On building a multilingual Parallel Computing Environment for Python, R, and MATLAB(r)

March 27, 2008
2:50 pm - 4:00 pm
Halligan 111
Speaker: Alan Edelman, MIT

Abstract

The lack of mathematical standards for scientfic computing languages is hindering free growth and transparent comparisons. Should your numerical language be allowed to give wrong answers? Different answers? We will also demonstrateStar-P from Interactive Supercomputing if internet conditions allow. As a puzzle for MATLAB experts, try explaining why x and y which seem equal in every way (both are "identical" 3x1 vectors) >> x=[1 -1 i]'; y=[1;-1;-i]; yield different answers for sort(x) and sort(y).