Randomly generated Gaussian distribution sequence data randn

I personally think that randn is approximate when the data is very long. The details are as follows:
Assume that you want to generate a random sequence with the mean of B and the variance of A, inMATLABThe following table is provided in help.
X = B + SQRT (a) * randn (1,000000 );
(You can use mean (X) and VAR (X) to check the mean value and variance. We will find that the mean value is approximately equal to B and ).
The mean and variance of data generated in a short period of time are generally not met. I usually do this (that is, I also produce the data when the data is long) for reference only:
According to your requirements, assume that the mean value of the 200 column in a row is B, and the variance is:
X = randn (1,200 );
X = x-mean (X );
X = B + SQRT (a) * x/SQRT (VAR (x ));