Knowledge Discovery in a Virtual Universe

1
Knowledge Discovery in a Virtual Universe
Project Summary
Last Years Highlights

• Wrote 3-point correlation function and
  friends-of-friends group finder using the Ntropy
  framework.
• Implemented automatic dynamic workload
  management.
• Solidified Ntropy API
• Developed new parallel group finding algorithm

• Develop a framework, called Ntropy, for the rapid
  development of massively parallel data analysis
  applications for algorithms that use data trees.

http//www.psc.edu/gardnerj/ntropy
Impact on Science
Jeffrey P. Gardner
The cost of developing scalable parallel programs
is generally prohibitive for data analysis
applications. Ntropy lowers the cost of
development by over a factor of 10, thereby
enabling the scientists to harness the power of
massively parallel supercomputers.
Pittsburgh Supercomputing Center
Plans for Next Year
gardnerj_at_psc.edu

• Generate and publish science results for Ntropy
  applications
• Publish work on Ntropy and new parallel group
  finding algorithm.
• Develop and implement parallel density estimation
  algorithm

Andrew Connolly
University of Pittsburgh
ajc_at_phyast.pitt.edu
NASA AISR 2004-07