Research - Network & Data Science Management

While a brief and coarse introduction of common research directions of our working group is on the landing page, please find below some more specific example domains of our research.

Exact algorithms for Operations Research and interdisciplinary applications

We develop exact methods for decision making an optimization problems that have dedicated applications in, e.g., logistics, transportation, health-care, or production and manufacturing, or that serve as a model of interdisciplinary or even overarching relevance. Thereby, we have a particular expertise in assignment, matching, ordering, layout, scheduling, and routing applications.

• Sophisticated methods based on integer programming and combinatorial algorithms, like e.g. enhanced separation procedures and branch-and-cut algorithms.
• Modeling and reformulation towards a better practical solvability (push frontiers of instances solved).
• Linear or quadratic programming relaxations and polynomial-time algorithms (flows, shortest-paths, matchings, ...).

Methods for quadratic optimization problems

Many operations research problems depend (regarding the objective or the feasibility of a solution) on simultaneous decisions, and thus on quadratic terms. We address this by different problem-specific methods, such as e.g.:

• Linearization and convexification techniques, for instance the Inductive Linearization Technique.
• Integrating techniques from Quadratic Unconstrained Binary Optimization (QUBO) and corresponding resolution methods.
• Direct Techniques based on convex relaxations, or inexact non-convex methods.

Algorithm Engineering

Typically, our developments and publications follow well the Algorithm Engineering paradigm, in particular in terms of a repeated integration of and refinement based on the following ingredients:

• Theoretical investigations: Polyhedral Theory, Graph Theory, Complexity Theory.
• Algorithmic Implementations of High-Performance.
• Qualified Experimental Studies.