Difference between lhsnorm and normrnd

I am trying to understand the differnce between lhsnorm and normrnd. I generate two samples using same mean and std butthese metods are giving very different sample? The code is below

n = 50; % number of observations

noise_std = .02; % standard deviation of noise

d = 30;

mu = 0*ones(d,1);

sd = 0.2*ones(d,1);

sigma = 0.2*eye(d);

xtrain = lhsnorm(mu,sigma,n);

xtrain1 = normrnd(0,0.2,n,d);

I expected xtrain and xtrain1 to me similar but it is not. 30 features all with 0 mean and 0.2 std. Plese let me know if i am making mistake with lhsnorm.

Answer 1

[Paul]

The doc page for lhsnorm is surprisingly sparse. But it does talk about the multi-variable normal distribution, in which case sigma is usally the covariance of the distribution. So for xtrain try

Sigma = 0.2^2 * eye(d)
xtrain = lhsnorm(mu,Sigma,n);

The sigma input to normrnd is the standard deviation, as you've done in the code.

[Ankush Kumar Mishra]

Thanks