mmvn_gen(m,varargin )

function [X idx theta] = mmvn_gen(m,varargin )
% MMVN_GEN generates random data drawn from a mixture of multivariate guassians
%
% function X = mmvn_gen(m, theta)          % theta is a mmvn struct
%     generates X, an m x d matrix of data specified by theta
% function X = mmvn_gen(m, M );            % M is a k x n matrix of means.
% function X = mmvn_gen(m, M, V);          % provide M and V
% function X = mmvn_gen(m, M, V, W );
% function X = mmvn_gen(IDX, ... );         
%       IDX is a grouping variable in 1..k that specifies which class
%       X(i,:) is drawn from 
% function [X idx] = mmvn_gen(...)         
%       returns idx, the source cluster for each observation
% function [X idx theta] = mmvn_gen(...)
%       returns theta, a standard parameter structre for mmvn_xxx functions
% example:
%   [X idx theta] = mmvn_gen( 200, [0 5; 5 0; 5 5; 0 0] );
%   %this example creates 200 points randomly selected from 4, 2 dimensional
%   %clusters with centers on a 5x5 cube
%   scatter( X(:,1), X(:,2), [], idx, 'filled' );
%
% See also
%       mmvn_getTheta, mmvn_fit

%% $Id: mmvn_gen.m,v 1.3 2007/07/08 19:17:07 Mike Exp $
% Copyright 2006 Mike Boedigheimer
% Amgen Inc.
% Department of Computational Biology
% [e-mail]

theta = mmvn_getTheta( varargin{:} );

[k d] = size(theta.M);  % k sources

if length(m) > 1
    idx   = m;
    m     = length(idx);
else
    idx   = randsample( k, m, 'true', theta.W );  % random discrete process to
                                                  % generate source
end

X     = nan(m,d);     % preallocate memory for result


% for each source
for i = 1:k
    [r p] = chol( theta.V(:,:,i) );
    if p ~= 0
        error('linstats:mmvn_gen:InvalidCovarianceMatrix', 'covariance matrix must be positive definite');
    end
    X(idx==i,:) = randn(sum(idx==i),d)*r;
end

X = X + theta.M(idx,:);