Green simulation with database Monte Carlo

We develop a green simulation procedure that reuses simulation outputs of the current experiment to improve the computational efficiency of future experiments. We consider practical situations where idle computational resource is available after delivering a simulation answer within given time limit. When used correctly, such idle resource can be valuable simulation investment that can benefit future experiments. In repeated simulations when the same simulation model is run with different inputs at different times, our green simulation procedure repeatedly invests idle computations into databases, which are then used to improve the accuracy of future experiments. Our numerical results show that, as more and more outputs are reused, our proposed estimator has improving accuracy within fixed time limit.