Page 1 of 1

Bug in fnn.m

Posted: Mon Aug 5, 2013 08:45
by martin
Hello,

I recently discovered a bug in the False Nearest Neighbour-function fnn.m (rev. 5.8), which is part of the CRP-Toolbox. Obviously in ll. 149 and 154 the identifiers “t” (embedding delay) and “r” (neighbourhood criterion) are mixed up. A correct version might be:

Code: Select all

t = varargin{i_x(2)}; % line 149, 154 respectively
r = varargin{i_x(3)}; % line 150, 155 respectively
This bug is crucial when using method = 'silent' or 'nogui', because without the gui you can not control if the parameters are interpreted correctly.

Best regards,
Martin