Sunday, January 22, 2012

Faster FFT !?


The above is a graph accompanied an article in SciTechDaily a few days ago entitled  "New Algorithm Faster Than Fourier Transform."  It is a picture I believe all wave aficionados are familiar with.  The article attracted my attention because I think the title should be more accurately stated as "Faster than Fast Fourier Transform".  As one who was around as a bystander at the beginning of the FFT algorithm in the late 1960's and FFT became an integral part of wave data analysis -- long before the present day super computer / faster PC era.  Since everything is getting faster and faster noadays, it is really not surprising to see a Faster FFT -- will that be FFFT?!  But what going through my mind is what difference would that be made to our data analysis or daily life?  Can we really notice the difference in speed by any chance?  Or in other words, hasn't our technology advancement reached the stage of irrelevance?

John Tukey (1915-2000)

I still remember the excitement of FFT's early days, wall to wall papers about it in every conference, and the new household name of Cooley-Tukey -- even though no one knew or cared who Cooley really is!  Now after nearly 5 decades, it is just made faster! So?  It is going to be presented in Association for Computing Machinery’s Symposium on Discrete Algorithms (SODA) this week.  Again, So?  I would be more interested in hearing what John Tukey (1915-2000) thinks about it.  Would he be excited?

No comments: