CODEWITHSUNDEEP
r
Matata
Matata

Reputation: 271

generate multi-normal data in R

I was trying to genrate multi-normal data in R using mvrnorm, however, I got errors as:

> epsiloni <- mvrnorm(n = 1, rep(0,8), diag(1), tol = 1e-6, empirical = FALSE, EISPACK = FALSE)
Error in mvrnorm(n = 1, rep(0, 8), diag(1), tol = 1e-06, empirical = FALSE,  
: incompatible arguments

But it works fine for other expectation and variance, like

betai <- mvrnorm(n = 1, mu, D, tol = 1e-6, empirical = FALSE, EISPACK = FALSE)

where mu and D are specified otherwise.

The only difference between these two are just the expectations and the variance, but I did not see anything wrong with the mean and variance in epsiloni, they are just multi-standard normal.

Thanks!

Upvotes: 2

Views: 1449

Answers (1)

Julius Vainora
Julius Vainora

Reputation: 48241

Setting Sigma to diag(1), which is a 1x1 matrix, is the problem. If you want an identity covariance matrix for an 8-dimensional vector, you need diag(8).

mvrnorm(n = 1, rep(0, 8), diag(8))
# [1] -0.3554192  0.6051595  0.3926595  0.2752819  0.8610572  0.2679094 -1.3581420 -0.2814057

Upvotes: 3

Related Questions