多变量正态分布亦称为多变量高斯分布。它是单维正态分布向多维的推广。它同矩阵正态分布有紧密的联系。 事实速览 记号, 参数 ...多元正态分布 概率密度函数 Many samples from a multivariate (bivariate) Gaussian distribution centered at (1,3) with a standard deviation of 3 in roughly the (0.878, 0.478) direction (longer vector) and of 1 in the second direction (shorter vector, orthogonal to the longer vector).记号 N ( μ , Σ ) {\displaystyle {\mathcal {N}}({\boldsymbol {\mu }},\,{\boldsymbol {\Sigma }})} 参数 μ ∈ RN — 位置Σ ∈ RN×N — 协方差矩阵 (半正定)值域 x ∈ μ+span(Σ) ⊆ RN概率密度函数 ( 2 π ) − N 2 | Σ | − 1 2 e − 1 2 ( x − μ ) ′ Σ − 1 ( x − μ ) , {\displaystyle (2\pi )^{-{\frac {N}{2}}}|{\boldsymbol {\Sigma }}|^{-{\frac {1}{2}}}\,e^{-{\frac {1}{2}}(\mathbf {x} -{\boldsymbol {\mu }})'{\boldsymbol {\Sigma }}^{-1}(\mathbf {x} -{\boldsymbol {\mu }})},} (仅当 Σ 为正定矩阵时)累积分布函数 解析表达式不存在期望值 μ众数 μ方差 Σ熵 1 2 ln ( ( 2 π e ) N | Σ | ) {\displaystyle {\frac {1}{2}}\ln((2\pi e)^{N}|{\boldsymbol {\Sigma }}|)} 矩生成函数 exp ( μ ′ t + 1 2 t ′ Σ t ) {\displaystyle \exp \!{\Big (}{\boldsymbol {\mu }}'\mathbf {t} +{\tfrac {1}{2}}\mathbf {t} '{\boldsymbol {\Sigma }}\mathbf {t} {\Big )}} 特征函数 exp ( i μ ′ t − 1 2 t ′ Σ t ) {\displaystyle \exp \!{\Big (}i{\boldsymbol {\mu }}'\mathbf {t} -{\tfrac {1}{2}}\mathbf {t} '{\boldsymbol {\Sigma }}\mathbf {t} {\Big )}} 关闭 一般形式 N维随机向量 X = [ X 1 , … , X N ] T {\displaystyle \ X=[X_{1},\dots ,X_{N}]^{T}} 如果服从多变量正态分布,必须满足下面的三个等价条件: 任何线性组合 Y = a 1 X 1 + ⋯ + a N X N {\displaystyle \ Y=a_{1}X_{1}+\cdots +a_{N}X_{N}} 服从正态分布。 存在随机向量 Z = [ Z 1 , … , Z M ] T {\displaystyle \ Z=[Z_{1},\dots ,Z_{M}]^{T}} ( 它的每个元素服从独立标准正态分布),向量 μ = [ μ 1 , … , μ N ] T {\displaystyle \ \mu =[\mu _{1},\dots ,\mu _{N}]^{T}} 及 N × M {\displaystyle N\times M} 矩阵 A {\displaystyle \ A} 满足 X = A Z + μ {\displaystyle \ X=AZ+\mu } . 存在 μ {\displaystyle \mu } 和一个对称半正定阵 Σ {\displaystyle \ \Sigma } 满足 X {\displaystyle \ X} 的特征函数 ϕ X ( u ; μ , Σ ) = e i μ T u − 1 2 u T Σ u {\displaystyle \phi _{X}\left(u;\mu ,\Sigma \right)=\mathrm {e} ^{i\mu ^{T}u-{\frac {1}{2}}u^{T}\Sigma u}} 如果 Σ {\displaystyle \ \Sigma } 是非奇异的,那么该分布可以由以下的概率密度函数来描述:[1] f x ( x 1 , … , x k ) = 1 ( 2 π ) k | Σ | e − 1 2 ( x − μ ) T Σ − 1 ( x − μ ) , {\displaystyle f_{\mathbf {x} }(x_{1},\ldots ,x_{k})={\frac {1}{\sqrt {(2\pi )^{k}|{\boldsymbol {\Sigma }}|}}}\mathrm {e} ^{-{\frac {1}{2}}({\mathbf {x} }-{\boldsymbol {\mu }})^{\mathrm {T} }{\boldsymbol {\Sigma }}^{-1}({\mathbf {x} }-{\boldsymbol {\mu }})},} 注意这里的 | Σ | {\displaystyle |\Sigma |} 表示协方差矩阵的行列式。 二元的情况 在二维非奇异的情况下(k = rank(Σ) = 2),向量 [X Y]′ 的概率密度函数为: f ( x , y ) = 1 2 π σ X σ Y 1 − ρ 2 e − 1 2 ( 1 − ρ 2 ) [ ( x − μ X σ X ) 2 − 2 ρ ( x − μ X σ X ) ( y − μ Y σ Y ) + ( y − μ Y σ Y ) 2 ] {\displaystyle f(x,y)={\frac {1}{2\pi \sigma _{X}\sigma _{Y}{\sqrt {1-\rho ^{2}}}}}\mathrm {e} ^{-{\frac {1}{2(1-\rho ^{2})}}\left[({\frac {x-\mu _{X}}{\sigma _{X}}})^{2}-2\rho ({\frac {x-\mu _{X}}{\sigma _{X}}})({\frac {y-\mu _{Y}}{\sigma _{Y}}})+({\frac {y-\mu _{Y}}{\sigma _{Y}}})^{2}\right]}} 其中 ρ 是 X 与 Y 之间的相关系数, σ X > 0 {\displaystyle \sigma _{X}>0} 且 σ Y > 0 {\displaystyle \sigma _{Y}>0} 。在这种情况下, μ = ( μ X μ Y ) , Σ = ( σ X 2 ρ σ X σ Y ρ σ X σ Y σ Y 2 ) . {\displaystyle {\boldsymbol {\mu }}={\begin{pmatrix}\mu _{X}\\\mu _{Y}\end{pmatrix}},\quad {\boldsymbol {\Sigma }}={\begin{pmatrix}\sigma _{X}^{2}&\rho \sigma _{X}\sigma _{Y}\\\rho \sigma _{X}\sigma _{Y}&\sigma _{Y}^{2}\end{pmatrix}}.} 参考文献 [1]UIUC, Lecture 21. The Multivariate Normal Distribution (页面存档备份,存于互联网档案馆), 21.5:"Finding the Density". Wikiwand in your browser!Seamless Wikipedia browsing. On steroids.Every time you click a link to Wikipedia, Wiktionary or Wikiquote in your browser's search results, it will show the modern Wikiwand interface.Wikiwand extension is a five stars, simple, with minimum permission required to keep your browsing private, safe and transparent.Wikiwand for ChromeWikiwand for EdgeWikiwand for Firefox
Every time you click a link to Wikipedia, Wiktionary or Wikiquote in your browser's search results, it will show the modern Wikiwand interface.Wikiwand extension is a five stars, simple, with minimum permission required to keep your browsing private, safe and transparent.