EPS2016: EURO PhD School 2016 is a gathering of PhD students and young researchers working on the field of Operations Research. During one week, participants will explore the challenging domain of Matheuristics along with its application to the Vehicle Routing Problem (VRP). Additionally, participants will learn how to make their algorithms available to the research community by means of web services. The EPS2016 will be hosted by the Operations Research Group of the Université de Bretagne-Sud in Lorient, France from April 20th to 28th, 2016. This event is organized with the financial support of EURO (The Association of European Operational Research Societies).
During EPS2016, experts in the domains of metaheuristics and mathematical programming will join the participants in an attempt to discover some of the best ways to create optimization algorithms. A computer architect will demonstrate how to implement web services efficiently in order to make the algorithms accessible to other users. The academic programme consist of a series of lectures together with practical hands-on sessions. In these sessions, participants will develop their own approach to solve one variant of the VRP, and will be able to compare their approach to those implemented by other students.
Matheuristics combine the best of two radically different approaches to optimization: mathematical programming and metaheuristics. On the one hand, mathematical programming focuses on understanding and exploiting the underlying structure of the optimization problem, and is generally used to create exact methods (i.e., methods that guarantee to find an optimal solution). On the other hand, metaheuristics are more pragmatical in that they focus on finding good solutions in a reasonable amount of computing time, and for this reason forsake the guarantee of optimality.
In recent years, both schools of optimization have grown towards each other. Researchers in the field of metaheuristics have been attracted towards the sturdy mathematical and theoretical underpinning of the field of mathematical programming, to the existence of ever more powerful solvers, and to the ability of mathematical programming approaches to solve a large number of well-defined problems to optimality. On the other hand, the field of mathematical programming has been attempting to integrate ideas from the field of metaheuristics in an attempt to render its algorithms more efficient and approach the usability of metaheuristics on real-life optimization problems.
For many years, vehicle routing problems have attracted researchers and they have developed several methods to solve a large number of variants of these problems. With this EURO PhD School, we would like to contribute to the field as well by developing some matheuristics algorithms for specific variants of the VRP.
Another important aspect of the development of optimization tools is
their availability. In most cases, researchers develop their own
computer programs and optimization solvers, perfectly tailored to solve
one specific class of problems. As soon as they (or somebody else) need
to adapt their solver to another type/class of problems, they have to
start over again and most probably, they will throw away 80% of the work
done previously to keep only small pieces of former computer programs.
To make these programs available, one possibility is to transform them
into web-services, reachable from any platform and from anywhere in the
world where an Internet connection is open. Part of the program will be dedicated to the development of a platform that will be available for free for small companies or developers.