. Even with priors from $Planck$, $B_0^g$ improves all of the cosmological constraints by $\gtrsim2\times$. In fact, for $P_\ell^g$ and $B_0^g$ out to $k_{\rm max}=0.5h/Mpc$ with $Planck$ priors, we achieve a $1σ$ $M_ν$ constraint of 0.048 eV, which is tighter than the current best cosmological constraint. While effects such as survey geometry and assembly bias will have an impact, these constraints are derived for $(1h^{-1}{\rm Gpc})^3$, a substantially smaller volume than upcoming surveys. Therefore, we conclude that the galaxy bispectrum will significantly improve cosmological constraints for upcoming galaxy surveys -- especially for $M_ν$.

astro-ph.CO↗

arXiv2020

dm2gal: Mapping Dark Matter to Galaxies with Neural Networks

Maps of cosmic structure produced by galaxy surveys are one of the key tools for answering fundamental questions about the Universe. Accurate theoretical predictions for these quantities are needed to maximize the scientific return of these programs. Simulating the Universe by including gravity and hydrodynamics is one of the most powerful techniques to accomplish this; unfortunately, these simulations are very expensive computationally. Alternatively, gravity-only simulations are cheaper, but do not predict the locations and properties of galaxies in the cosmic web. In this work, we use convolutional neural networks to paint galaxy stellar masses on top of the dark matter field generated by gravity-only simulations. Stellar mass of galaxies are important for galaxy selection in surveys and thus an important quantity that needs to be predicted. Our model outperforms the state-of-the-art benchmark model and allows the generation of fast and accurate models of the observed galaxy distribution.

astro-ph.CO↗

arXiv2020

Fast and Accurate Non-Linear Predictions of Universes with Deep Learning

Cosmologists aim to model the evolution of initially low amplitude Gaussian density fluctuations into the highly non-linear "cosmic web" of galaxies and clusters. They aim to compare simulations of this structure formation process with observations of large-scale structure traced by galaxies and infer the properties of the dark energy and dark matter that make up 95% of the universe. These ensembles of simulations of billions of galaxies are computationally demanding, so that more efficient approaches to tracing the non-linear growth of structure are needed. We build a V-Net based model that transforms fast linear predictions into fully nonlinear predictions from numerical simulations. Our NN model learns to emulate the simulations down to small scales and is both faster and more accurate than the current state-of-the-art approximate methods. It also achieves comparable accuracy when tested on universes of significantly different cosmological parameters from the one used in training. This suggests that our model generalizes well beyond our training set.

astro-ph.CO↗

arXiv2020

Modeling assembly bias with machine learning and symbolic regression

Upcoming 21cm surveys will map the spatial distribution of cosmic neutral hydrogen (HI) over unprecedented volumes. Mock catalogues are needed to fully exploit the potential of these surveys. Standard techniques employed to create these mock catalogs, like Halo Occupation Distribution (HOD), rely on assumptions such as the baryonic properties of dark matter halos only depend on their masses. In this work, we use the state-of-the-art magneto-hydrodynamic simulation IllustrisTNG to show that the HI content of halos exhibits a strong dependence on their local environment. We then use machine learning techniques to show that this effect can be 1) modeled by these algorithms and 2) parametrized in the form of novel analytic equations. We provide physical explanations for this environmental effect and show that ignoring it leads to underprediction of the real-space 21-cm power spectrum at $k\gtrsim 0.05$ h/Mpc by $\gtrsim$10\%, which is larger than the expected precision from upcoming surveys on such large scales. Our methodology of combining numerical simulations with machine learning techniques is general, and opens a new direction at modeling and parametrizing the complex physics of assembly bias needed to generate accurate mocks for galaxy and line intensity mapping surveys.

