Galaxy formation through the lens of galaxystructure with semi-empirical models and deep learning - PixelCNN output log-likelihood ratios

This is the dataset that supports the University of Southampton Doctoral Thesis "Galaxy formation through the lens of galaxystructure with semi-empirical models and deep learning". This dataset relates to Chapter 8. It stores the log-likelihood ratio values for galaxies in: the Sloan Digital Sky Survey (SDSS), the Illustris simulation, the Illustris TNG 100,the Illustris TNG 50 simulations and the "Sersic blobs" simulations. The columns are the value of the likelihood ratio, the likelihood of the individual pixelCNN models (trained on SDSS and the Sersic blobs) and the ID identifying each objects. For SDSS, two files are provided: one includes the whole dataset ('SDSS_all.csv') and one only the test data ('SDSS_test.csv'). The SDSS catalogue used in this paper is coupled to the Meert et al. 2015 catalogs (the SDSSsee http://alan-meert-website-aws.s3-website-us-east-1.amazonaws.com/fit_catalog/index.html for the Meert et al. catalog) based on the SDSS Data Release 7. In the two SDSS files the SDSS Data Release 7 ID is reported ("objid") as well as the ID for the Meert et al. catalogue ("galcount"), for convenience . See https://www.illustris-project.org/ and https://www.tng-project.org/ for more information concerning the Illustris and Illustris TNG simulations respectively, as well as for open source data to be matched with the likelihood ratios provided here.