On the Online Shortest Path Problem

This paper is concerned with the stochastic shortest path problem with recourse when limited forms of spatial and temporal arc cost dependencies are accounted for. Recourse is defined as the opportunity for a decision-maker to re-evaluate his or her remaining path when en route information is available. Typically, accounting for recourse within a model yields benefits due to the information obtained between problem stages. The impact and applicability of information, here modeled as arc cost dependencies, is of vital importance when dealing with such problems. System properties and algorithm development are explored under limited cases of spatial and temporal arc cost dependencies; the numerical results verify some of the theoretical insights and demonstrate the applicability of the algorithms