= Bivariate normal distribution
{wiki}
= Bivariate normal density
{synonym}
A <bivariate normal distribution> is a <multivariate normal distribution> with two coordinates. It is determined by <mean>[means] $\mu_1,\mu_2$, positive <standard deviation>[standard deviations] $\sigma_1,\sigma_2$, and <correlation coefficient> $\rho\in[-1,1]$. For $|\rho|<1$ it has joint <probability density function>[probability density]
$$
f(x_1,x_2)=\frac1{2\pi\sigma_1\sigma_2\sqrt{1-\rho^2}}\exp\left[-\frac1{2(1-\rho^2)}\left\{\frac{(x_1-\mu_1)^2}{\sigma_1^2}-\frac{2\rho(x_1-\mu_1)(x_2-\mu_2)}{\sigma_1\sigma_2}+\frac{(x_2-\mu_2)^2}{\sigma_2^2}\right\}\right].
$$
One construction uses <independent random variables>[independent] <standard normal distribution>[standard normal] variables $U,V$: set $X_1=\mu_1+\sigma_1U$ and $X_2=\mu_2+\sigma_2(\rho U+\sqrt{1-\rho^2}V)$. This verifies the <mean>[means], <variances> and <covariance> $\rho\sigma_1\sigma_2$, and also explains the limiting cases. At $\rho=\pm1$ the law is supported on a line and has no two-dimensional Lebesgue density. At $\rho=0$ the density factors into its marginal normal densities and the coordinates are <independent random variables>[independent].
Each <conditional distribution>[conditional law] is normal: $X_2\mid X_1=x$ has <mean> $\mu_2+\rho(\sigma_2/\sigma_1)(x-\mu_1)$ and <variance> $\sigma_2^2(1-\rho^2)$. For example, equally scaled standardized variables with correlation $1/2$ have <conditional variance> $3/4$, quantifying the uncertainty left after observing the first coordinate.
Back to article page