[SciPy-User] scipy.signal.filtfilt small off by one error
Paul Northug
pnorthug at gmail.com
Wed Feb 23 23:03:49 EST 2011
I am using scipy 0.8.0 on a Mac. I am not sure, but there may be very
small error in this function in the line,
s=r_[2*x[0]-x[edge:1:-1],x,2*x[-1]-x[-1:-edge:-1]]
should be,
s=r_[2*x[0]-x[edge-1:0:-1],x,2*x[-1]-x[-1:-edge:-1]]
The intent is to take a signal x and pad it on both sides with negative
mirrors of the ends of x of length edge. I have checked that this function
doesn't produce the same output as matlab's filtfilt on the boundaries
and it looks from the code that the intention was to match matlab.
If someone with better python knowledge could check this, that would
be great and I can put in a bug report if necessary.
More information about the SciPy-User
mailing list