astro-ph.CO↗

arXiv2020

Neural networks as optimal estimators to marginalize over baryonic effects

Many different studies have shown that a wealth of cosmological information resides on small, non-linear scales. Unfortunately, there are two challenges to overcome to utilize that information. First, we do not know the optimal estimator that will allow us to retrieve the maximum information. Second, baryonic effects impact that regime significantly and in a poorly understood manner. Ideally, we would like to use an estimator that extracts the maximum cosmological information while marginalizing over baryonic effects. In this work we show that neural networks can achieve that. We made use of data where the maximum amount of cosmological information is known: power spectra and 2D Gaussian density fields. We also contaminate the data with simplified baryonic effects and train neural networks to predict the value of the cosmological parameters. For this data, we show that neural networks can 1) extract the maximum available cosmological information, 2) marginalize over baryonic effects, and 3) extract cosmological information that is buried in the regime dominated by baryonic physics. We also show that neural networks learn the priors of the data they are trained on. We conclude that a promising strategy to maximize the scientific return of cosmological experiments is to train neural networks on state-of-the-art numerical simulations with different strengths and implementations of baryonic effects.

astro-ph.CO↗

arXiv2020

Baryonic effects on the matter bispectrum

The large-scale clustering of matter is impacted by baryonic physics, particularly AGN feedback. Modelling or mitigating this impact will be essential for making full use of upcoming measurements of cosmic shear and other large-scale structure probes. We study baryonic effects on the matter bispectrum, using measurements from a selection of state-of-the-art hydrodynamical simulations: IllustrisTNG, Illustris, EAGLE, and BAHAMAS. We identify a low-redshift enhancement of the bispectrum, peaking at $k\sim 3h\,{\rm Mpc}^{-1}$, that is present in several simulations, and discuss how it can be associated to the evolving nature of AGN feedback at late times. This enhancement does not appear in the matter power spectrum, and therefore represents a new source of degeneracy breaking between two- and three-point statistics. In addition, we provide physical interpretations for other aspects of these measurements, and make initial comparisons to predictions from perturbation theory, empirical fitting formulas, and the response function formalism. We publicly release our measurements (including estimates of their uncertainty due to sample variance) and bispectrum measurement code as resources for the community.

astro-ph.CO↗

arXiv2020

Teaching neural networks to generate Fast Sunyaev Zel'dovich Maps

The thermal Sunyaev-Zel'dovich (tSZ) and the kinematic Sunyaev-Zel'dovich (kSZ) effects trace the distribution of electron pressure and momentum in the hot Universe. These observables depend on rich multi-scale physics, thus, simulated maps should ideally be based on calculations that capture baryonic feedback effects such as cooling, star formation, and other complex processes. In this paper, we train deep convolutional neural networks with a U-Net architecture to map from the three-dimensional distribution of dark matter to electron density, momentum and pressure at ~ 100 kpc resolution. These networks are trained on a combination of the TNG300 volume and a set of cluster zoom-in simulations from the IllustrisTNG project. The neural nets are able to reproduce the power spectrum, one-point probability distribution function, bispectrum, and cross-correlation coefficients of the simulations more accurately than the state-of-the-art semi-analytical models. Our approach offers a route to capture the richness of a full cosmological hydrodynamical simulation of galaxy formation with the speed of an analytical calculation.

astro-ph.CO↗

arXiv2020

The Effective Halo Model: Creating a Physical and Accurate Model of the Matter Power Spectrum and Cluster Counts

We introduce a physically-motivated model of the matter power spectrum, based on the halo model and perturbation theory. This model achieves 1\% accuracy on all $k-$scales between $k=0.02h\,\mathrm{Mpc}^{-1}$ to $k=1h\,\mathrm{Mpc}^{-1}$. Our key ansatz is that the number density of halos depends on the non-linear density contrast filtered on some unknown scale $R$. Using the Effective Field Theory of Large Scale Structure to evaluate the two-halo term, we obtain a model for the power spectrum with only two fitting parameters: $R$ and the effective `sound speed', which encapsulates small-scale physics. This is tested with two suites of cosmological simulations across a broad range of cosmologies and found to be highly accurate. Due to its physical motivation, the statistics can be easily extended beyond the power spectrum; we additionally derive the one-loop covariance matrices of cluster counts and their combination with the matter power spectrum. This yields a significantly better fit to simulations than previous models, and includes a new model for super-sample effects, which is rigorously tested with separate universe simulations. At low redshift, we find a significant ($\sim 10\%$) exclusion covariance from accounting for the finite size of halos which has not previously been modeled. Such power spectrum and covariance models will enable joint analysis of upcoming large-scale structure surveys, gravitational lensing surveys and cosmic microwave background maps on scales down to the non-linear scale. We provide a publicly released Python code.

astro-ph.CO↗