[Nipy-devel] Hello nipy-devel

Ariel Rokem arokem@berkeley....
Tue Jan 11 15:18:22 CST 2011

Hi Matthew and Eli,

On Tue, Jan 11, 2011 at 7:42 AM, Matthew Brett <matthew.brett@gmail.com> wrote:
> Ariel, Fernando, have you got any time to scan pbrain to see if
> there's any overlap between nitime and pbrain in terms of algorithms?
> I imagine there is - and that might be a good way to build up a
> central core of algorithms across the packages.

There is definitely some overlap. Some of the algorithms in nitime are
simply wrappers around calls to matplotlib.mlab and I suspect that
pbrain includes earlier implementations of these same mlab functions.
In particular, we use the Welch PSD/CSD implementation from mlab.
However, we tried to make it so that the implementation of other
algorithms, which rely on calculation of CSD/PSD, such as coherence
are flexible and allow using other algorithms (not the Welch windowed
periodogram) for calculation of  CSD/PSD.

Eli - welcome to the list! I am wondering about the time-local
coherence algorithm you described. Do I understand correctly that you
calculate a time-point by time-point coherence value between pairs of
electrodes? How do you do that? Do you use a sliding window?



Ariel Rokem
Helen Wills Neuroscience Institute
University of California, Berkeley

More information about the Nipy-devel mailing list