skip to main content

Title: An observationally driven multifield approach for probing the circum-galactic medium with convolutional neural networks

The circum-galactic medium (CGM) can feasibly be mapped by multiwavelength surveys covering broad swaths of the sky. With multiple large data sets becoming available in the near future, we develop a likelihood-free Deep Learning technique using convolutional neural networks (CNNs) to infer broad-scale physical properties of a galaxy’s CGM and its halo mass for the first time. Using CAMELS (Cosmology and Astrophysics with MachinE Learning Simulations) data, including IllustrisTNG, SIMBA, and Astrid models, we train CNNs on Soft X-ray and 21-cm (H i) radio two-dimensional maps to trace hot and cool gas, respectively, around galaxies, groups, and clusters. Our CNNs offer the unique ability to train and test on ‘multifield’ data sets comprised of both H i and X-ray maps, providing complementary information about physical CGM properties and improved inferences. Applying eRASS:4 survey limits shows that X-ray is not powerful enough to infer individual haloes with masses log (Mhalo/M⊙) < 12.5. The multifield improves the inference for all halo masses. Generally, the CNN trained and tested on Astrid (SIMBA) can most (least) accurately infer CGM properties. Cross-simulation analysis – training on one galaxy formation model and testing on another – highlights the challenges of developing CNNs trained on a single model to marginalize over astrophysical uncertainties and perform robust inferences on real data. The next crucial step in improving the resulting inferences on the physical properties of CGM depends on our ability to interpret these deep-learning models.

more » « less
Author(s) / Creator(s):
; ; ; ;
Publisher / Repository:
Oxford University Press
Date Published:
Journal Name:
Monthly Notices of the Royal Astronomical Society
Medium: X Size: p. 10038-10058
["p. 10038-10058"]
Sponsoring Org:
National Science Foundation
More Like this
  1. Abstract

    We present CAMELS-ASTRID, the third suite of hydrodynamical simulations in the Cosmology and Astrophysics with MachinE Learning (CAMELS) project, along with new simulation sets that extend the model parameter space based on the previous frameworks of CAMELS-TNG and CAMELS-SIMBA, to provide broader training sets and testing grounds for machine-learning algorithms designed for cosmological studies. CAMELS-ASTRID employs the galaxy formation model following the ASTRID simulation and contains 2124 hydrodynamic simulation runs that vary three cosmological parameters (Ωm,σ8, Ωb) and four parameters controlling stellar and active galactic nucleus (AGN) feedback. Compared to the existing TNG and SIMBA simulation suites in CAMELS, the fiducial model of ASTRID features the mildest AGN feedback and predicts the least baryonic effect on the matter power spectrum. The training set of ASTRID covers a broader variation in the galaxy populations and the baryonic impact on the matter power spectrum compared to its TNG and SIMBA counterparts, which can make machine-learning models trained on the ASTRID suite exhibit better extrapolation performance when tested on other hydrodynamic simulation sets. We also introduce extension simulation sets in CAMELS that widely explore 28 parameters in the TNG and SIMBA models, demonstrating the enormity of the overall galaxy formation model parameter space and the complex nonlinear interplay between cosmology and astrophysical processes. With the new simulation suites, we show that building robust machine-learning models favors training and testing on the largest possible diversity of galaxy formation models. We also demonstrate that it is possible to train accurate neural networks to infer cosmological parameters using the high-dimensional TNG-SB28 simulation set.

    more » « less

    We present a new suite of over 1500 cosmological N-body simulations with varied warm dark matter (WDM) models ranging from 2.5 to 30 keV. We use these simulations to train Convolutional Neural Networks (CNNs) to infer WDM particle masses from images of DM field data. Our fiducial setup can make accurate predictions of the WDM particle mass up to 7.5 keV with an uncertainty of ±0.5 keV at a 95 per cent confidence level from (25 h−1Mpc)2 maps. We vary the image resolution, simulation resolution, redshift, and cosmology of our fiducial setup to better understand how our model is making predictions. Using these variations, we find that our models are most dependent on simulation resolution, minimally dependent on image resolution, not systematically dependent on redshift, and robust to varied cosmologies. We also find that an important feature to distinguish between WDM models is present with a linear size between 100 and 200 h−1 kpc. We compare our fiducial model to one trained on the power spectrum alone and find that our field-level model can make two times more precise predictions and can make accurate predictions to two times as massive WDM particle masses when used on the same data. Overall, we find that the field-level data can be used to accurately differentiate between WDM models and contain more information than is captured by the power spectrum. This technique can be extended to more complex DM models and opens up new opportunities to explore alternative DM models in a cosmological environment.

