Please update your Flash Player to view content.

[View PDF]

 

Chun-Ming Chen, (Kitware, Inc.)
Han-Wei Shen, (OSU)
Berk Geveci (Kitware, Inc.)
Tom Peterka (ANL)

Objectives

  • Visualize large scale flow simulation data such as Nek 5000 and MJO climate simulations
  • Make available advanced parallel and out-of-core integral curve computation algorithms to  the application scientists
  • Support three-dimensional time-varying vector data in a wide variety of file formats and data types (curvilinear, unstructured, regular grids)

Impact

  • SDAV (OSU, Kitware, ANL) researchers can more quickly dissimilate their cutting-edge  flow visualization research results through VTK data models
  • Application scientists can plug in their data to the next generation SDAV visualization software more easily
  • SDAV software (OSUFlow, VTK, Paraview) are more tightly integrated together   

Accomplishments

Software: A new and improved OSUFlow software design for DOE’s leadership computing facility

Publications

  • Chun-Ming Chen and Han-Wei Shen, "Graph-based Seed Scheduling for Out-of-core FTLE and Pathline Computation", IEEE Symposium on Large Data Analysis and Visualization 2013
  • Chun-Ming Chen, Boonthanome Nouanesengsy, Teng-Yok Lee, Han-Wei Shen, "Flow-Guided File Layout for Out-of-core Pathline Computation", IEEE symposium on Large Data Analysis and Visualization
Unstructured Grid Curvilinear grid
Unstructured grid Curvilinear grid

 

Notes:

Objectives:

SDAV researchers are currently putting a great effort to help analyze the flow field data generated by DOE’s large scale simulations. One such example is OSUFlow, a collaborative project among OSU,  ANL, and Kitware. OSUFlow is a software library for high performance parallel particle tracing and flow data analysis. The software includes several novel parallel load balancing and I/O optimization algorithms that have  been previously published in the top venues such as ACM Supercomputing and IEEE Visualization conferences. One limitation of OSUFlow before was that it used a propriety data format so data sets need to be converted to the propriety format first before analysis can take place. This limitation causes some problems for certain applications because of the data conversion and movement cost.

In the past 6 months, researchers in OSU, ANL, and Kitware have gotten together to integrate OSUFlow with VTK’s data model. Our goal is to make available the state-of-the-art parallel and out-of-core flow analysis algorithms to the application scientists who have already adopt the VTK data model. The type of data that are supported are three-dimensional time-varying vector field defined in regular, curvilinear, and unstructured meshes. Any file formats that can be supported by VTK are to be read by OSUFlow.

Impacts:

The adoption of the VTK data model has been completed in October 2013. It is now possible for SDAV researchers to more easily dissimilate the research results to be used in production analysis tasks. It also provides a tighter collaboration platform to further the research of flow analysis and visualization among the SDAV researchers.

Accomplishments:

OSUFlow is downloadable from a software repository located at Argonne National Laboratory. We will continue to report our research finding in top quality conferences and journals. Two of our recent publications related to OSUFlow are listed on the slide.