A random vector has a multivariate normal distribution when every real linear combination of its coordinates is normal, allowing zero variance. Its mean and covariance matrix determine the characteristic functionIf is positive definite, its probability density function is . A singular covariance matrix gives a law supported on a proper affine subspace, without such a full-dimensional density. Joint normality makes zero cross-covariance equivalent to independence, unlike general distributions.
If are independent and identically distributed random variables with a centered multivariate normal distribution and covariance matrix , then has the same probability law as . Its two marginal covariance matrices are and its cross-covariance is zero. The pair has a multivariate normal distribution, so the two components are independent. This remains valid when is singular.
A Gaussian random vector has a multivariate normal distribution; equivalently, every real linear combination of its coordinates is a normal random variable, allowing zero variance. Its expected value vector and covariance matrix determine its law. Subvectors are independent when their cross-covariances vanish, a feature particular to joint Gaussianity.
For centered, unit-variance jointly Gaussian random variables with correlation coefficient , the displayed identity holds. Realize the pair as projections onto unit vectors of angle . Rotational symmetry of the two-dimensional Gaussian random vector direction makes sign disagreement have probability . The expectation is therefore . Perfectly correlated and anticorrelated endpoints follow directly; the sign at zero does not affect the result.
A Gaussian random vector with independent normal coordinates of expectation zero and variance one. Its distribution is invariant under orthogonal transformations. In particular, its direction in a fixed two-dimensional plane is uniform, which underlies the Gaussian sign-correlation identity.
For a jointly Gaussian vector, conditional independence is equivalent to vanishing conditional cross-covariance. When the full covariance matrix is positive definite, two coordinates are conditionally independent given all the others exactly when their off-diagonal entry in the precision matrix is zero.
A bivariate normal distribution is a multivariate normal distribution with two coordinates. It is determined by means , positive standard deviations , and correlation coefficient . For it has joint probability densityOne construction uses independent standard normal variables : set and . This verifies the means, variances and covariance , and also explains the limiting cases. At the law is supported on a line and has no two-dimensional Lebesgue density. At the density factors into its marginal normal densities and the coordinates are independent.
Each conditional law is normal: has mean and variance . For example, equally scaled standardized variables with correlation have conditional variance , quantifying the uncertainty left after observing the first coordinate.
A Bivariate standard normal distribution is a bivariate normal distribution whose two marginal means are zero and marginal variances are one; its remaining parameter is their correlation.
For independent Gaussian random vectors with known zero mean and positive-definite covariance matrix , the candidate maximum-likelihood covariance estimator isWhen is positive definite, the Gaussian log likelihood, up to a constant, is . Writing the precision matrix as gives the concave objective , whose derivative vanishes uniquely at . This proves the displayed estimator. It is unbiased because the mean is known: .
If the sample vectors do not span the whole -dimensional space, is singular. Shrinking a candidate covariance eigenvalue to zero along an unobserved direction makes the likelihood unbounded, so there is no maximum within the positive-definite parameter space. In particular, cannot yield a positive-definite maximum-likelihood estimate. For a positive-definite Gaussian population and , the vectors span the space almost surely. When the mean is instead estimated from the same data, centering changes the rank condition and the unbiased covariance uses denominator rather than .
Let be independent, identically distributed centered bivariate normal vectors with covariance matrix , and let . ThenThe asymptotic covariance belongs to the -scaled error; the unscaled estimator has covariance , in fact exactly for every . The distinction prevents a missing-sample-size error.
To derive it, apply the multivariate central limit theorem to the independent vectors . The Isserlis theorem gives , and . Subtracting products of means gives, for example, and . In general .
Conditionally on ,
When has full column rank, its Gram matrix is positive-definite and . The stacked normal vector therefore has zero cross covariance, so its blocks are independent. Since , the linear and quadratic statistics are independent. This is the projection argument behind independence of regression coefficient and variance estimators in a normal linear model.
Articles by others on the same topic
There are currently no matching articles.