R/AR1Est.R

In FitAR: Subset AR Model Fitting

`AR1Est` <-
function(z, MeanValue=0){
stopifnot(length(z)>1)
n=length(z)
m <- MeanValue
mz <- rep(m,n)
a <- sum((z-mz)^2)
b <- sum((z[-1]-mz[-1])*(z[-n]-mz[-n]))
c <- sum((z[c(-1,-n)]-mz[c(-1,-n)])^2)
i <- complex(1,0,1)
x <- ((-16)*b^3+18*a*b*c+24*b^3*n-27*a*b*c*n-9*b*c^2*n-12*b^3*n^2+9*a*b*c*n^2+27*b*c^2*n^2+2*b^3*n^3-18*b*c^2*n^3)
y <- (-b^2*(-2+n)^2+3*c*(-1+n)*(-a-c*n))
f <- complex(1,x^2+4*y^3,0)
z <- (x+sqrt(f))^(1/3)
g <- x^2+4*y^3
z1 <- (x+(-g)^(1/2)*i)^(1/3)
part1 <- (n-2)*b/(3*c*(n-1))
part2 <- (1-sqrt(3)*i)*y/(3*2^(2/3)*c*(n-1)*z)
part3 <- (1+sqrt(3)*i)*z/(6*2^(1/3)*c*(n-1))
Re(part1+part2-part3) 
}