1
用Gamma分布构造Dirichlet分布
令是相互独立的随机变量,每个变量的伽玛分布参数为表示,与X1,…,Xk+1X1,…,Xk+1X_1,\dots,X_{k+1}αi,i=1,2,…,k+1αi,i=1,2,…,k+1\alpha_i,i=1,2,\dots,k+1Yi=XiX1+⋯+Xk+1,i=1,…,kYi=XiX1+⋯+Xk+1,i=1,…,kY_i=\frac{X_i}{X_1+\cdots+X_{k+1}},i=1,\dots,kDirichlet(α1,α2,…,αk;αk+1)Dirichlet(α1,α2,…,αk;αk+1)\text{Dirichlet}(\alpha_1,\alpha_2,\dots,\alpha_k;\alpha_{k+1}) 的联合PDF。然后找到关节(y_1,\ dots,Y_ {k + 1})的 pdf文件,我找不到jacobian即J(\ frac {x_1,\ dots,x_ {k + 1}} {y_1,\ dots,y_ {k + 1} })(Ý1,...,ÿķ+1)Ĵ(X1,...,X ķ + 1(X1,…,Xk+1)=e−∑k+1i=1xixα1−11…xαk+1−1k+1Γ(α1)Γ(α2)…Γ(αk+1)(X1,…,Xk+1)=e−∑i=1k+1xix1α1−1…xk+1αk+1−1Γ(α1)Γ(α2)…Γ(αk+1)(X_1,\dots,X_{k+1})=\frac{e^{-\sum_{i=1}^{k+1}x_i}x_1^{\alpha_1-1}\dots x_{k+1}^{\alpha_{k+1}-1}}{\Gamma(\alpha_1)\Gamma(\alpha_2)\dots \Gamma(\alpha_{k+1})}(Y1,…,Yk+1)(Y1,…,Yk+1)(Y_1,\dots,Y_{k+1})J(x1,…,xk+1y1,…,yk+1)J(x1,…,xk+1y1,…,yk+1)J(\frac{x_1,\dots,x_{k+1}}{y_1,\dots,y_{k+1}})