91̽»¨

Abstract:

Cosmological N-body simulations of the dark matter component of the universe typically use initial conditions with a fixed power spectrum and random phases of the density field, leading to structure consistent with the local distribution of galaxies only in a statistical sense. It is, however, possible to infer the initial phases which lead to the configuration of galaxies and clusters that we see around us. We analyse the CSiBORG suite of 101 simulations, formed by constraining the density field within 155 Mpc h^−1 with dark matter particle mass 4.38 × 10⁹ M_⊙, to quantify the degree to which constraints imposed on 2.65 Mpc h^−1 scales reduce variance in the halo mass function and halo–halo cross-correlation function on a range of scales. This is achieved by contrasting CSiBORG with a subset of the unconstrained Quijote simulations and expectations for the ΛCDM average. Using the FOF, PHEW, and HOP halofinders, we show that the CSiBORG suite beats cosmic variance at large mass scales (≳10¹⁴ M_⊙ h^−1), which are most strongly constrained by the initial conditions, and exhibits a significant halo–halo cross-correlation out to ∼30 Mpc h^−1. Moreover, the effect of the constraints percolates down to lower mass objects and to scales below those on which they are imposed. Finally, we develop an algorithm to ‘twin’ haloes between realizations and show that approximately 50 per cent of haloes with mass greater than 10¹⁵ M_⊙ h^−1 can be identified in all realizations of the CSiBORG suite. We make the CSiBORG halo catalogues publicly available for future applications requiring knowledge of the local halo field.

Abstract:

We apply machine learning (ML), a powerful method for uncovering complex correlations in high-dimensional data, to the galaxy-halo connection of cosmological hydrodynamical simulations. The mapping between galaxy and halo variables is stochastic in the absence of perfect information, but conventional ML models are deterministic and hence cannot capture its intrinsic scatter. To overcome this limitation, we design an ensemble of neural networks with a Gaussian loss function that predict probability distributions, allowing us to model statistical uncertainties in the galaxy-halo connection as well as its best-fitting trends. We extract a number of galaxy and halo variables from the Horizon-AGN and IllustrisTNG100-1 simulations and quantify the extent to which knowledge of some subset of one enables prediction of the other. This allows us to identify the key features of the galaxy-halo connection and investigate the origin of its scatter in various projections. We find that while halo properties beyond mass account for up to 50 per cent of the scatter in the halo-To-stellar mass relation, the prediction of stellar half-mass radius or total gas mass is not substantially improved by adding further halo properties. We also use these results to investigate semi-Analytic models for galaxy size in the two simulations, finding that assumptions relating galaxy size to halo size or spin are not successful.