SIM_RL: A New Approach for Integrating Simulation with Reinforcement Learning

Computer simulation, the process of mathematical modelling performed on a computer, is designed to predict the behavior of a real-world system. As a system becomes more complex, the simulation engine must run numerous times in response to the increasing complexity of the input and the simulation process. Additionally, an expensive physical experiment needs to be performed to validate the results. This paper demonstrates an innovative, general-purpose simulation approach strengthened by refinement learning (RL), formalized in the SIM_RL algorithm, and using epidemic spread (COVID-19) test data. The main advantages of this approach are computational resource savings, reduced need for physical experiments, and the ability to predict system behavior based on actual results. Moreover, this approach can be used in various disciplines to solve complex simulation problems.