Francesco Montorsi

Hi,
great toolbox, thanks.
By the way, I agree with Allan's comment (see below) that regarding the Von Mises distribution, it may be useful to have an implementation with higher numerical stability. In particular, I added this trivial function, which returns the log-pdf of the Von Mises distrib:

function [p alpha] = circ_vm_logpdf(alpha, thetahat, kappa)

% if no angles are supplied, 100 evenly spaced points around the circle are
% chosen
if nargin < 1 || isempty(alpha)
alpha = linspace(0, 2*pi, 101)';
alpha = alpha(1:end-1);
end
if nargin < 3
kappa = 1;
end
if nargin < 2
thetahat = 0;
end

alpha = alpha(:);

% evaluate pdf
C = -log( 2*pi*besseli(0,kappa) );
p = C + kappa*cos(alpha-thetahat);

Thanks to the greater numerical stability log-pdfs are often used in place of pdfs, so this little function may be of help to others...

Works great! perfect to start interfacing C++ object-oriented code to MATLAB

Very nice function!
Note however there is a typo, at line 29, in the comments containing an example to run, it should be "timebar(h,i/100)" instead of "timebar(h,1/100)", otherwise the progress signaled to the timebar is always 1/100 instead of proceeding toward 100 :)

I've updated the ZIP file with corrected files!

yes, you're right. I will replace global with persistent keyword asap :)