Inhalt des Dokuments
|Project heads:||Nicole Megow (until March 2016)|
|Student research assistant:||Jacob Focke|
|Duration:||June 2014 - May 2017|
This research is carried out in the framework of Matheon supported by Einstein Foundation Berlin
Uncertainty in the input data is an omnipresent issue in most real world planning processes. Metropolitan infrastructure -its design, operation and maintenance- induces complex planning processes where data uncertainty lies, e. g. in processing durations, transit times, cost, market prices, customer demands, capacity, bandwidth, energy consumption, et cetera. Since decisions on the infrastructure are typically very cost-intensive and of long-term impact, there is an urgent need of best possible mathematical support in this decision making process.
The quality of solutions for optimization problems (e. g. in infrastructure networks) with uncertain input data crucially depends on the amount of uncertainty. More information, or even knowing the exact data, allows for significantly improved solutions. It is impossible to fully abolish/avoid uncertainty. Nevertheless, it is sometimes possible to obtain exact data, but it may involve certain exploration cost in time, money, energy, bandwidth, etc.
In telecommunication networks planning, for example, information on the existing infrastructure (copper lines, optical fiber etc.) or the transmission range might not be easily available. The challenging major task of this project is to develop a structural understanding and algorithmic methods on how to balance the cost for data exploration with the actual benefit for the quality of solution to the optimization problem under consideration.
- Experimental evluation of algorithms for Minimum Spanning Tree under Uncertainty, Code and Data (ZIP, 14,4 MB)
- Organization of the BIMoS Day on Combinatorial Optimization and Efficient Algorithms, November 2015
- Organization of Autumn School on Approximation Algorithms for Stochastic Optimization, September 2014