SmoothViz: Visualization of Smoothed Particle Hydrodynamics Simulation Data

• Applicants: Professor Dr.-Ing. Lars Linsen; Professor Dr. Paul Rosenthal
• Subject Area: Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
• Term: from 2008 to 2016
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 69608725

• Final Report Year: 2015

Final Report Abstract

The SmoothViz project was concerned with the visualization and interactive visual analysis of smoothed particle hydrodynamics (SPH) simulation data with application in astrophysics. Within the duration of the project, many scientific contributions have been made. Within the second funding period (a two-year extension of the first funding period), methods have been developed that allow for handling large data sizes, analyzing multiple fields simultaneously (included multiple fields varying over time), analyzing multiple simulation runs in comparison to each other, and putting all approaches in an integrated system including the achievements of the first funding period. Handling large data sizes using a level-of-detail approach for particle data sets proved to be helpful for visualization purposes such that less data can be considered with minimized errors leading to faster rendering computation times. When analyzing physical simulations, it is desirable to consider all physical variables that contribute to the process simultaneously. We have shown that an interactive visual analysis approach based on coordinated views between physical space renderings and attribute space renderings of the multidimensional attribute space formed by the multiple fields can be effective. We contributed in multiple ways to improve the analysis process. In particular, we proposed a generalization of continuous scatterplots to projected views of higherdimensional data. Such continuous representations infer the spatial coherence given in the physical space to the attribute space, effectively allowing for the application of interpolation processes. Operating in the multidimensional attribute space can further be facilitated using control points for each cluster of data points. We also proved this approach to be helpful in the analysis of medical imaging data. Furthermore, we showed that the attribute space exploration using projected views is still useful when not concentrating at a single time step of the simulation, but when considering temporally changing data, i.e., the entire multi-field spatio-temporal data set. Physical simulation typically depend on multiple input parameters or initial configurations. One important aspect of the analysis process is to understand the influence of the input parameter values on the outcome of the simulation. Hence, multiple simulation runs are executed and need to be analyzed in a comparative fashion. We have extended our approach to the analysis of multi-run simulation data. We have shown that our approach allows for a comparative analysis of individual simulation runs leading to overall patterns and a comprehension of the input parameter’s influence. This allows for a more detailed interactive analysis than just providing statistical, i.e., aggregated, information about all runs. We have integrated the visual analysis components we developed into the SmoothViz software system, which was made available publically and presented to SPH domain scientists at multiple occasions.

Publications

• SmoothViz: Visualization of Smoothed Particle Hydrodynamics Data. In Hydrodynamics Optimizing Methods and Tools, pp. 328, 2011
  Lars Linsen, Vladimir Molchanov, Petar Dobrev, Stephan Rosswog, Paul Rosenthal, Tran Van Long
  
• Variational Level-Set Detection of Local Isosurfaces from Unstructured Point-based Volume Data. In Scientific Visualization: Interactions, Features, Metaphors, Dagstuhl Follow-Ups, vol. 2, pp. 222239, 2011
  Vladimir Molchanov, Paul Rosenthal, Lars Linsen
  
• Continuous Representation of Projected Attribute Spaces of Multifields over Any Spatial Sampling, Computer Graphics Forum, vol. 33, iss. 3, pp. 301310, 2013
  Vladimir Molchanov, Alexey Fofonov, Lars Linsen
  
• SmoothViz: An Interactive Visual Analysis System for SPH Data. In Proceedings of the 8th International Smoothed Particle Hydrodynamics European Research Interest Community (SPHERIC) Workshop, pp. 350356, 2013
  Vladimir Molchanov, Alexey Fofonov, Stephan Rosswog, Paul Rosenthal, Lars Linsen
  
• Visual Exploration of Patterns in Multi-run Time-varying Multi-field Simulation Data Using Projected Views. In Proceedings of WSCG, The 22nd International Conference on Computer Graphics, Visualization and Computer Vision, pp. 39 48, 2014
  Vladimir Molchanov, Lars Linsen
  
• Error-minimizing SPH Particle Merging for Constructing Multi-resolution Hierarchies, In Proceedings of the 10th International Smoothed Particle Hydrodynamics European Research Interest Community (SPHERIC) Workshop, 2015
  Jonathan Fischer, Lars Linsen, Paul Rosenthal