A Machine-learning Approach to Predict Missing Flux Densities in Multiband Galaxy Surveys

We present a new method based on information theory to find the optimal number of bands required measure physical properties galaxies with desired accuracy. As proof concept, using recently updated COSMOS catalog (COSMOS2020), we identify most relevant wavebands for measuring in Hawaii Two-0 (H20)- and UVISTA-like survey sample $i<25$ AB mag galaxies. that available $i$-band fluxes, $r$, $u$, IRAC/$ch2$ $z$ provide regarding redshift importance decreasing from $r$-band $z$-band. also same sample, IRAC/$ch2$, $Y$, $r$ $u$ are stellar mass measurements order importance. Investigating inter-correlation between bands, train model predict UVISTA observations near-IR H20-like observations. magnitudes $YJH$ can be simulated/predicted an accuracy $1\sigma$ scatter $\lesssim 0.2$ brighter than 24 bands. One should note these conclusions depend selection criteria sample. For any different selection, results remeasured. Our suggest presence limited machine learning trained over population observed extensive spectral coverage outperforms template-fitting. Such maximally comprises acquired surveys breaks degeneracies parameter space template-fitting inevitable few