authorea Authorea 10.36227/techrxiv.171078237.77836600/v1 Multi-Discounting Reinforcement Learning Based on Reward Decomposition 0009-0007-4307-5618 Chen Pengbin Liu Qi Li Yanjie Yan Kejian Ma Shuaikang 18 3 2024 This preprint is available at https://doi.org/10.36227/techrxiv.171078237.77836600/v1 computing and processing discount factor mdp reinforcement learning reward decomposition robotics and control systems