A Practical Visualization Strategy for Large-Scale Supernovae CFD Simulations

Simulating the expansion of a Type II supernova using an adap- tive computational fluid dynamics (CFD) engine yields a complex mixture of turbulent flow with dozens of physical properties. The dataset shown in this sketch was initially simulated on iVEC's EPIC supercomputer (a 9600 core Linux cluster) using FLASH [Fryx- ell et al. 2000] to model the thermonuclear explosion, and later post-processed using a novel integration technique to derive the ra- dio frequency emission spectra of the expanding shock-wave front [Potter et al. 2011]. Model parameters have been chosen to simu- late the asymmetric properties of the SN 1987A remnant [Potter et al. 2009].

This offline workflow takes several hundred machine-hours to com- plete, and results in a volumetric time-series dataset stored on an adaptive mesh refinement grid with a total storage allocation of >= 10 terabytes. This dataset consists of several thousand time-steps (adaptively outputted at non-linear time intervals), each containing a dozen simulation variables stored as floating-point vector fields. Due to the intricate nature of the flow, visualizing these datasets requires a rendering engine capable of high-quality image recon- struction in order to maintain the underlying visual complexity.

In this sketch, we describe a practical approach we've developed which enables explorative visualization for studying large-scale time-series astrophysical CFD simulations. This is part of an ongo- ing data-intensive research project within our group to support the visualization of large-scale astrophysics datasets for the scientists at the International Centre for Radio Astronomy Research (ICRAR).

In particular, we discuss the application of progressive stochastic sampling and adjustable workloads to insure a consistent response time and a fixed frame-rate to guarantee interactivity. The user is permitted to adjust all rendering parameters while receiving contin- uous visual feedback, facilitating explorative visualization of our complex volumetric time-series datasets.