Deutsch Intern
Institute of Computer Science

28.10.2024

Informatik-Kolloquium

Auf Einladung von Prof. Dr. Marie Schmidt findet der folgende Vortrag statt:

Montag, 28. Oktober 2024, 16:15 Uhr, Übungsraum I, Informatikgebäude, Am Hubland

Prof. Dr. Clemens Thielen
TUM Campus Straubing

The Multiobjective Temporal Shortest Path Problem

 

Abstract

When making decisions, we usually pursue multiple goals and strive to achieve all of these goals as well as possible. Optimization problems that arise in practice are therefore oftenmultiobjective problems in which several conflicting objectives have to be taken into account simultaneously. When choosing a path for a train journey, for example, one may want to minimize not only the travel time, but also other objectives such as the cost of the ticket or the number of transfers. Moreover, since trains have fixed departure times, this application naturally yields a shortest path problem in a so-called temporal graph, in which each arc can only be entered at a specified start time (which corresponds to the departure time of the train). The resulting shortest path problem with multiple objectives is therefore called the multiobjective temporal shortest path problem.

This talk presents a general label setting algorithm for the single-source version of the multiobjective temporal shortest path problem that can handle a large variety of different objectives. The only condition imposed on the objectives is a monotonicity property that generalizes the nonnegativity of the arc costs required for the well-known label setting algorithm for solving the static single-source shortest path problem in both the singleobjective and the multiobjective case. The worst-case running time of our algorithm is polynomial in the sum of the input size of the problem instance and the number of nondominated images (vectors of objective values of efficient paths), which means that it runs in polynomial time for all tractable versions of the problem (i.e., for all versions where the number of nondominated images remains polynomial in the input size).

To complement this result, we provide a complete classification into tractable and intractable problems for all single-source multiobjective temporal shortest path problems involving a large variety of objectives.

URL:  https://www.professoren.tum.de/thielen-clemens