R的优化

Question

我是R的新手,我的优化问题.

d <- c(0, 9.017, -9.017, 0, 9.017, 0, -8.579, -7.849, 0, 0, -7.849, 
    -9.017, 0, -7.849, -7.849, 0, 0, 0, 8.579, 1.168, 8.579, 8.579, 
    -7.849, 0.729, 8.579, 9.017, 0, -0.438)

x <- c(0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 1, 1, 0, 
    1, 1, 0, 1, 1, 1, 0, 1)

log.like<-function (sigma){ 
    theta = pnorm(d,mean=0,sd=sigma) 
    logl = sum(log((theta ^ x) * ((1-theta)^(1-x))))
    -logl
}

optim(0,fn=log.like,method="L-BFGS-B",lower=0,upper=1)

它给了我以下错误:

 Error in optim(0, fn = log.like, method = "L-BFGS-B", lower = 0, upper = 1) : 
  L-BFGS-B needs finite values of 'fn'

Answer 1

> log.like(0)
[1] Inf

您使用0作为起始值,但为您的函数提供无限值.这就是函数抱怨的原因.选择一个实际适当的起始值,它应该工作正常.我不明白你为什么要为参数设置上限1.您可能也希望增加它.