The distribution of the sum of independent gamma random variables (Q120481): Difference between revisions

Let \(\{X_ j\}\), \(j=1,...,n\), be a set of mutually independent gamma variates with parameters \(\alpha_ j>0\) and \(\beta_ j>0\). Then the density of \(X_ j\) is given by \(f_ j(x_ j)=x_ j^{\alpha_{j- 1}}e^{-x_ j/\beta_ j}[\beta_ j^{\alpha_ j}\Gamma (\alpha_ j)],\) \(x_ j>0\) and \(f_ j(x_ j)=0\) elsewhere. In this note the exact density of the sum \(y=x_ 1+...+x_ n\) is given and this density can be expressed as \[ g(y)=C\sum^{\infty}_{k=0}\delta_ ky^{\rho +k- 1}e^{-y/\beta_ 1}[\Gamma (\rho +k)\beta_ 1^{\rho +k}],\quad y>0,\quad and\quad 0\quad elsewhere, \] where \(\rho =\sum^{n}_{j=1}\alpha_ j\), \(C=\prod^{n}_{j=1}(\beta_ 1/\beta_ j)^{\alpha_ j}\), \((\beta_ 1=\min_{j}\beta_ j)\) and \(\delta_{k+1}=(k+1)^{-1}\sum^{k+1}_{j=1}j\gamma_ j\delta_{k+1-j}\) \((k=0,1,2,...)\).