MATLAB: Matlab random number generation, normrnd VS mvnrnd

Question

Suppose I need to generate 5 iid RV's with mean 0 and stdev 0.5. I am aware of the distinction between normrnd and mvnrnd because the former is for a single random variable while mvnrnd is for multiple random variables.

I am however wondering if both normrnd(0,0.5,[1 5]) and mvnrnd(zeros(1,5),0.5^2*eye(5)) will suit the above objective.

I was told before that if the random number generator is good then the random variates obtained from normrnd(0,0.5,[1 5]) ought to be independent. Is this true or would mvnrnd(zeros(1,5),0.5^2*eye(5)) be the safer bet or would either way accomplish my objective?

Answer 1

These will both accomplish the same thing. mvnrnd() is more for creating correlated random variables (with uncorrelated as a special case, as you are using it).

normrnd() will definitely create iid distributions (with the general constraints of pseudorandom generation). Also, it has the merit of being significantly faster.