A machine learning approach to galaxy properties: joint redshift-stellar mass probability distributions with Random Forest
S. Mucesh, W. G. Hartley, A. Palmese, O. Lahav, L. Whiteway, A. F. L. Bluck, A. Alarcon, A. Amon, K. Bechtol, G. M. Bernstein, A. Carnero Rosell, M. Carrasco Ki… 展开作者
S. Mucesh, W. G. Hartley, A. Palmese, O. Lahav, L. Whiteway, A. F. L. Bluck, A. Alarcon, A. Amon, K. Bechtol, G. M. Bernstein, A. Carnero Rosell, M. Carrasco Kind, A. Choi, K. Eckert, S. Everett, D. Gruen, R. A. Gruendl, I. Harrison, E. M. Huff, N. Kuropatkin, I. Sevilla-Noarbe, E. Sheldon, B. Yanny, M. Aguena, S. Allam, D. Bacon, E. Bertin, S. Bhargava, D. Brooks, J. Carretero, F. J. Castander, C. Conselice, M. Costanzi, M. Crocce, L. N. da Costa, M. E. S. Pereira, J. De Vicente, S. Desai, H. T. Diehl, A. Drlica-Wagner, A. E. Evrard, I. Ferrero, B. Flaugher, P. Fosalba, J. Frieman, J. García-Bellido, E. Gaztanaga, D. W. Gerdes, J. Gschwend, G. Gutierrez, S. R. Hinton, D. L. Hollowood, K. Honscheid, D. J. James, K. Kuehn, M. Lima, H. Lin, M. A. G. Maia, P. Melchior, F. Menanteau, R. Miquel, R. Morgan, F. Paz-Chinchón, A. A. Plazas, E. Sanchez, V. Scarpine, M. Schubnell, S. Serrano, M. Smith, E. Suchyta, G. Tarle, D. Thomas, C. To, T. N. Varga, R. D. Wilkinson
备注：
18 pages, 8 figures, Accepted by MNRAS
发表信息：
Monthly Notices of the Royal Astronomical Society, Volume 502, Issue 2, April 2021, Pages 2770-2786
We demonstrate that highly accurate joint redshift-stellar mass probability distribution functions (PDFs) can be obtained using the Random Forest (RF) machine learning (ML) algorithm, even with few photometric bands available. As an example, we use the Dark Energy Survey (DES), combined with the COSMOS2015 catalogue for redshifts and stellar masses. We build two ML models: one containing deep photometry in the $griz$ bands, and the second reflecting the photometric scatter present in the main DES survey, with carefully constructed representative training data in each case. We validate our joint PDFs for $10,699$ test galaxies by utilizing the copula probability integral transform and the Kendall distribution function, and their univariate counterparts to validate the marginals. Benchmarked against a basic set-up of the template-fitting code BAGPIPES, our ML-based method outperforms template fitting on all of our predefined performance metrics. In addition to accuracy, the RF is extremely fast, able to compute joint PDFs for a million galaxies in just under $6$ min with consumer computer hardware. Such speed enables PDFs to be derived in real time within analysis codes, solving potential storage issues. As part of this work we have developed GALPRO, a highly intuitive and efficient Python package to rapidly generate multivariate PDFs on-the-fly. GALPRO is documented and available for researchers to use in their cosmology and galaxy evolution studies.