    more » « less
  3. Abstract We train graph neural networks to perform field-level likelihood-free inference using galaxy catalogs from state-of-the-art hydrodynamic simulations of the CAMELS project. Our models are rotational, translational, and permutation invariant and do not impose any cut on scale. From galaxy catalogs that only contain 3D positions and radial velocities of ∼1000 galaxies in tiny ( 25 h − 1 Mpc ) 3 volumes our models can infer the value of Ω m with approximately 12% precision. More importantly, by testing the models on galaxy catalogs from thousands of hydrodynamic simulations, each having a different efficiency of supernova and active galactic nucleus feedback, run with five different codes and subgrid models—IllustrisTNG, SIMBA, Astrid, Magneticum, SWIFT-EAGLE—we find that our models are robust to changes in astrophysics, subgrid physics, and subhalo/galaxy finder. Furthermore, we test our models on 1024 simulations that cover a vast region in parameter space—variations in five cosmological and 23 astrophysical parameters—finding that the model extrapolates really well. Our results indicate that the key to building a robust model is the use of both galaxy positions and velocities, suggesting that the network has likely learned an underlying physical relation that does not depend on galaxy formation and is valid on scales larger than ∼10 h −1 kpc. 
    more » « less
  4. Abstract We train graph neural networks on halo catalogs from Gadget N -body simulations to perform field-level likelihood-free inference of cosmological parameters. The catalogs contain ≲5000 halos with masses ≳10 10 h −1 M ⊙ in a periodic volume of ( 25 h − 1 Mpc ) 3 ; every halo in the catalog is characterized by several properties such as position, mass, velocity, concentration, and maximum circular velocity. Our models, built to be permutationally, translationally, and rotationally invariant, do not impose a minimum scale on which to extract information and are able to infer the values of Ω m and σ 8 with a mean relative error of ∼6%, when using positions plus velocities and positions plus masses, respectively. More importantly, we find that our models are very robust: they can infer the value of Ω m and σ 8 when tested using halo catalogs from thousands of N -body simulations run with five different N -body codes: Abacus, CUBEP 3 M, Enzo, PKDGrav3, and Ramses. Surprisingly, the model trained to infer Ω m also works when tested on thousands of state-of-the-art CAMELS hydrodynamic simulations run with four different codes and subgrid physics implementations. Using halo properties such as concentration and maximum circular velocity allow our models to extract more information, at the expense of breaking the robustness of the models. This may happen because the different N -body codes are not converged on the relevant scales corresponding to these parameters. 
    more » « less

    The hot component of the circumgalactic medium (CGM) around star-forming galaxies is detected as diffuse X-ray emission. The X-ray spectra from the CGM depend on the temperature and metallicity of the emitting plasma, providing important information about the feeding and feedback of the galaxy. The observed spectra are commonly fitted using simple one-temperature (1-T) or two-temperature (2-T) models. However, the actual temperature distribution of the gas can be complex because of the interaction between galactic outflows and halo gas. Here, we demonstrate this by analysing 3D hydrodynamical simulations of the CGM with a realistic outflow model. We investigate the physical properties of the simulated hot CGM, which shows a broad distribution in density, temperature, and metallicity. By constructing and fitting the simulated spectra, we show that, while the 1-T and 2-T models are able to fit the synthesized spectra reasonably well, the inferred temperature(s) does not bear much physical meaning. Instead, we propose a lognormal distribution as a more physical model. The lognormal model better fits the simulated spectra while reproducing the gas temperature distribution. We also show that when the star formation rate is high, the spectra inside the biconical outflows are distinct from those outside, as outflows are generally hotter and more metal enriched. Finally, we produce mock spectra for future missions with the eV-level spectral resolution, such as Athena, Lynx, the Hot Universe Baryon Surveyor, and theX-ray Imaging and Spectroscopy Mission.

    more » « less