Rho (weight) function as in the robust literature. It
is called by some routines in
lmRobMM, but not supposed to be called by the
users directly.
rho.weight(svals,ipsi,xk)
ipsi is
1, the optimal loss (weight) function will be used;
if
ipsi is
2, the rescaled bisquare loss (weight) function will be used.
Rho function.
Marazzi, A. (1993): Algorithms, routines and S functions for robust statistics, Wadsworth & Brooks/Cole, Pacific Grove, CA.