Here is an example implementation of Auto Adaptive M-Estimation (AAME) in R, finding the mean of a data and the error density. The data could be with a heavy-tailed and/or comtaminated error.

1. Run this in R: AAME_onesamp functions

2. Load necessary library:

   library(quadprog)
   

3. Generate a data, then fit the model:

   set.seed(1)
   mu <- 100
   y <- mu + rt(500,df=2)
   fit <- onesamp(y)
   

4. Estimated mean of the data, and its confidence interval:

   fit$muhat 
   fit$confidence.intervals 
   

5. Plot the estimated error density function:

   xp <- seq(-max(fit$knots),max(fit$knots),length=100)
   hist(y,xaxt='n',yaxt='n',xlab='',ylab = '',main="",probability = T,nclass = 100)
   lines(fit$fhat(xp)~c(xp+fit$muhat),lwd=2)