We compare the low redshift (z ~ 0.1) Lyman-alpha forest from hydrodynamical simulations with data from the Cosmic Origin Spectrograph (COS). We find tension between the observed number of lines with b-parameters in the range 25-45 km/s and the predictions from simulations that incorporate either vigorous feedback from active galactic nuclei or that exclude feedback altogether. The gas in these simulations is, respectively, either too hot to contribute to the Lyman-alpha absorption or too cold to produce the required line widths. Matching the observed b-parameter distribution therefore require…
Under the assumption that galaxies and Quasi-Stellar Objects (QSOs) dominate the metagalactic ultra-violet (UV) background, it is shown that at high redshifts fluctuations in the UV background are dominated by QSO shot noise and have an auto-correlation length of a few to several comoving Mpcs, depending on the bright end of the QSO luminosity function. The correlations create long range spatial coherence in the neutral hydrogen fraction. Using a semi-analytic model, it is demonstrated that the coherence may account for the broad distribution in effective optical depths measured in the Ly$α$ f…
We present a 30h integration Very Large Telescope X-shooter spectrum of the Lyman series forest towards the $z = 7.084$ quasar ULAS J1120+0641. The only detected transmission at ${\rm S/N}>5$ is confined to seven narrow spikes in the Ly$α$ forest, over the redshift range $5.858<z<6.122$, just longward of the wavelength of the onset of the Ly$β$ forest. There is also a possible detection of one further unresolved spike in the Ly$β$ forest at $z=6.854$, with ${\rm S/N}=4.5$. We also present revised Hubble Space Telescope F814W photometry of the source. The summed flux from the transmission spike…
We study the effect of different feedback prescriptions on the properties of the low redshift ($z\leq1.6$) Ly$α$ forest using a selection of hydrodynamical simulations drawn from the Sherwood simulation suite. The simulations incorporate stellar feedback, AGN feedback and a simplified scheme for efficiently modelling the low column density Ly$α$ forest. We confirm a discrepancy remains between Cosmic Origins Spectrograph (COS) observations of the Ly$α$ forest column density distribution function (CDDF) at $z \simeq 0.1$ for high column density systems ($N_{\rm HI}>10^{14}\rm\,cm^{-2}$), as wel…
A long (110 cMpc/$h$) and deep absorption trough in the Ly$α$ forest has been observed extending down to redshift 5.5 in the spectrum of ULAS J0148+0600. Although no Ly$α$ transmission is detected, Ly$β$ spikes are present which has led to claims that the gas along this trough must be ionized. Using high resolution cosmological radiative transfer simulations in large volumes, we show that in a scenario where reionization ends late ($z \sim 5.2$), our simulations can reproduce troughs as long as observed. In this model, we find that the troughs are caused by islands of neutral hydrogen. Small i…
We present an analysis of the evolution of the Lyman-series forest into the epoch of reionization using cosmological radiative transfer simulations in a scenario where reionization ends late. We explore models with different midpoints of reionization and gas temperatures. We find that once the simulations have been calibrated to match the mean flux of the observed Lyman-$α$ forest at $4 < z < 6$, they also naturally reproduce the distribution of effective optical depths of the Lyman-$β$ forest in this redshift range. We note that the tail of the largest optical depths that is most challenging …
Active galactic nuclei (AGN) feedback models are generally calibrated to reproduce galaxy observables such as the stellar mass function and the bimodality in galaxy colors. We use variations of the AGN feedback implementations in the IllustrisTNG (TNG) and Simba cosmological hydrodynamic simulations to show that the low redshift Lyman-$α$ forest can provide constraints on the impact of AGN feedback. We show that TNG over-predicts the number density of absorbers at column densities $N_{\rm HI} < 10^{14}$ cm$^{-2}$ compared to data from the Cosmic Origins Spectrograph (in agreement with previous…
Blazars are some of the brightest UV and X-ray sources in the sky and are valuable probes of the elusive warm-hot intergalactic medium (WHIM; $T{\simeq} 10^5-10^7$ K). However, many of the brightest blazars$-$called BL Lac objects such as 1ES1553+113$-$have quasi-featureless spectra and poorly constrained redshifts. Here, we significantly improve the precision of indirect redshift constraints for blazars based on the edge of the $\rm{H\,I}$ Ly$α$ forest observed in their UV spectra. We develop a robust technique to constrain the redshift of a $z<0.5$ AGN or blazar with a $1σ$ uncertainty of ${…
