可以计算仿真，测试成功 %for i=1:n sample=samplex(nn:n,:); t1=sum(sample); t2=mean(sample); t3=std(sample,0); %ratio p=samplex(:,2)./samplex(:,3); e=mean(p); va=std(p,1); t4=(1-e)*e/va^2; a=e*(t4-1) b=(1-e)*(t4-1) %pa=betafit(sam(:,1),0.05); %pas=gamfit(sam(:,2),1) %a=pas(1) %b=1/pas(2)