OptimalPolicies.jl

Overview

Parallel Tempering is a computational method to sample complex probability distributions, which is common in simulating physical and chemical processes. The Parallel Tempering algorithm runs multiple Markov chain Monte Carlo (MCMC) simulations at different temperatures in parallel to explore high-dimensional spaces, where the temperature refers to a parameter that controls the acceptance rate of the MCMC simulation.

Optimal policy search in economics refers to finding policies that maximize or minimize certain objectives, such as maximizing profit and minimizing costs. The economic models in this context can be highly complex, involving numerous variables and parameters. Traditional optimization methods could be struggled by those models due to their nature of high-dimensionality. However, Parallel Tempering can be adapted to solve those problems by running multiple simulations of the economic model with different parameter settings defined as different temperatures. Therefore, Parallel Tempering has the potential to be a systematic approach of economic policy search for policymakers. This Julia package aims to utilize the Parallel Tempering algorithm to find optimal transportation networks.