Re: [Numpy-discussion] numpy FFT memory accumulation
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] d 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. Thanks, Ray
On 11/1/07, Ray S <subscriber100@rjs.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.
d 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. Chuck
participants (2)
-
Charles R Harris -
Ray S