vector of values, in nonlinear regression residuals are used.
Value
Minimized Sum of rho function, by reweighting.
Details
Robust M-estimate of scale using rodust loss function rho, to acheive 50% breakdown. Tunning constant k1 should be stored in nl.form object of the robust function.
This is called from nlmest.NLM function to derive MM.estimate.
References
Yohai (1987) Annals, Stromberg (1993) JASA. GKS 2 June 99