Monte Carlo Sampling for the Probabilistic Orienteering Problem
The Probabilistic Orienteering Problem is a variant of the orienteering problem where customers are available with a certain probability. Given a solution, the calculation of the objective function value is complex since there is no linear expression for the expected total cost. In this work we approximate the objective function value with a Monte Carlo Sampling technique and present a computational study about precision and speed of such a method. We show that the evaluation based on Monte Carlo Sampling is fast and suitable to be used inside heuristic solvers. Monte Carlo Sampling is also used as a decisional tool to heuristically understand how many of the customers of a tour can be effectively visited before the given deadline is incurred.
KeywordsProbabilistic Orienteering Problem Monte Carlo Sampling Heuristic algorithms
Xiaochen Chou was supported by the Swiss National Science Foundation through grant 200020\(\_\)156259: “Hybrid Sampling-based metaheuristics for Stochastic Optimization Problems with Deadlines”.