Venue H10-31.

Nov. 12

Casper de Vries (Erasmus Universiteit Rotterdam)


The herodotus paradox

The Babylonian bridal auction, described by Herodotus, is regarded as one of the earliest uses of an auction in history. Yet, to our knowledge, the literature lacks a formal equilibrium analysis of this auction. We provide such an analysis for the two-player case with complete and incomplete information, and in so doing identify what we call the "Herodotus Paradox."

Time: 12:00h.

Oct. 29

Frank Karsten (Technical University of Eindhoven)


Analysis of resource pooling games via a new extension of the Erlang loss function.

We study a situation where several independent service firms, such as call centers, hold a number of servers for their customer populations. Each firm is modeled as an Erlang loss system and faces holding costs for their servers and penalty costs for lost customers. The firms may collaborate by full pooling of their servers and individual customer streams. We examine the allocation of costs of the pooled system amongst the firms by formulating a cooperative cost game in which each coalition optimizes the number of servers. We identify a cost allocation that is in the core of this game, i.e., no subset of firms has an incentive to split off and form a separate pooling group. For our analysis we define a new extension of the Erlang loss function to a non-integral number of servers and establish several of its properties.

Time: 12:00h.

Apr. 16

Anita Schöbel (Georg-August Universität Göttingen)


New approaches to Robust Optimization

Many practical problems suffer from inaccurate, missing, or unreliable input data. This is a severe problem, since even small changes can make an optimal solution completely useless for practice. Robust optimization approaches try to hedge against uncertain data. The goal is to find solutions which are good for all scenarios contained in some given uncertainty set.

In this talk we will give an overview about models on robust optimization including the "classical" concepts as well as the more recent concepts of light robustness and recovery robustness. We will then present a new generic approach which enables us to generate robust solutions if a solution procedure for the certain optimization problem is known. Properties and first numerical results will be presented.

We will also discuss the applicability of the various approaches in particular with respect to applications occurring in public transportation such as line planning, timetabling, and delay management.




Remy Spliet
Room: 11-05
Phone: 010-4081342