High Performance Analytics

The work of the department High Performance Analytics concentrates on two reserach fields,

  • network analysis, simulation, optimization and graph mining, for instance for electrical circuits, gas transport, cooling circuits, oil and energy,
  • robust design for statistical analysis and optimization of parameter-dependent problems, particularly in the areas of networks, production processes and industrial products.

Cooperation partners and customers stem from different areas such as oil and gas, chemistry, microelectronics, automotive and engineering industry.

We develop mathematical methods and software products. Our range of services includes studies, licenses of our products, tailored software solutions as well as integration of our software modules into other software products.

Originally stemming from the analysis of business processes, »High Performance Analytics« means highly efficient creation, storage and analysis of large amounts of data for gaining novel, nontrivial insights into processes, allowing for steering these processes, and / or optimizing them.

Applications arise from a variety of areas including but not limited to natural sciences, engineering, economical or social questions. Some examples:

  • networks: modeling, simulation, analysis / optimization in such areas as circuits, gas, water, energy, oil etc.
  • analysis of data stemming from measurements for controlling gas or water transport networks etc.
  • analysis and optimization of parameter-dependent industrial production processes or products taking parameter variations and tolerances into account (parameters describing the process, material properties, geometry etc.)
  • condition monitoring of machines or process monitoring of production processes
  • process monitoring for handling scanned documents and their OCR in the context of data management systems (DMS)
  • network analysis / control for crisis management

Methods from different areas have to be considered and further developed:

  • data mining, machine learning, metamodeling: the department develops and implements, for instance, interpolation-based data models (response surfaces) and statistical methods (DesParO).
  • robust multiobjective optimization: the department develops and implements efficient methods to be used mainly for metamodels (DesParO).
  • statistical analysis and robust optimization of parameter-dependent chains of (simulation) processes (PRO-CHAIN).
  • networks and graphs: the department develops and implements a framework for network simulation (MYNTS for electrical circuits, gas transport etc.) as well as software for analyzing, manipulating, matching of networks and graphs (net'O'graph).

In order to allow for large-scale data analysis, algorithms as well as data transport and storage methods have to get the most out of modern system architectures. Depending on the application at hand, technical computing and / or cloud computing play a decisive role.

The following flyers and articles are available:



Simulation and Optimization of Data Center Energy Flows

Energy management (DIN EN ISO 50001) and optimization based on online process mining

Hierarchical Simulation of Nanoelectronic Systems for Control of Process Variations

Further Material:

High Performance Analytics at a Glance