Abstract
Context. The galaxy total mass inside the effective radius is a proxy of the galaxy dark matter content and the star formation efficiency. As such, it encodes important information on the dark matter and baryonic physics.
Aims. Total central masses can be inferred via galaxy dynamics or gravitational lensing, but these methods have limitations. We propose a novel approach based on machine learning to make predictions on total and dark matter content using simple observables from imaging and spectroscopic surveys.
Methods. We used catalogs of multiband photometry, sizes, stellar mass, kinematic measurements (features), and dark matter (targets) of simulated galaxies from the Illustris-TNG100 hydrodynamical simulation to train a Mass Estimate machine Learning Algorithm (MELA) based on random forests.
Results. We separated the simulated sample into passive early-type galaxies (ETGs), both normal and dwarf, and active late-type galaxies (LTGs) and showed that the mass estimator can accurately predict the galaxy dark masses inside the effective radius in all samples. We finally tested the mass estimator against the central mass estimates of a series of low-redshift (z ≲ 0.1) datasets, including SPIDER, MaNGA/DynPop, and SAMI dwarf galaxies, derived with standard dynamical methods based on the Jeans equations. We find that MELA predictions are fully consistent with the total dynamical mass of the real samples of ETGs, LTGs, and dwarf galaxies.
Conclusions. MELA learns from hydro-simulations how to predict the dark and total mass content of galaxies, provided that the real galaxy samples overlap with the training sample or show similar scaling relations in the feature and target parameter space. In this case, dynamical masses are reproduced within 0.30 dex (∼2σ), with a limited fraction of outliers and almost no bias. This is independent of the sophistication of the kinematical data collected (fiber vs. 3D spectroscopy) and the dynamical analysis adopted (radial vs. axisymmetric Jeans equations, virial theorem). This makes MELA a powerful alternative to predict the mass of galaxies of massive stage IV survey datasets using basic data, such as aperture photometry, stellar masses, fiber spectroscopy, and sizes. We finally discuss how to generalize these results to account for the variance of cosmological parameters and baryon physics using a more extensive variety of simulations and the further option of reverse engineering this approach and using model-free dark matter measurements (e.g., via strong lensing), plus visual observables, to predict the cosmology and the galaxy formation model.
Cited by
1 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献