Parallel Large Data Visualization Course in Fall 2014
In Fall 2014, Dr. Philip Rhodes is teaching CSci 581-1, a Special Topics course on Visualization. This course will focus on a new topic: parallel large data visualization.
Scientific data comes from a variety of sources. Satellites produce terabytes of geospatial data each day, while mechanical and aerospace engineers run simulations of ignition data and fluid flow over a surface to create more efficient engines and airplane wings. Doctors use CT and MRI data to diagnose patients and find tumors. In all of these cases, producing an effective visual representation of the raw data is an important and necessary part of the process of understanding the significance of a dataset. However, increasingly large dataset size and complexity presents a growing bottleneck for the visualization of scientific data.
This course is an overview of issues relating to representation, handling, and visualization of scientific data. Topics include 2D and 3D visualization techniques, data formats for representation of regular and unstructured scientific data, and multidimensional access methods. In previous iterations of the course, visualization was conducted through our own Java code, and through the Visualization Tool Kit (vtk). This time around parallel visualization for large data will be a new focus. We’ll be using ParaView, a parallel variant of vtk, running on our own cluster. Information on both vtk and paraview (including image galleries) can be found at www.kitware.com.
The images are from an assignment given in a previous offering of the Visualization course.