Sigma for variance?

stanc

#1

Usually sigma stands for stddev, and sigma^2 is the variance. However, I see this notation in the manual

(y - mu)’ * Sigma * (y - mu)

this tell me Sigma is actually the variance. Should I assume in the notation

y ~ normal(x, sigma);
sigma means variance?


#2

For the regular normal, the second argument is the standard deviation. For the multivariate normal (where you might see a product that looks like (y - mu)' Sigma^-1 (y - mu)), it’s the covariance matrix.

Where in the manual is this?

When it comes to distributions, whatever software/language you’re using, it’s worth looking up the parameterization just to be clear.

Hope that helps.


#3

ok, That sounds better. The product is given on page 57. I do see that it was declared on page 39

corr_matrix[3] Sigma;
declares Sigma to be a 3 × 3 correlation matrix.


#4

You should read the manual. It’s all well documented and uses conventional variable naming.


#5

we all do our best to RTFM, but it really helps to jump start with some experts help.


#6

Unfortunately, we don’t have unlimited time, so we appreciate people trying to answer their own questions before asking on the list.