Ntropy

From Wikipedia, the free encyclopedia
Jump to: navigation, search

NTropy is a computer library under development at the Pittsburgh Supercomputing Center and the University of Pittsburgh Department of Physics and Astronomy that enables the rapid development of any algorithm that uses a kd-tree for running on a parallel computer. In its current form, it is geared towards astrophysics data, which is generally point-like in nature meaning that most efficient analysis algorithms tend to employ tree data structures.

The idea behind NTropy, like most libraries, is that the derive the benefits of an efficient tree implementation and parallel computing without having to write a tree structure nor a parallel program from scratch. Instead, they take NTropy's existing tree structure and customize the data stored in each node to the needs of their application. Then they can write a serial piece of code that describes their tree walk algorithm. NTropy will then execute these tree walks in parallel on any sort of parallel computer, from a Beowulf cluster to a Massively Parallel Processing computer (MPP) to machines in a computational grid.

NTropy is being developed under a grant from NASA's Applied Information Systems Research Program.

External links[edit]