|
From: | al davis |
Subject: | Re: [Gnucap-devel] parallel |
Date: | Fri, 7 Mar 2014 01:40:54 -0500 |
User-agent: | KMail/1.13.7 (Linux/3.2.0-4-amd64; KDE/4.8.4; x86_64; ; ) |
Here is an exercise .. a homework assignment .. to enhance understanding of the algorithms. Consider the well known Fast Fourier Transform. Compare it to a plain (non-fast) discrete Fourier Transform that takes the same data and computes the same result. What is it that makes the "Fast" Fourier Transform fast? How was it derived? What is the principle behind it? How much faster is it? How does that concept relate to the matrix ordering being discussed here? Where else might this concept be used to make it faster?
[Prev in Thread] | Current Thread | [Next in Thread] |