Date: Mon, 3 Aug 1998 15:08:11 +0200 (CEST) From: Malte Lance <malte.lance@gmx.net> To: Luoqi Chen <luoqi@watermarkgroup.com> Cc: reilly@zeta.org.au, hackers@FreeBSD.ORG, jgrosch@mooseriver.com, shocking@prth.pgs.com Subject: Re: Fast FFT routines with source? Message-ID: <13765.46057.61247.598795@neuron.webmore.de> In-Reply-To: <199808031101.HAA12632@lor.watermarkgroup.com> References: <199808031101.HAA12632@lor.watermarkgroup.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Luoqi Chen writes: > > > No, Numerical Recipes is not even a little bit optimised, and > > > although the Fortran version is OK, the C version is horrible, > > > being a translitteration from the Fortran version, for the most > > > part. The hackery to get Fortran-style offset-1 array indexing is > > > particularly nasty. > > > > Had the same feelings abouot NR. More of an recipe "how to write a > > fft (... and introduce maximal confusion by strange indexing)". > > > Code samples (yes, they are samples, I never use them as they are) in NR > are horrible, but no other book could beat NR on explaining how an > algorithm works (have you read the FFT section in Sedgewick's Algorithms? > Instead of explaining how *FFT* works, it tries to explain what are > roots of unity, does that belong to a high school Algebra book?) Don't know about US-highschools. To know "how FFT works", you have to know what this nasty numbers in the frequency-domain stand for and where they come from. Also you need to know why you are able to reuse intermediary calculation-results (bit-reversion/reordering). The answers to this questions are easy, when you have knowledge about unit-roots and exponentials. Unit-roots and exponentials are really not that hard, that they shouldn't be explained in a basic-level analysis or algebra book. Malte. > > -lq To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?13765.46057.61247.598795>