Networks are a ubiquitous tool to model the growing amount of data collected in science and business. In areas such as telecommunications, location theory, or social networks analysis (SNA), the size of the resulting networks and application data is ever increasing and analysis methods for large-scale data are crucial to deal with them in a meaningful way. Typically there are also inherent uncertainties associated with the input data and the implied optimization problems often face multiple objectives. These three aspects (large-scale data, uncertainty and multiple objectives) limit the applicability of available optimization algorithms that are capable to deal with at most one (if any) of these aspects. Following our recent
thinning out modeling approach, we aim to develop novel mathematical models and algorithmic solutions for solving highly relevant problems from operations management, telecommunications, and SNA at the large scale. We plan to exploit parallelization techniques, sparse mathematical models and general purpose heuristics, to reconsider decomposition approaches, and apply them in the context of high performance computing. The obtained results will enable the consideration of uncertain input data and / or multiple objectives at the large scale. To this end, various robust optimization concepts and their applicability in multi-objective settings will be analyzed. Results will be used to derive high-performance solution methods aiming to solve realistic, large-scale problem instances. These approaches will be based on mixed-integer (non-linear) optimization, heuristics and parallelization methods.