Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userPainting halos from cosmic density fields of dark matter with physically motivated neural networks
99
0
0.0
(
0
)
Added by
Doogesh Kodi Ramanah
Publication date
2019
fields
Physics
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
We present a novel halo painting network that learns to map approximate 3D dark matter fields to realistic halo distributions. This map is provided via a physically motivated network with which we can learn the non-trivial local relation between dark matter density field and halo distributions without relying on a physical model. Unlike other generative or regressive models, a well motivated prior and simple physical principles allow us to train the mapping network quickly and with relatively little data. In learning to paint halo distributions from computationally cheap, analytical and non-linear density fields, we bypass the need for full particle mesh simulations and halo finding algorithms. Furthermore, by design, our halo painting network needs only local patches of dark matter density to predict the halos, and as such, it can predict the 3D halo distribution for any arbitrary simulation box size. Our neural network can be trained using small simulations and used to predict large halo distributions, as long as the resolutions are equivalent. We evaluate our models ability to generate 3D halo count distributions which reproduce, to a high degree, summary statistics such as the power spectrum and bispectrum, of the input or reference realizations.
rate research
Read More
The simplest scheme for predicting real galaxy properties after performing a dark matter simulation is to rank order the real systems by stellar mass and the simulated systems by halo mass and then simply assume monotonicity - that the more massive halos host the more massive galaxies. This has had some success, but we study here if a better motivated and more accurate matching scheme is easily constructed by looking carefully at how well one could predict the simulated IllustrisTNG galaxy sample from its dark matter computations. We find that using the dark matter rotation curve peak velocity, $v_{max}$, for normal galaxies reduces the error of the prediction by 30% (18% for central galaxies and 60% for satellite systems) - following expectations from Faber-Jackson and the physics of monolithic collapse. For massive systems with halo mass $>$ 10$^{12.5}$ M$_{odot}$ hierarchical merger driven formation is the better model and dark matter halo mass remains the best single metric. Using a new single variable that combines these effects, $phi$ $=$ $v_{max}$/$v_{max,12.7}$ + M$_{peak}$/(10$^{12.7}$ M$_{odot}$) allows further improvement and reduces the error, as compared to ranking by dark matter mass at $z=0$ by another 6% from $v_{max}$ ranking. Two parameter fits -- including environmental effects produce only minimal further impact.
Cosmological simulations play an important role in the interpretation of astronomical data, in particular in comparing observed data to our theoretical expectations. However, to compare data with these simulations, the simulations in principle need to include gravity, magneto-hydrodyanmics, radiative transfer, etc. These ideal large-volume simulations (gravo-magneto-hydrodynamical) are incredibly computationally expensive which can cost tens of millions of CPU hours to run. In this paper, we propose a deep learning approach to map from the dark-matter-only simulation (computationally cheaper) to the galaxy distribution (from the much costlier cosmological simulation). The main challenge of this task is the high sparsity in the target galaxy distribution: space is mainly empty. We propose a cascade architecture composed of a classification filter followed by a regression procedure. We show that our result outperforms a state-of-the-art model used in the astronomical community, and provides a good trade-off between computational cost and prediction accuracy.
The multicomponent dark matter model with self-scattering and inter-
We develop a new method to reconstruct the cosmic density field from the distribution of dark matter haloes above a certain mass threshold. Our motivation is that well-defined samples of galaxy groups/clusters, which can be used to represent the dark halo population, can now be selected from large redshift surveys of galaxies, and our ultimate goal is to use such data to reconstruct the cosmic density field in the local universe. Our reconstruction method starts with a sample of dark matter haloes above a given mass threshold. Each volume element in space is assigned to the domain of the nearest halo according to a distance measure that is scaled by the virial radius of the halo. The distribution of the mass in and around dark matter haloes of a given mass is modelled using the cross-correlation function between dark matter haloes and the mass distribution within their domains. We use N-body cosmological simulations to show that the density profiles required in our reconstruction scheme can be determined reliably from large cosmological simulations, and that our method can reconstruct the density field accurately using haloes with masses down to $sim 10^{12}msun$ (above which samples of galaxy groups can be constructed from current large redshift surveys of galaxies). Working in redshift space, we demonstrate that the redshift distortions due to the peculiar velocities of haloes can be corrected in an iterative way. We also describe some applications of our method.
We perform an analysis of the three-dimensional cosmic matter density field traced by galaxies of the SDSS-III/BOSS galaxy sample. The systematic-free nature of this analysis is confirmed by two elements: the successful cross-correlation with the gravitational lensing observations derived from Planck 2018 data and the absence of bias at scales $k simeq 10^{-3}-10^{-2}h$ Mpc$^{-1}$ in the a posteriori power spectrum of recovered initial conditions. Our analysis builds upon our algorithm for Bayesian Origin Reconstruction from Galaxies (BORG) and uses a physical model of cosmic structure formation to infer physically meaningful cosmic structures and their corresponding dynamics from deep galaxy observations. Our approach accounts for redshift-space distortions and light-cone effects inherent to deep observations. We also apply detailed corrections to account for known and unknown foreground contaminations, selection effects and galaxy biases. We obtain maps of residual, so far unexplained, systematic effects in the spectroscopic data of SDSS-III/BOSS. Our results show that unbiased and physically plausible models of the cosmic large scale structure can be obtained from present and next-generation galaxy surveys.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
