Learning the Relationship between Galaxies Spectra and their Star Formation Histories using Convolutional Neural Networks and Cosmological Simulations

Abstract

We present a new method for inferring galaxy star formation histories (SFH) using machine learning methods coupled with two cosmological hydrodynamic simulations. We train Convolutional Neural Networks to learn the relationship between synthetic galaxy spectra and high resolution SFHs from the EAGLE and Illustris models. To evaluate our SFH reconstruction we use Symmetric Mean Absolute Percentage Error (SMAPE), which acts as a true percentage error in the low-error regime. On dust-attenuated spectra we achieve high test accuracy (median SMAPE $=10.5\%$). Including the effects of simulated observational noise increases the error ($12.5\%$), however this is alleviated by including multiple realisations of the noise, which increases the training set size and reduces overfitting ($10.9\%$). We also make estimates for the observational and modelling errors. To further evaluate the generalisation properties we apply models trained on one simulation to spectra from the other, which leads to only a small increase in the error (median SMAPE $\sim 15\%$). We apply each trained model to SDSS DR7 spectra, and find smoother histories than in the VESPA catalogue. This new approach complements the results of existing SED fitting techniques, providing star formation histories directly motivated by the results of the latest cosmological simulations.

20 pages, 21 figures, accepted for publication in Monthly Notices of the Royal Astronomical Society (MNRAS)

References in corpus (29)

• Multi-messenger Observations of a Binary Neutron Star Merger
• The EAGLE project: Simulating the evolution and assembly of galaxies and their environments
• Introducing the Illustris Project: Simulating the coevolution of dark and visible matter in the Universe
• The propagation of uncertainties in stellar population synthesis modeling I: The relevance of uncertain aspects of stellar evolution and the IMF to the derived physical properties of galaxies
• Overview of the SDSS-IV MaNGA Survey: Mapping Nearby Galaxies at Apache Point Observatory
• The EAGLE simulations of galaxy formation: calibration of subgrid physics and model variations
• Introducing the Illustris Project: the evolution of galaxy populations across cosmic time
• How to Measure Galaxy Star Formation Histories II: Nonparametric Models
• The Universal Relation of Galactic Chemical Evolution: The Origin of the Mass-Metallicity Relation
• How to measure galaxy star-formation histories I: Parametric models
• Recovering galaxy star formation and metallicity histories from spectra using VESPA
• The star formation main sequence and stellar mass assembly of galaxies in the Illustris simulation
• Non-parametric Star Formation History Reconstruction with Gaussian Processes I: Counting Major Episodes of Star Formation
• Galaxy And Mass Assembly: Deconstructing Bimodality - I. Red ones and blue ones
• Synthetic Galaxy Images and Spectra from the Illustris Simulation
• Machine Learning in Astronomy: a practical overview
• On the SFR-M${}_{\ast }$ main sequence archetypal star-formation history and analytical models
• Far-infrared and dust properties of present-day galaxies in the EAGLE simulations
• The 1-1000 micron SEDs of far-infrared galaxies
• SpectRes: A Fast Spectral Resampling Tool in Python
• On the importance of using appropriate spectral models to derive physical properties of galaxies at 0.7<z<2.8
• The spatially-resolved star formation histories of CALIFA galaxies: Implications for galaxy formation
• Black hole mass estimation for Active Galactic Nuclei from a new angle
• The EAGLE simulations of galaxy formation: Public release of particle data
• A Selective Overview of Deep Learning
• Galaxy formation in the PLANCK cosmology - III. The high-redshift universe
• Parametrising Star Formation Histories
• Simultaneous Estimation of Photometric Redshifts and SED Parameters: Improved Techniques and a Realistic Error Budget
• Comparison of Observed Galaxy Properties with Semianalytic Model Predictions using Machine Learning

Cited by in corpus (30)

• The VANDELS survey: the star-formation histories of massive quiescent galaxies at 1.0 < z < 1.3
• How Well Can We Measure the Stellar Mass of a Galaxy: The Impact of the Assumed Star Formation History Model in SED Fitting
• The DAWES review 10: The impact of deep learning for the analysis of galaxy surveys
• The CAMELS project: public data release
• Predicting star formation properties of galaxies using deep learning
• {\sc mirkwood:} Fast and Accurate SED Modeling Using Machine Learning
• COSMOS2020: Manifold Learning to Estimate Physical Parameters in Large Galaxy Surveys
• Surrogate modelling the Baryonic Universe I: The colour of star formation
• Stellar Populations With Optical Spectra: Deep Learning vs. Popular Spectrum Fitting Codes
• AstraeusIV : Quantifying the star formation histories of galaxies in the Epoch of Reionization
• Galaxy classification: deep learning on the OTELO and COSMOS databases
• MillimeterDL: Deep Learning Simulations of the Microwave Sky
• Multiwavelength cluster mass estimates and machine learning
• Deriving the star formation histories of galaxies from spectra with simulation-based inference
• DIGS: Deep Inference of Galaxy Spectra with Neural Posterior Estimation
• Constraining stellar population parameters from narrow band photometric surveys using convolutional neural networks
• ERGO-ML: Towards a robust machine learning model for inferring the fraction of accreted stars in galaxies from integral-field spectroscopic maps
• Starduster: A multi-wavelength SED model based on radiative transfer simulations and deep learning
• The miniJPAS survey quasar selection II: Machine learning classification with photometric measurements and uncertainties
• De-noising of galaxy optical spectra with autoencoders
• Lightning: An X-ray to Submillimeter Galaxy SED-Fitting Code With Physically-Motivated Stellar, Dust, and AGN Models
• Learning the Universe: Cosmological and Astrophysical Parameter Inference with Galaxy Luminosity Functions and Colours
• Euclid preparation. LXVIII. Extracting physical parameters from galaxies with machine learning
• pop-cosmos: Insights from generative modeling of a deep, infrared-selected galaxy population
• Euclid preparation. Spatially resolved stellar populations of local galaxies with Euclid: a proof of concept using synthetic images with the TNG50 simulation
• Harvesting the Lyα forest with convolutional neural networks
• Sengi: a small, fast, interactive viewer for spectral outputs from stellar population synthesis models
• Predicting far-infrared maps of galaxies via machine learning techniques
• Flexible Simulation Based Inference for Galaxy Photometric Fitting with Synthesizer
• Debunking Generalization Error or: How I Learned to Stop Worrying and Love My Training Set