Automatic visualization and control of arbitrary numerical simulations

Authors’ preprint version as submitted to ECCOMAS Congress 2016, Minisymposium 505 - Interactive Simulations in Computational Engineering. Abstract: Visualization of numerical simulation data has become a cornerstone for many industries and research areas today. There exists a large amount of software support, which is usually tied to specific problem domains or simulation platforms. However, numerical simulations have commonalities in the building blocks of their descriptions (e. g., dimensionality, range constraints, sample frequency). Instead of encoding these descriptions and their meaning into software architecures we propose to base their interpretation and evaluation on a data-centric model. This approach draws much inspiration from work of the IEEE Simulation Interoperability Standards Group as currently applied in distributed (military) training and simulation scenarios and seeks to extend those ideas. By using an extensible self-describing protocol format, simulation users as well as simulation-code providers would be able to express the meaning of their data even if no access to the underlying source code was available or if new and unforseen use cases emerge. A protocol definition will allow simulation-domain experts to describe constraints that can be used for automatically creating appropriate visualizations of simulation data and control interfaces. Potentially, this will enable leveraging innovations on both the simulation and visualization side of the problem continuum. We envision the design and development of algorithms and software tools for the automatic visualization of complex data from numerical simulations executed on a wide variety of platforms (e. g., remote HPC systems, local many-core or GPU-based systems). We also envisage using this automatically gathered information to control (or steer) the simulation while it is running, as well as providing the ability for fine-tuning representational aspects of the visualizations produced