DOXIV: DIY-OSUFlow eXtension and Integration with VTK ¾ Toward Parallel Visualization of Large Flow Data through MPI

Zhanping Liu
Seminar

CT

Argonne National Laboratory
Building 240/Room 4301
Host Tom Peterka
Event Sponsor Mathematics and Computing Science Seminar

Scientific visualization is a very important application of computer graphics to scientific computing by providing deep insight into the pattern underlying large-scale data. Flow visualization plays a crucial role in a wide variety of areas such as oceanographic-atmospheric modeling, computational fluid dynamics simulation, and electro-magnetic field analysis, to name only a few. As a geometry-based approach, flow lines (e.g., streamlines and pathlines) remain a straightforward, effective, and efficient representation for visualizing flows ranging from steady to unsteady (time-varying) and from 2D to surface and volume. As flow data grows at exponential rates to exa-scale (1018) computing, it is nearly impossible to comprehend, interpret, and communicate complex phenomena resulting from sophisticated modeling and massive simulation. Thus, it is necessary and very promising to resort to parallel computing to achieve high-performance visualization to meet the increasing demands of domain scientists.

This talk presents Dr. Liu’s recent work on DOXIV (DIY-OSUFlow eXtension and Integration with VTK) toward parallel visualization of large flow data through MPI (Message Passing Interface). Specifically, VTK (Visualization Tool Kit: a large-scale, cross-platform, open-source, general-purpose visualization package developed by Kitware, Inc.) I/O has been integrated with DIY (Do It Yourself: a cross-platform, versatile, easy-to-use, MPI-based parallel computing library developed by Dr. Thomas Peterka here at ANL-MCS) and OSUFlow (a flow line and surface generation library developed by Ohio State University, initially connected but in part with DIY), involving volume partitioning and blocks distribution yet with a particular focus on demonstrating the practical applicability of DIY-OSUFlow to parallel visualization of steady/time-varying flow data of various formats coupled with flexible analysis output and powerful graphics rendering. Furthermore, DIY-OSUFlow has been extended to support parallel flow line generation over curvilinear grids in addition to Cartesian grids. Also briefly reported is Dr. Liu’s research in progress on load balancing of flow visualization over curvilinear grids.

Biography:

Dr. Zhanping Liu received the BS degree in mathematics from Nankai University (1992) and the PhD degree in computer science from Peking University (2000), P. R. China. He is a tenure-track assistant professor of computer science at Kentucky State University, a 2013 ANL-MCS FRPP (Faculty Research Participation Program) visitor hosted by Dr. Thomas Peterka. Previously, he was a research scientist with the Section of Biomedical Image Analysis under the School of Medicine at the University of Pennsylvania (2010 ~ 2011), the Scientific Computing and Visualization Group at Kitware, Inc. (“Leaders in Visualization Technology”, 2008 ~ 2010), and the Visualization Analysis and Imaging Lab of the High-Performance Computing Collaboratory (HPC2: formerly the NSF Engineering Research Center for Computational Field Simulation) at Mississippi State University (2001 ~ 2008), and much earlier a post-doctoral associate with the Micro-CT Image Reconstruction and Volume Visualization Lab of the School of Medicine at the University of Iowa (2000 ~ 2001). His research interests lie in computer graphics, scientific visualization (particularly flow/vector field visualization), image processing, data analysis, and parallel visualization as the integration of high-performance computing with scientific visualization, while he has several papers published in IEEE TVCG (Transactions on Visualization and Computer Graphics). More information about him is available at www.zhanpingliu.org.