The dark energy survey 5-year photometrically identified Type Ia supernovae

As part of the cosmology analysis using Type Ia Supernovae (SN Ia) inthe Dark Energy Survey (DES), we present photometrically identified SNIa samples using multi-band light-curves and host galaxy redshifts. Forthis analysis, we use the photometric classification frameworkSuperNNova (SNN; Möller et al. 2019) trained on realistic DES-likesimulations. For reliable classification, we process the DES SNprogramme (DES-SN) data and introduce improvements to the classifierarchitecture, obtaining classification accuracies of more than 98 percent on simulations. This is the first SN classification to make use ofensemble methods, resulting in more robust samples. Using photometry,host galaxy redshifts, and a classification probability requirement, weidentify 1,863 SNe Ia from which we select 1,484 cosmology-grade SNe Iaspanning the redshift range of 0.07 < z < 1.14. We find goodagreement between the light-curve properties of thephotometrically-selected sample and simulations. Additionally, we createsimilar SN Ia samples using two types of Bayesian Neural Networkclassifiers that provide uncertainties on the classificationprobabilities. We test the feasibility of using these uncertainties asindicators for out-of-distribution candidates and model confidence.Finally, we discuss the implications of photometric samples andclassification methods for future surveys such as Vera C. RubinObservatory Legacy Survey of Space and Time (LSST)