An $L^2$ Analysis of Reinforcement Learning in High Dimensions with Kernel and Neural Network Approximation. CSIAM-AM. 2022;3(2):191-220. doi:10.4208/csiam-am.SO-2021-0026