# EXAMPLE 4: Introduction to Statistics for Astrophysicists
# SS 2012
# JOCHEN WELLER
# The Lighthouse Problem
alpha0 = 1 #km true position at coastline
beta = 1 #km position out at see
# Sequence of positions (use already prior range from -10 to 10 km)
alpha=seq(-10,10,by = 0.05)
# number of trials
N=512
# generate random sample distributed according to a Cauchy distribution (see lecture)
# with N trial
pos=rcauchy(N,alpha0,beta)
# prior distribution
# uniform between -10 and +10 km
prior = 1.0/20.0
sum = 0.0
i = 1
logpost <- function(alpha) { while (i<=N){sum=sum+log(beta^2+(pos[i]-alpha)^2)
i=i+1 }; sum}
#find minimum of log likelihood
lmin=min(logpost(alpha))
# normalized posterior
post <- function(alpha) exp(-logpost(alpha)+lmin)
# plot posterior
par(font.lab=2)
par(font.axis=2)
plot(alpha,post(alpha),type='l', xlab=expression(paste(alpha,"(km)")),ylab=expression(prob(~alpha~"|{"~x[k]~"}",~beta,I)), lwd=2, main=paste("N=",N),xlim=c(-12,12),ylim=c(0,1.2))
# plot points of samples at vertical position 1.1
normpoint = (pos-pos)+1.1
points(pos,normpoint)
# vertical line at mean of sample
abline(v=mean(pos),lwd=2,lty=2)