[Numpy-discussion] Pruned / sparse FFT 2D

[Daπid]

Hi,

I have a sparse cloud of points of ~10^4 points randomly scattered around a
triangular domain [1] from which I want to take the Fourier transform. I
have been looking for algorithms and found one library, but only appears to
be for the 1-D case (and seems there is no documentation). In [3] there is
C code for the FFTW, but seems it code is itself pruned (pun intended).

Does anyone have a clue about how to perform it? Speed is not a big issue,
but accuracy is quite important.


Thanks,

David.


[1] http://www.asaaf.org/~davidmh/coverage.png
[2] https://pypi.python.org/pypi/pynfftls/1.0
[3] http://www.fftw.org/pruned.html
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/numpy-discussion/attachments/20130415/c720c9df/attachment.html>