commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Endo, Roger" <re...@amgen.com>
Subject RE: [math] Complex dilemmas
Date Wed, 11 Jun 2003 23:45:32 GMT
Hi again.  This is a fun subject.  Brings back memories.

Back in 1999 I struggled to see the "Fast" in any Java FFT implementation
compared to a C implementation, especially when iterating over the x and y
axis for a 2D FFT.  Hopefully things can be better these days.  

And it is true that FFT usually can be kept as re and im separate doubles.
But usually you want to do some sort of complex math before or after the FFT
so at some point complex datastructures and complex matrix math would be
nice.

I'll check back in later on this year when complex math can get more
attention.

Cheers,
Roger Endo


> -----Original Message-----
> From: Al Chou [mailto:hotfusionman@yahoo.com]
> Sent: Wednesday, June 11, 2003 4:04 PM
> To: Jakarta Commons Developers List
> Subject: Re: [math] Complex dilemmas
> 
> 
> --- "J.Pietschmann" <j3322ptm@yahoo.de> wrote:
> > Al Chou wrote:
> > > Thanks for the input, Roger.  I think for FFT's in 
> particular, we would
> > want to
> > > profile whether using a complex number class would be 
> fast enough compared
> > to
> > > using two primitive doubles.  The "Fast" in "Fast Fourier 
> Transform" should
> > be
> > > preserved as much as possible, IMHO.
> > 
> > Don't waste your time: FFT is actually easier to implement
> > using double primitives, it's also likely to be faster, and
> > if you really want to get the last performance percent, cache
> > locality is a major factor which is even more a point against
> > using a Complex class.
> 
> Thanks for the commentary!  I've never coded FFT's myself, so 
> I didn't want to
> come down too hard on the idea of using complex number 
> objects in the absence
> of experience, even though I suspected what you said.
> 
> 
> > Study the Great Internet Mersenne Prime search sources for
> > an interesting read about the problems (high precision
> > number multiplication is basically FFT too).
> 
> Cool reference, thanks.  Did you mean "study the source 
> code"?  There don't
> seem to be any expositive materials on the subject at their site
> (http://www.mersenne.org/).
> 
> 
> > I personally found usage of complex numbers only natural for
> > certain EE calculations (involving alternate current) and
> > advcanced scientific computation. All other areas of financial
> > math, statistics, civil and mechanical engineering can do
> > quite comfortably without using complex numbers.
> 
> That's my general feeling, too.
> 
> 
> 
> Al
> 
> =====
> Albert Davidson Chou
> 
>     Get answers to Mac questions at http://www.Mac-Mgrs.org/ .
> 
> __________________________________
> Do you Yahoo!?
> Yahoo! Calendar - Free online calendar with sync to Outlook(TM).
> http://calendar.yahoo.com
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: commons-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: commons-dev-help@jakarta.apache.org
> 

---------------------------------------------------------------------
To unsubscribe, e-mail: commons-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: commons-dev-help@jakarta.apache.org


Mime
View raw message