Doctor of Philosophy (PhD)
This research develops a comprehensive framework for interactive walkthrough involving one billion particles in an immersive virtual environment to enable interrogative visualization of large atomistic simulation data. As a mixture of scientific and engineering approaches, the framework is based on four key techniques: adaptive data compression based on space-filling curves, octree-based visibility and occlusion culling, predictive caching based on machine learning, and scalable data reduction based on parallel and distributed processing. In terms of parallel rendering, this system combines functional parallelism, data parallelism, and temporal parallelism to improve interactivity. The visualization framework will be applicable not only to material simulation, but also to computational biology, applied mathematics, mechanical engineering, and nanotechnology, etc.
Document Availability at the Time of Submission
Release the entire work immediately for access worldwide.
Liu, Xinlian, "Fast scalable visualization techniques for interactive billion-particle walkthrough" (2002). LSU Doctoral Dissertations. 2664.