Ifft and fft for sparse vectors

Asked May 10 '16 at 01:12

Active Aug 31 '18 at 16:43

Viewed 577 times

I am trying to optimize some code for performance and i notice that i am forced to make a conversion from sparse to full vectors since the built-in ifft and fft functions do not support the sparse matrix type. My signal is however sparse and I want to exploit this fact.

Does anybody have a suggestion what you can do here?

Nithin

asked May 10 '16 at 01:12

Nithin Govindarajan

1

How does the sparceness ratio compare to log(N)/N ? – hotpaw2 May 10 '16 at 02:01
You could always calculate the Fourier-integral which goes O(N^2) rather than O (N log N). Per se the FFT is well defined over equally spaced data, so there is no trivial `sparse FFT`. – roadrunner66 May 10 '16 at 04:39
This varies per problem, but typically it will be very small compared log(N)/N. – Nithin Govindarajan May 10 '16 at 16:42

Ifft and fft for sparse vectors

0 Answers0