Machine learning and cosmological simulations – II. Hydrodynamical simulations

Abstract

We extend a machine learning (ML) framework presented previously to model galaxy formation and evolution in a hierarchical universe using N-body + hydrodynamical simulations. In this work, we show that ML is a promising technique to study galaxy formation in the backdrop of a hydrodynamical simulation. We use the Illustris Simulation to train and test various sophisticated machine learning algorithms. By using only essential dark matter halo physical properties and no merger history, our model predicts the gas mass, stellar mass, black hole mass, star formation rate, g − r color, and stellar metallicity fairly robustly. Our results provide a unique and powerful phenomenological framework to explore the galaxy-halo connection that is built upon a solid hydrodynamical simulation. The promising reproduction of the listed galaxy properties demonstrably place ML as a promising and a signicantly more computationally efficient tool to study small-scale structure formation. We find that ML mimics a full-blown hydrodynamical simulation surprisingly well in a computation time of mere minutes. The population of galaxies simulated by ML, while not numerically identical to Illustris, is statistically robust and physically consistent with Illustris galaxies and follows the same fundamental observational constraints. Machine learning offers an intriguing and promising technique to create quick mock galaxy catalogs in the future.

Painting galaxies into dark matter haloes using machine learning

“…Comparing to the Illustris-based ML study of Kamdar et al (2016), we find that our values of R 2 and ρ for M * and SFR agree quite well with their predictions, showing that this level of agreement is generally independent of the details of ML algorithm and training simulation. However, for gas mass and metallicity, our correlation measures are somewhat worse than their quoted values.…”

“…The approach of using machine learning to populate dark matter halos by training on hydrodynamic simulations has previously been examined by Kamdar et al (2016), who used the Illustris simulation (Vogelsberger et al 2014) as the training set. Our approach will be generally similar to theirs, and in areas of overlap our results quantitatively corroborate theirs.…”

A machine learning approach to mapping baryons on to dark matter haloes using the eagle and C-EAGLE simulations

“…They used dark matter properties from each halo as features, and baryonic properties as predictors, and trained the machine to learn the mapping between the two. They then followed this up by applying the same technique to the Illustris hydrodynamic simulation (Kamdar et al 2016b). Agarwal et al (2018) presented a similar model applied to the MUFASA simulation.…”

Learning the relationship between galaxies spectra and their star formation histories using convolutional neural networks and cosmological simulations

“…We would also like to thank Rita Tojeiro for help understanding Vespa. We wish to acknowledge the use of the following open source software packages not mentioned directly in the text: Scipy (Jones et al 2001) and Astropy (Astropy Collaboration et al 2013). CCL (ORCID 0000-0001-7964-5933) acknowledges the support of a PhD studentship from the UK Science and Technology Facilities Council (STFC).…”