Interactive Visualization and Data Analysis of Extreme Scale Science

Within the last year, the research group “Interactive Visualization and Analysis for Extreme Scale Science” consequently continued to work on the their scientific goals: The efficient and effective visualization of large datasets. The group’s activities can be categorized in three areas of research: First, works concerning the direct integration of efficient visualization algorithms into supercomputer-based simulations, the so-called “in-situ methods”. Secondly, the research and development of highly adaptive rendering algorithms for a wide range of usage scenarios. A particular focus here lies on mobile devices to allow the visualization of data for a ubiquitous access. Third, the research and evaluation of novel visualization approaches to improve perception and visual analysis of the ever-growing amount of data.

In the first area – the in-situ visualization – the research group published two papers at internationally renowned conferences. The paper “An Approach to Lowering the In Situ Visualization Barrier” was published at the “ISAV’15: In Situ Infrastructures for Enabling Extreme-scale Analysis and Visualization“-workshop and it demonstrated, that it is possible for he most common types of simulation data to extract the visualization data directly from the executable without any access to the source code or the necessity of a modification thereof. This approach allows avoiding having to integrate the visualization code with the simulation tool, which is in many scenarios the reason to apply not adopting in-situ methods.

Furthermore, a second paper “Towards in Situ Visualization of Extreme Scale, Agent-Based, World-wide Disease Spreading Simulations” was published at the “SIGGRAPH Asia’15 Symposium on Visualization in High Performance Computing”, demonstrating that using an optimized space partition scheme allows a supercomputer to simulate large populations up to the point where every single human on the planet is considered separately. Of particular interested is that this partition scheme leads to a very efficient query-driven visualization of the simulation while it is running.

In the second sub-area, the mobile visualization, the research group continued to refine their basic concepts for mobile rendering. At the EuroVIS’15 conference, this was demonstrated in the publication “State of the Art in Mobile Volume Rendering on iOS Devices”. Focusing particularly on novel APIs such as Metal, a number of adaptive rendering approaches were developed and compared. In addition to this publication, the research group acquired funding for the five-year project “Mobile Decision Support System for Nurse Management of Neuromodulation Therapy“, which is funded by the National Institute of Nursing Research. The goal of this project is to validate the performance of our mobile visualization approach in a clinical setting. This project is carried out in collaboration with the Scientific Computing and Imaging Institute in Utah.

In the third area of visualization baseline research, the research group of Prof. Krüger collaborated with the University of Zurich and Clemson University on the paper “A Testbed Combining Visual Perception Models for Geographic Gaze Contingent Displays”, which was published at “EuroVIS’15” and which won the “Best Short Paper Award”. In this project, a test bed for the evaluation of the influence of different parameters of the human visual system (HVS) on visualization approaches was developed. It has been shown that the influence of the HVS can
have a disruptive effect on visualization techniques, ranging from enhances in perception to novel perceptually lossless compression techniques.

Finally, the group also continued their research in the area of dense flow visualization methods. At the “IEEE VIS’15” conference, the paper “Line Integral Convolution with Adaptive Frequency Management” was published, which demonstrates for the first time ever how to reliably control the frequencies of flow visualization.

Project Team

Principal Investigators
Prof. Dr. Jens Krüger