Re: [eigen] A complex FFT for Eigen |

[ Thread Index | Date Index | More lists.tuxfamily.org/eigen Archives ]

*To*: eigen@xxxxxxxxxxxxxxxxxxx*Subject*: Re: [eigen] A complex FFT for Eigen*From*: Tim Molteno <tim@xxxxxxxxxxxxxxxxxxx>*Date*: Fri, 28 Nov 2008 23:02:00 +1300

Gael Guennebaud wrote:

First of all, I'm 100% OK to provide a Fourier module in Eigen. Myissues are more about how to do it. Looking at FFTW benchmarks, it ispretty clear to me that we *must* provide support for it.

So, in the same vein of the other modules, I would provide a defaultimplementation based on Tim's lightweight source code such that we canmake the dependency to FFTW optional for most use cases of thismodule. Of course this requires to extend a bit Tim's code to supportat least: real data, dynamic sizes, 1D and 2D data (I guess all aretrivial to add). IMO non pot sizes is not mandatory.About current Tim's code, what about adding a specialization ofRadix2_Decimation<> for sizes greater than, let's say 32, which woulduse normal recursive calls untill the size is 32. This wouldautomatically add support for dynamic sizes, and significantly reducethe size of the generated code.

Cheers Tim ---

**Follow-Ups**:**Re: [eigen] A complex FFT for Eigen***From:*Gael Guennebaud

**References**:**[eigen] A complex FFT for Eigen***From:*Tim Molteno

**Re: [eigen] A complex FFT for Eigen***From:*Benoit Jacob

**Re: [eigen] A complex FFT for Eigen***From:*Tim Molteno

**Re: [eigen] A complex FFT for Eigen***From:*Benoit Jacob

**Re: [eigen] A complex FFT for Eigen***From:*Gael Guennebaud

**Messages sorted by:**[ date | thread ]- Prev by Date:
**Re: [eigen] A complex FFT for Eigen** - Next by Date:
**Re: [eigen] A complex FFT for Eigen** - Previous by thread:
**Re: [eigen] A complex FFT for Eigen** - Next by thread:
**Re: [eigen] A complex FFT for Eigen**

Mail converted by MHonArc 2.6.19+ | http://listengine.tuxfamily.org/ |