We present a method of selecting quasars up to redshift $\approx$ 6 with random forests, a supervised machine learning method, applied to Pan-STARRS1 and WISE data. We find that, thanks to the increasing set of known quasars we can assemble a training set that enables supervised machine learning algorithms to become a competitive alternative to other methods up to this redshift. We present a candidate set for the redshift range 4.8 to 6.3 which includes the region around z = 5.5 where quasars are difficult to select due to photometric similarity to red and brown dwarfs. We demonstrate that und…
We investigate the effects of AGN heating and the ultraviolet background on the low-redshift Lyman-$α$ forest column density distribution (CDD) using the Illustris simulation. We show that Illustris reproduces observations at $z =0.1$ in the column density range $10^{12.5} - 10^{13.5}$cm$^{-2}$, relevant for the "photon underproduction crisis." We attribute this to the inclusion of AGN feedback, which changes the gas distribution so as to mimic the effect of extra photons, as well as the use of the Faucher-Giguère ultraviolet background, which is more ionizing at $z=0.1$ than the Haardt & Mada…
Recent observations of the Lyman-alpha forest show large-scale spatial variations in the intergalactic Lyman-alpha opacity that grow rapidly with redshift at z>5, far in excess of expectations from empirically motivated models. Previous studies have attempted to explain this excess with spatial fluctuations in the ionizing background, but found that this required either extremely rare sources or problematically low values for the mean free path of ionizing photons. Here we report that much -- or potentially all -- of the observed excess likely arises from residual spatial variations in tempera…
We study the sensitivity of the $z=0.1$ Lyman-$α$ Forest observables, such as the column density distribution function (CDD), flux PDF, flux power spectrum, and line width distribution, to sub-grid models of active galactic nuclei (AGN) feedback using the Illustris and IllustrisTNG (TNG) cosmological simulations. The two simulations share an identical Ultraviolet Background (UVB) prescription and similar cosmological parameters, but TNG features an entirely reworked AGN feedback model. Due to changes in the AGN radio mode model, the original Illustris simulations have a factor of 2-3 fewer Lym…
A late end to reionisation at redshift $z\simeq 5.3$ is consistent with observed spatial variations in the Ly$α$ forest transmission and the deficit of Ly$α$ emitting galaxies around extended Ly$α$ absorption troughs at $z=5.5$. In this model, large islands of neutral hydrogen should persist in the diffuse intergalactic medium (IGM) until $z\simeq 6$. We use a novel, hybrid approach that combines high resolution cosmological hydrodynamical simulations with radiative transfer to predict the incidence of strong 21 cm forest absorbers with optical depths $τ_{21}>10^{-2}$ from the diffuse IGM in t…
We examine the statistics of the low-redshift Ly-alpha forest in an adaptive mesh refinement hydrodynamic cosmological simulation of sufficient volume to include distinct large-scale environments. We compare our HI column density distribution of absorbers both with recent work and between two highly-refined regions of our simulation: a large-scale overdensity and a large-scale underdensity (on scales of approximately 20 Mpc). We recover the average results presented in Kollmeier et al. (2014) using different simulation methods. We further break down these results as a function of environment t…
We measure the effective optical depth in the Lyman-alpha (Ly$α$) forest using 40,035 quasar spectra from the Twelfth Data Release (DR12) of the Baryon Oscillation Spectroscopic Survey (BOSS) of SDSS-IV. A rigorous selection based on spectral index and the equivalent width of the C IV emission line is applied to choose seven uniform samples with minimal intrinsic variations across redshifts. Modeling the redshift evolution of the effective optical depth with a power-law, $τ_\mathrm{eff}=τ_0 (1 + z)^γ$ produces $τ_0=(5.54 \pm 0.64)\times 10^{-3}$ and $γ=3.182 \pm 0.074$. The 2.2% precision esti…
Lyman-Alpha Forest and its Cross-Correlation with High-Redshift Galaxies in Effective Field Theory at the Field Level
Center for Theoretical Physics – a Leinweber Institute, Massachusetts Institute of Technology（麻省理工学院理论物理中心）；The NSF AI Institute for Artificial Intelligence and Fundamental Interactions（美国国家科学基金会人工智能基础相互作用研究所）；Department of Physics, Columbia University（哥伦比亚大学物理系）；NASA Hubble Fellowship Program（NASA 哈勃奖学金计划）；Theory Center, Institute of Particle and Nuclear Studies, High Energy Accelerator Research Organization (KEK)（高能加速器研究机构粒子与核研究学院理论中心）
We examine the column density distribution function (CDDF) and Doppler parameter distribution from hydrodynamical simulations and Cosmic Origins Spectrograph (COS) observations of the Lyman-alpha forest at redshift $0\leq z\leq 0.2$. Allowing for a factor of two uncertainty in the metagalactic HI photoionisation rate, our hydrodynamical simulations are in good agreement ($1$-$1.5σ$) with the shape and amplitude of the observed CDDF at HI column densities $10^{13.3}\rm\,cm^{-2}\leq N_{\rm HI}\leq 10^{14.5}\rm\,cm^{-2}$. However, the Doppler widths of the simulated lines remain too narrow with r…
