Interactive Visual Analysis of Multi-Parameter Scientific Data
Increasing complexity and a large number of control parameters make the design and understanding of modern engineering systems impossible without simulation today. Recent advancements in simulation and computing make it possible to compute large simulation ensembles. A simulation ensemble consists of multiple simulation runs of the same model with different values of control parameters. In order to cope with ensemble data, a modern analysis methodology is necessary.

In this talk, we present our experience with simulation ensemble exploration and steering by means of interactive visual analysis. We present how interactive visual analysis can be used to gain a deep understanding in the heterogeneous ensemble simulation data. A single data point in our case does not contain scalars or vectors only, as usual. Instead, a single data point contains scalars, time series, and other types of mappings. Interactive visual analysis utilizes a tight feedback loop of computation/visualization and user interaction to facilitate knowledge discovery in complex datasets.

Most of the analysis techniques consider the data as a static source. Such an approach often hinders the analysis. We introduce a concept of interactive visual steering for simulation nsembles. We link the data generation and data exploration and analysis tasks in a single workflow. We employ interactive and automatic methods simultaneously. This makes it possible to tune and optimize complex systems having high dimensional parameter space and complex outputs.

The research results have been evaluated in a tight collaboration with domain experts from industry. Very positive feedback from them indicates the usefulness of newly proposed analysis methods.
Lecturer:
Dr. Kresimir Matkovic, VRVis Forschungs-GmbH
Dates:
Fr. 29.04.2016, 13:00 c.t., G29-335