As is well known, the uniform distribution on the unit sphere $\mathbb{S}^{n-1}$ is equivalent to the normalized standard Gaussian distribution (cf. this link). That is, if a random vector $\boldsymbol{x}\sim\mathcal{N}(0,\boldsymbol{I}_n)$, then $\frac{\boldsymbol{x}}{\|\boldsymbol{x}\|_2}$ is uniformly distributed on the unit sphere $\mathbb{S}^{n-1}$. My question is, if there is an equivalent representation for the uniform distribution on the nonnegative part of the unit sphere $\mathbb{S}^{n-1}\cap\mathbb{R}_+^n$?
Asked
Active
Viewed 45 times
0
-
Can't you generate $x \sim N(0, I)$ then take absolute values, then normalize? – Qiaochu Yuan Sep 29 '22 at 05:48
-
@QiaochuYuan Wow! Yes, this is an elegant way that I have not come up with. Thank you. – jwguan Oct 02 '22 at 03:06