function L = log_dirichlet( H, A);%% L = log_dirichlet( H, A);%% H ( k x N) : N k-dimensional distributions - NOTE: have to be LOGS of normalized distribution values !!!% A ( k x C) : parameters of C Dirichlets with dimension k% % L ( C x N) : log-likelihoods under the different DirichletsN = size(H,2);L = (A-1)' * H;L = L + repmat( gammaln(sum(A,1)') - sum( gammaln(A),1)', 1, N);