DE | EN
Home
About Us
Overview
Facts and Figures
Organization
Scientists
Contact
Approach
Situations offered
Research
Overview
Application Fields
Projects
Publications
Scientists
Preprints
Institutional Cooperation
Archiv 02-14
Transfer
Overview
Industry
References
MODAL-AG
Spin Offs
Software
Patents
Schools
Overview
MathInside
MATHEATHLON
Matheon-Kalender
What'sMath
Training for Teachers
Summer Schools
Events
Press
Overview
Releases
News
Overview
Matheon Head
Number of the week
News 2002 - 2014
Activities
Overview
Workshops
15 Years Matheon
Media
Overview
Photos
Videos
Audios
Booklets
Books
News from around the world

Dr. Kai-Helge Becker

becker@zib.de


Projects as a member

  • MI10

    Acyclic network flows

    Dr. Benjamin Hiller / Prof. Dr. Thorsten Koch / Prof. Dr. Martin Skutella

    Project heads: Dr. Benjamin Hiller / Prof. Dr. Thorsten Koch / Prof. Dr. Martin Skutella
    Project members: Dr. Kai-Helge Becker
    Duration: -
    Status: running
    Located at: Konrad-Zuse-Zentrum für Informationstechnik Berlin

    Description

    Utility and infrastructure networks are at the heart of our daily life and we are taking their proper working for granted. To provide this service, network operators face difficult planning and operational problems. The complexity of the considered optimization problems increases for several reasons, for instance because geographically bigger networks are considered or the detail level is increased. For large networks, providing globally optimal solutions or at least good bounds for assessing solution quality is still a big challenge. This project addresses this challenge for so-called potential-driven nonlinear network flow problems that are a key model for infrastructure networks for fluids, e.g. water and gas. These problems feature a so-called potential for each node, and the flow on an arc is related to the difference of the potential of its end nodes. In particular, flow is always directed from higher to lower potential, i.e. the potentials induce an acylic orientation of the arcs. This observation is the motivation for this project: We study network flow problems with the additional requirement of acyclic flows: If each network arc is oriented in the direction of flow over this arc, then there is no directed cycle in the resulting network. This is an interesting combinatorial structure that arises from nonconvex nonlinear constraints and thus links combinatorial and continuous optimization. The aim of this project is to develop algorithmic techniques to exploit this structure to improve global optimization methods for large-scale nonlinear flow problems.

    http://www.zib.de/projects/acyclic-network-flows

Projects as a guest