[Numpy-discussion] numpy FFT memory accumulation
Charles R Harris
Thu Nov 1 20:50:29 CDT 2007
On 11/1/07, Ray S <firstname.lastname@example.org> wrote:
> At 09:00 AM 11/1/2007, Chuck wrote:
> In Python, collections.deque makes a pretty good circular buffer.
> Numpy will
> make an array out of it, which involves a copy, but it might be
> better than what you are doing now.
> hmmm, I'll think more about that - and the copy is only at program
> start, it seems
> the fft can always be fft(d[:-N])
> >>> from collections import deque
> >>> import numpy as N
> >>> d = deque(N.zeros(10,))
> >>> d.extend(N.ones(4,))
> >>> d
> deque([0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0, 1.0,
> 1.0, 1.0])
> >>> [d.pop() for i in range(4)]
> [1.0, 1.0, 1.0, 1.0]
Yeah, I noticed that inconvenience.
> deque([0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0])
> An additional complication is that I pass the numpy (or Numeric)
> array address to the ctypes library call so that the data is placed
> directly into the array from the call. I use the if/else end wrap
> logic to determine whether I need to do a split and copy if the new
> data wraps.
OK. Hmm, I wonder if you would lose much by taking a straight forward
radix-2 fft and teaching it to use modular indices? Probably not worth the
trouble, but an fft tailored to a ring buffer might be useful for other
things. Probably the easiest thing is to just copy the ring buffer out into
a linear array.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Numpy-discussion