Abstract: Reinforcement learning (RL) has been widely applied in decision-making and decision-optimization tasks. This work proposes a feasible framework which adapts the RL to achieve scientific ...