Date of Award

2010

Degree Type

Thesis

Degree Name

Master of Science

Program

Computer Science

Supervisor

Dr. Marc Moreno Maza

Abstract

In scientific computing, it is often required to evaluate a polynomial expression (or a matrix depending on some variables) at many points which are not known in advance or with coordinates containing “symbolic expressions”. In these circumstances, standard evaluation schemes, such as those based on Fast Fourier Transforms do not apply. Given a polynomial f expressed as the sum of its terms, we propose an algorithm which generates a representation of f optimizing the process of evaluating f at some points. In addition, this evaluation of f can be done efficiently in terms of data locality and parallelism. We have implemented our algorithm in the Cilk++ concurrency platform and our implementation achieves nearly linear speedup on 16 cores with large enough input. For some large polynomials, the generated schedule can be evaluated at least 10 times faster than the schedules produced by other available software solutions. Moreover, our code can handle much larger input polynomials.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.