Catalog-free modeling of galaxy types in deep images: Massive dimensional reduction with neural networks

Abstract

Current models of galaxy evolution are constrained by the analysis of catalogs containing the flux and size of galaxies extracted from multiband deep fields carrying inevitable observational and extraction-related biases which can be highly correlated. In practice, taking all of these effects simultaneously into account is difficult, and derived models are inevitably biased. To address this issue, we use robust likelihood-free methods for the inference of luminosity function parameters, made possible via massive compression of multiband images using artificial neural networks. This technique makes the use of catalogs unnecessary when comparing observed and simulated multiband deep fields and constraining model parameters. A forward modeling approach generates galaxies of multiple types depending on luminosity function parameters and paints them on photometric multiband deep fields including both the instrumental and observational characteristics. The simulated and the observed images present the same selection effects and can therefore be properly compared. We train a fully-convolutional neural network to extract the most model-parameter-sensitive summary statistics out of these realistic simulations, shrinking down the dimensionality of the summary space. Finally, using the trained network to compress both observed and simulated deep fields, the model parameter values are constrained through Population Monte Carlo likelihood-free inference. Using synthetic photometric multiband deep fields similar to the CFHTLS and D1/D2 deep fields and massively compressing them through the convolutional neural network, we demonstrate the robustness, accuracy and consistency of this new catalog-free inference method. We are able to constrain the parameters of luminosity functions of different types of galaxies and our results are fully compatible with the classic catalog extraction approaches.

arXiv comment23 pages, 20 figures

References in corpus (38)

• The Seventh Data Release of the Sloan Digital Sky Survey
• High Redshift Galaxies in the Hubble Deep Field. Color Selection and Star Formation History to z=4
• Galaxy Luminosity Functions to z~1: DEEP2 vs. COMBO-17 and Implications for Red Galaxy Formation
• Deep Big Simple Neural Nets Excel on Handwritten Digit Recognition
• An Analysis of the Shapes of Interstellar Extinction Curves. V. The IR-Through-UV Curve Morphology
• The size evolution of galaxies since z~3: combining SDSS, GEMS and FIRES
• The CANADA-FRANCE REDSHIFT SURVEY VI: Evolution of the galaxy luminosity function to z ~ 1
• The Evolving Luminosity Function of Red Galaxies
• The overdensities of galaxy environments as a function of luminosity and color
• Massive Lossless Data Compression and Multiple Parameter Estimation from Galaxy Spectra
• The evolving relations between size, mass, surface density, and star formation in 3x10^4 galaxies since z=2
• Stellar populations in the Milky Way bulge region : Towards solving the Galactic bulge and bar shapes using 2MASS data
• Modelling the spectral energy distribution of galaxies. V. The dust and PAH emission SEDs of disk galaxies
• The WIRCam Deep Survey I: Counts, colours and mass-functions derived from near-infrared imaging in the CFHTLS deep fields
• Constraining the thick disc formation scenario of the Milky Way
• The local space density of Sb-Sdm galaxies as function of their scalesize, surface brightness and luminosity
• Insights into the content and spatial distribution of dust from the integrated spectral properties of galaxies
• The Bimodal Galaxy Stellar Mass Function in the COSMOS Survey to z~1: A Steep Faint End and a New Galaxy Dichotomy
• The VIMOS VLT Deep Survey - Evolution of the luminosity functions by galaxy type up to z=1.5 from first epoch data
• Nuisance hardened data compression for fast likelihood-free inference
• Automatic physical inference with information maximising neural networks
• Likelihood-Free Cosmological Inference with Type Ia Supernovae: Approximate Bayesian Computation for a Complete Treatment of Uncertainty
• Generalized massive optimal data compression
• Approximate Bayesian Computation for Forward Modeling in Cosmology
• Evolution over time of the Milky Way's disc shape
• Pégase.3: A code for modeling the UV-to-IR/submm spectral and chemical evolution of galaxies with dust
• Quasi integral of motion for axisymmetric potentials
• Comparing PyMorph and SDSS photometry. II. The differences are more than semantics and are not dominated by intracluster light
• Measurement of the B-band Galaxy Luminosity Function with Approximate Bayesian Computation
• The red and blue galaxy populations in the GOODS field: evidence for an excess of red dwarfs
• Accelerating Approximate Bayesian Computation with Quantile Regression: Application to Cosmological Redshift Distributions
• The Effect of Surface Brightness Dimming in the Selection of High-z Galaxies
• The $z<1.2$ optical luminosity function from a sample of $\sim 410000$ galaxies in bootes
• The ALHAMBRA survey : $B-$band luminosity function of quiescent and star-forming galaxies at $0.2\le z<1$ by PDF analysis
• Inferring the photometric and size evolution of galaxies from image simulations
• Simulating Deep Hubble Images With Semi-empirical Models of Galaxy Formation
• The First Source Counts at 18 microns from the AKARI NEP Survey
• A Preferential Attachment Model for the Stellar Initial Mass Function

Cited by in corpus (5)

• The Cosmic Graph: Optimal Information Extraction from Large-Scale Structure using Catalogues
• The PAU Survey: Measurement of Narrow-band galaxy properties with Approximate Bayesian Computation
• How informative are summaries of the cosmic 21-cm signal?
• CLAP. I. Resolving miscalibration for deep learning-based galaxy photometric redshift estimation
• Hybrid summary statistics: neural weak lensing inference beyond the power spectrum