首先要定义你要构造你的似然函数，这个是先验的，是你根据分布和结合理论知识所选择的模型，比如一个分布差不多符合正态分布，那么你就可以构造一个正态分布的似然函数：
normal <- function(theta,x){
  mu <- theta[1]
  sigma2 <- theta[2]
  n <- length(x)
  logL <- -0.5*n*log(2*pi)-0.5*n*log(sigma2)-(1/(2*sigma2))*sum((x-mu)**2)
  return (-logL)
}
然后就是把你的数据，似然函数，和一个你认为最接近结果的起始值一股脑放进optim就可以了
result <- optim(c(0,1),normal,x=x)

snakepointid explained very well

> loglik <- function(param) {
+   p <- param[1]
+   k1 <- param[2]
+   k2<-param[3]
+   u1<-param[4]
+   u2<-param[5]
+   ll <-  sum(log(p*exp(k1)*cos(x-u1)/(2*pi*besselj(x))+(1-p)*exp(k2)*cos(x-u2)/(2*pi*besselj(x))))
+   ll
+ }
> res <- maxLik(loglik, start=c(0,0,0,0,0)) # use 'wrong' start values
Show Traceback

Rerun with Debug
Error in Math.data.frame(x - u1) :
  non-numeric variable in data frame: chushichulizhong In addition: Warning messages:
1: In Ops.factor(left, right) : ‘-’ not meaningful for factors
2: In Ops.factor(left, right) : ‘-’ not meaningful for factors
3: In Ops.factor(left, right) : ‘-’ not meaningful for factors
请问这个代码是哪错了，极大似然估计5个参数的