Professor, RWTH Aachen University
For me, becoming an associate member of GERAD felt like coming home, after many years of collaboration. It began exactly 20 years ago, when I needed an external PhD reviewer. Montréal has many column-generation experts, but I wanted one in particular: Jacques Desrosiers. Together we have now published 10 “Cahiers,” we co-organize (along with Guy Desaulniers) a biannual event (alternating between a school and a workshop) on column generation, and we are currently co-writing a book—the reference—on branch-and-price.
Since 2010, I have been a professor and the chair of operations research at RWTH Aachen University in Germany. I am a mathematician by training, having completed my PhD in 2001 at the Institute for Mathematical Optimization in Braunschweig (with Uwe Zimmermann), and my habilitation in 2007 at the Combinatorial Optimization and Graph Algorithms Group at TU Berlin (with Rolf Möhring). In 2009–10, I was a visiting professor on discrete optimization at Darmstadt.
My research interests lie in computational integer (linear) programming along with its various decomposition methods, discrete and combinatorial optimization, and everything relating to graphs, algorithm engineering, approximation algorithms and machine learning. I have worked, often in partnership with industry, in many areas of application, such as production, logistics, traffic, energy, healthcare, education, sports and politics.
Two topics are currently of special interest to me. The first is the overlap between machine learning and computational integer programming. In particular, we make use of learning techniques to improve optimization algorithms. For example, we worked on learning when to apply branch-and-cut vs. branch-and-price. The second area is GCG, a generic research code automating branch-price-and-cut, which aims to make the useful Dantzig-Wolfe reformulation available to everyone. It is important to recognize the model structure that can be exploited in reformulations, so we provide tools to the community that help understand this structure. In the long run, I would like to get hold of model semantics directly from an LP/MPS file in order to help users formulate better models.
I am very active on Twitter (@mluebbecke) and it is easy to reach me there. If you would like to spend time researching in Germany, as a PhD student, for a postdoc or as visiting faculty, be it for a short or long period, please don’t hesitate to contact me.
Associate member since december 2018
Cahiers du GERAD
This paper addresses the solution of the capacitated minimum cost flow problem on a network containing
\(n\) nodes and
\(m\) arcs. Satisfying necessary ...
This paper describes a vector space decomposition algorithmic framework for linear programming guided by dual feasibility considerations. The resolution pro...BibTeX reference