MIT researchers have made a mathematical breakthrough by making FFTs faster and requiring less data
This is fascinating news for all you maths buffs or anyone doing signal processing: The venerable Fourier Transform, a mathematical technique for decomposing a signal into its component frequencies, has been considerably improved.
In 2012 researchers at MIT, discovered an algorithm that in some circumstances can perform hundreds of times faster than the fast Fourier transform (FFT) and now, in a paper to be presented at the ACM-SIAM Symposium on Discrete Algorithms in January, they’ve gone even further reducing the number of samples required to the theoretical minimum number of samples.
It is expected that this advance will significantly reduce how long it takes to make such things as MRI scans and produce radio telescope images as well as improve imaging accuracy.
Signal your thoughts below or to gearhead@gibbs.com then follow me on Twitter, App.net, and Facebook.
