Covariance matrix is positive definite, does the variable have density?

133 Views Asked by At

I know that if a random $n$-dimensional variable $Y= [Y_, ..., Y_n]^T$ has density $f$, then its covariance matrix $\Sigma $ is positive definite: $\forall x \in \mathbb{R}^n \setminus \{0\}: x^T \Sigma x >0$.

I wonder if the reverse implication is true.

$\Sigma>0 \ \implies Y$ has a density function.

I've tried to find such a theorem, but I failed. I do not know whether it is false or very trivial.

Could you help me with that?

It appears that this is not true for $n-$dimensional random variables in general.

Will it help if we assume that $Y$ has an $n-$dimensional normal distribution?