Truncated Gaussian

% function X = TruncatedGaussian(sigma, range)
% X = TruncatedGaussian(sigma, range, n)

Generate a pseudo-random vector X of size n, X are drawn from the truncated Gaussian distribution in a RANGE braket; and satisfies std(X)=sigma.

RANGE is of the form [left,right] defining the braket where X belongs. For a scalar input RANGE, the braket is [-RANGE,RANGE].

If input SIGMA is negative, X will be forced to have the same "shape" of distribution function than the unbounded Gaussian with standard deviation -SIGMA: N(0,-SIGMA). It is similar to calling RANDN and throw away values ouside RANGE. In this case, the standard deviation of the truncated Gaussian will be different than -SIGMA. The *effective* mean and the effective standard deviation can be obtained by calling:

[X meaneffective sigmaeffective] = TruncatedGaussian(...)

http://en.wikipedia.org/wiki/Truncated_normal_distribution