Code covered by the BSD License  

Highlights from
Generation of Random Variates

image thumbnail

Generation of Random Variates

by

 

generates random variates from over 870 univariate distributions

nc_f_pdf(x,nu1,nu2,delta)
function Y = nc_f_pdf(x,nu1,nu2,delta)
%NC_F_PDF Noncentral F probability density function (pdf).
%   Y = NC_F_PDF(X,NU1,NU2,DELTA) returns the noncentral F pdf with numerator 
%   degrees of freedom (df), NU1, denominator df, NU2, and noncentrality
%   parameter, DELTA, at the values in X.
%
%   The size of Y is the common size of the input arguments. A scalar input  
%   functions as a constant matrix of the same size as the other inputs.     

%   Reference:
%      [1]  Johnson, Norman, and Kotz, Samuel, "Distributions in
%      Statistics: Continuous Univariate Distributions-2", Wiley
%      1970 p. 191. (equation 5)
%      [2]  Evans, Merran, Hastings, Nicholas and Peacock, Brian,
%      "Statistical Distributions, Second Edition", Wiley
%      1993 p. 73-74.

%   B.A. Jones 2-7-95, ZP You 9-17-98
%   Copyright 1993-2000 The MathWorks, Inc. 
%   $Revision: 2.13 $  $Date: 2000/05/26 18:53:05 $

if nargin < 4, 
    error('Requires four input arguments.');
end

[errorcode, x, nu1, nu2, delta] = distchck(4,x,nu1,nu2,delta);

if errorcode > 0
    error('Requires non-scalar arguments to match in size.');
end

[m,n] = size(x);

% Initialize Y to zero.
y = zeros(m,n);
Y = y;

k = (nu1 <= 0 | nu2 <= 0 | x < 0 | delta < 0);
k1 = ~k;
% set Y to NaN for indices where X is negative or NU1 or NU2 are not positives integers.
Y(k) = NaN;
if ~any(k1), return; end

if any(k) % reset variable so that we don't have to deal with bad indices.
   y = y(k1);
   nu1 = nu1(k1);
   nu2 = nu2(k1);
   x = x(k1);
   delta = delta(k1);
end

% to simply computation, pre-divide nu1,nu2 and delta
nu1 = nu1/2; nu2 = nu2/2; delta = delta/2;

% Set up for infinite sum.
j = 0;
g = nu1.*x./nu2;
% Sum the series.
while j < 1000  % so that we don't get into a infinit loop
   b      = beta(j + nu1,nu2);
   tmp    = poiss_pdf(j,delta).*(g.^(j-1+nu1))./((1+g).^(j + nu1 + nu2));
   deltay = tmp./b;
   y = y + deltay;
   
   % Convergence test.
   if all(deltay(:)./(y(:)+eps^(1/4)) < sqrt(eps)), break; end
   j = j + 1;
end

if (j == 1000) 
  warning('Failed to converge, use the result cautiously.\n');
end

Y(k1) = nu1.*y./nu2;

return

Contact us