设随机向量 X∈Rd 的均值为 μ=E[X],协方差矩阵定义为
Σ=Cov(X)=E[(X−μ)(X−μ)T].
任取向量 a∈Rd,有
aTΣa=aTE[(X−μ)(X−μ)T]a=E[aT(X−μ)(X−μ)Ta]=E[(aT(X−μ))2]=Var(aTX)≥0.
因此对任意 a 都有 aTΣa≥0,故 Σ 为半正定矩阵。
英文解析
Let the mean value of the random vector X∈Rd be μ=E[X], and the covariance matrix is defined as
Σ=Cov(X)=E[(X−μ)(X−μ)T].
Optional vector a∈Rd, with
aTΣa=aTE[(X−μ)(X−μ)T]a=E[aT(X−μ)(X−μ)Ta]=E[(aT(X−μ))2]=Var(aTX)≥0.
Therefore, for any a, there is aTΣa≥0, so Σ is a semi-positive definite matrix.