GalaxyNet: Connecting galaxies and dark matter haloes with deep neural networks and reinforcement learning in large volumes

Abstract

We present the novel wide & deep neural network GalaxyNet, which connects the properties of galaxies and dark matter haloes, and is directly trained on observed galaxy statistics using reinforcement learning. The most important halo properties to predict stellar mass and star formation rate (SFR) are halo mass, growth rate, and scale factor at the time the mass peaks, which results from a feature importance analysis with random forests. We train different models with supervised learning to find the optimal network architecture. GalaxyNet is then trained with a reinforcement learning approach: for a fixed set of weights and biases, we compute the galaxy properties for all haloes and then derive mock statistics (stellar mass functions, cosmic and specific SFRs, quenched fractions, and clustering). Comparing these statistics to observations we get the model loss, which is minimised with particle swarm optimisation. GalaxyNet reproduces the observed data very accurately (${\chi }_{\mathrm{red}}=1.05$), and predicts a stellar-to-halo mass relation with a lower normalisation and shallower low-mass slope at high redshift than empirical models. We find that at low mass, the galaxies with the highest SFRs are satellites, although most satellites are quenched. The normalisation of the instantaneous conversion efficiency increases with redshift, but stays constant above $z\gtrsim 0.7$. Finally, we use GalaxyNet to populate a cosmic volume of $(5.9\mathrm{Gpc}{)}^3$ with galaxies and predict the BAO signal, the bias, and the clustering of active and passive galaxies up to $z=4$, which can be tested with next-generation surveys, such as LSST and Euclid.

21 pages, 21 figures, 6 tables, submitted

References in corpus (13)

• The NumPy array: a structure for efficient numerical computation
• Introducing the Illustris Project: Simulating the coevolution of dark and visible matter in the Universe
• The clustering of galaxies in the completed SDSS-III Baryon Oscillation Spectroscopic Survey: Observational systematics and baryon acoustic oscillations in the correlation function
• Photometric redshift estimation via deep learning
• A robust morphological classification of high-redshift galaxies using support vector machines on seeing limited images. I Method description
• Physical properties underlying observed kinematics of satellite galaxies
• The Galah Survey: Classification and diagnostics with t-SNE reduction of spectral information
• Feature importance for machine learning redshifts applied to SDSS galaxies
• EMERGE: Empirical predictions of galaxy merger rates since $z\sim 6$
• Finding New High-Redshift Quasars by Asking the Neighbours
• Predicting fully self-consistent satellite richness, galaxy growth and starformation rates from the STastical sEmi-Empirical modeL STEEL
• Using Artificial Neural Networks to Constrain the Halo Baryon Fraction during Reionization
• Evidence for merging or disruption of red galaxies from the evolution of their clustering

Cited by in corpus (32)

• The CAMELS project: public data release
• A machine learning approach to mapping baryons onto dark matter haloes using the EAGLE and C-EAGLE simulations
• Lossless, Scalable Implicit Likelihood Inference for Cosmological Fields
• Robust Field-level Likelihood-free Inference with Galaxies
• Deep Realistic Extragalactic Model (DREaM) Galaxy Catalogs: Predictions for a Roman Ultra-Deep Field
• Machine Learning for Observational Cosmology
• Learning to Concentrate: Multi-tracer Forecasts on Local Primordial Non-Gaussianity with Machine-Learned Bias
• The scatter in the galaxy-halo connection: a machine learning analysis
• Augmenting astrophysical scaling relations with machine learning: application to reducing the Sunyaev-Zeldovich flux-mass scatter
• From EMBER to FIRE: predicting high resolution baryon fields from dark matter simulations with Deep Learning
• A Search for Satellite Galaxies of Nearby Star-Forming Galaxies with Resolved Stars in LBT-SONG
• Multi-Epoch Machine Learning 1: Unravelling Nature vs Nurture for Galaxy Formation
• Hybrid analytic and machine-learned baryonic property insertion into galactic dark matter haloes
• Modeling assembly bias with machine learning and symbolic regression
• How the Galaxy-Halo Connection Depends on Large-Scale Environment
• SHAMe-SF: Predicting the clustering of star-forming galaxies with an enhanced abundance matching model
• MAHGIC: A Model Adapter for the Halo-Galaxy Inter-Connection
• Deep learning insights into cosmological structure formation
• Classification algorithms applied to structure formation simulations
• How to empirically model star formation in dark matter halos: I. Inferences about central galaxies from numerical simulations
• Predictions on the stellar-to-halo mass relation in the dwarf regime using the empirical model for galaxy formation EMERGE
• Beyond Mass: Detecting Secondary Halo Properties with Galaxy-Galaxy Lensing
• A sparse regression approach to modeling the relation between galaxy stellar masses and their host halos
• Analysis of Dark Matter Halo Structure Formation in $N$-body Simulations with Machine Learning
• Anatomy of galactic star formation history: Roles of different modes of gas accretion, feedback, and recycling
• Multi-Epoch Machine Learning 2: Identifying physical drivers of galaxy properties in simulations
• Total and dark mass from observations of galaxy centers with Machine Learning
• Halo Properties from Observable Measures of Environment: I. Halo and Subhalo Masses
• A Conditional Abundance Matching Method of Extending Simulated Halo Merger Trees to Resolve Low-Mass Progenitors and Sub-halos
• A sparse regression approach for populating dark matter halos and subhalos with galaxies
• Insights into the dependence of galaxy properties on the environment with explainable machine learning models
• ARCHITECTS I: Impact of subgrid physics on the simulated properties of the circumgalactic medium