Measuring distances of cosmological sources such as galaxies, stars and quasars plays an increasingly critical role in modern cosmology. Obtaining the optical spectrum and consequently calculating the redshift as a distance indicator could instantly classify these objects. As long as spectroscopic observations are not available for many galaxies and the process of measuring the redshift is time-consuming and infeasible for large samples, machine learning (ML) approaches could be applied to determine the redshifts of galaxies from different features including their photometric colors. In this p…
Several recent works have focused on the search for bright, high-z quasars (QSOs) in the South. Among them, the QUasars as BRIght beacons for Cosmology in the Southern hemisphere (QUBRICS) survey has now delivered hundreds of new spectroscopically confirmed QSOs selected by means of machine learning algorithms. Building upon the results obtained by introducing the probabilistic random forest (PRF) for the QUBRICS selection, we explore in this work the feasibility of training the algorithm on synthetic data to improve the completeness in the higher redshift bins. We also compare the performance…
Studies of low redshift galaxy clusters suggest the intra-cluster medium (ICM) has experienced non-gravitational heating during the formation phase of the clusters. Using simple phenomenological heating prescriptions, we simulate the effect of this preheating of the nascent ICM in galaxy proto-clusters and examine its effect on Lyman-$α$ (Ly$α$) forest tomographic maps. We analyse a series of cosmological zoom-in simulations of proto-clusters within the framework of the Ly$α$ transmission-dark matter (DM) density distribution. We find that the more energy is injected into the proto-ICM at $z$ …
Ringing of the Reionization: A first direct measurement of the intergalactic pressure smoothing scale at redshift z>4.2 as imprinted onto small-scale peculiar velocities in the Lyman-alpha forest
University of Hertfordshire（赫特福德大学）；Kavli Institute for Cosmology, University of Cambridge（剑桥大学卡弗里宇宙学研究所）；SISSA- International School for Advanced Studies（国际高等研究学院）；INFN – National Institute for Nuclear Physics（意大利国家核物理研究所）；IFPU, Institute for Fundamental Physics of the Universe（宇宙基础物理研究所）；INAF, Osservatorio Astronomico di Trieste（的里雅斯特天文台）；ICSC - Centro Nazionale di Ricerca in High Performance Computing, Big Data e Quantum Computing（高性能计算、大数据与量子计算国家研究中心）；Institute of Astronomy, University of Cambridge（剑桥大学天文学研究所）
We explore the use of Deep Learning to infer the temperature of the intergalactic medium from the transmitted flux in the high redshift Lyman-alpha forest. We train Neural Networks on sets of simulated spectra from redshift z=2-3 outputs of cosmological hydrodynamic simulations, including high temperature regions added in post-processing to approximate bubbles heated by Helium-II reionization. We evaluate how well the trained networks are able to reconstruct the temperature from the effect of Doppler broadening in the simulated input Lyman-alpha forest absorption spectra. We find that for spec…
Cross-correlations between the CLAMATO Lyman-alpha forest and galaxies within the COSMOS field
University of Southern California（南加州大学）；The University of Tokyo（东京大学）；Lawrence Berkeley National Laboratory（劳伦斯伯克利国家实验室）；The Ohio State University（俄亥俄州立大学）；Institut de Física d’Altes Energies (IFAE)（高能物理研究所）；The Barcelona Institute of Science and Technology（巴塞罗那科学技术研究院）；Institució Catalana de Recerca i Estudis Avançats（加泰罗尼亚研究与高级研究学院）；Carnegie Observatories（卡内基观测站）
We present the first observations of foreground Lyman-$α$ forest absorption from high-redshift galaxies, targeting 24 star-forming galaxies (SFGs) with $z\sim 2.3-2.8$ within a $5' \times 15'$ region of the COSMOS field. The transverse sightline separation is $\sim 2\,h^{-1}\mathrm{Mpc}$ comoving, allowing us to create a tomographic reconstruction of the 3D Ly$α$ forest absorption field over the redshift range $2.20\leq z\leq 2.45$. The resulting map covers $6\,h^{-1}\mathrm{Mpc} \times 14\,h^{-1}\mathrm{Mpc}$ in the transverse plane and $230\,h^{-1}\mathrm{Mpc}$ along the line-of-sight with a…
Identifying merging galaxies is an important - but difficult - step in galaxy evolution studies. We present random forest classifications of galaxy mergers from simulated JWST images based on various standard morphological parameters. We describe (a) constructing the simulated images from IllustrisTNG and the Santa Cruz SAM, and modifying them to mimic future CEERS observations as well as nearly noiseless observations, (b) measuring morphological parameters from these images, and (c) constructing and training the random forests using the merger history information for the simulated galaxies av…
