Forward Modeling of Galaxy Populations for Cosmological Redshift Distribution Inference

Abstract We present a forward-modeling framework for estimating galaxy redshift distributions from photometric surveys. Our forward model is composed of: detailed population describing the intrinsic distribution of physical characteristics galaxies, encoding evolution physics; stellar synthesis connecting properties galaxies to their photometry; data characterizing observation and calibration processes given survey; explicit treatment selection cuts, both into main analysis sample subsequent sorting tomographic bins. This approach has appeal that it does not rely on spectroscopic data, provides control over modeling assumptions builds direct bridge between photo- z inference physics. In addition distributions, drawing robust inferences about statistical more generally. demonstrate utility by Galaxy And Mass Assembly (GAMA) survey Vimos VLT Deep Survey (VVDS), validating against redshifts. baseline able predict GAMA VVDS with respective biases ? ? 0.003 ? 0.01 mean redshift—comfortably accurate enough Stage III cosmological surveys—without any hyperparameter tuning (i.e., prior doing fitting those data). anticipate additional improvements, will provide path IV