rndKMn¶

Purpose¶

Returns a matrix of standard normal (pseudo) random variables and the state of the random number generator.

Format¶

{ y, newstate } = rndKMn(r, c, state)
Parameters: r (scalar) – row dimension. c (scalar) – column dimension. state (scalar or 500x1 vector) – scalar case state = starting seed value only. If -1, GAUSS computes the starting seed based on the system clock. 500x1 vector case state = the state vector returned from a previous call to one of the rndKM random number functions. y (RxC matrix) – Standard normal random numbers. newstate (500x1 vector) – the updated state.

Examples¶

This example generates two thousand vectors of standard normal random numbers, each with one million elements. The state of the random number generator after each iteration is used as an input to the next generation of random numbers.

state = 13;
n = 2000;
k = 1000000;
c = 0;
submean = {};

do while c < n;
{ y,state } = rndKMn(k,1,state);
submean = submean | meanc(y);
c = c + k;
endo;

mean = meanc(submean);
print mean;


Remarks¶

r and c will be truncated to integers if necessary.

Technical Notes¶

rndKMn() calls the uniform random number generator that is the basis for rndKMu() multiple times for each normal random number generated. This is the recur-with-carry KISS+Monster algorithm described in the rndKMi() Technical Notes. Potential normal random numbers are filtered using the fast acceptance-rejection algorithm proposed by Kinderman, A.J. and J.G. Ramage, “Computer Generation of Normal Random Numbers,” Journal of the American Statistical Association, December 1976, Volume 71, Number 356, pp. 893-896. It employs the error correction from Tirler et al. (2004), “An error in the Kinderman-Ramage method and how to fix it,” Computational and Data Analysis, Vol. 47, 433-40.