ABSTRACT
Y(t) = nt + g(t/T) +noise (5.15) t = 0, ... , T - I with a a parameter related to diffusion motion and g(-) a step function. The model in Eq. (5.15) will be approximated by
for some 11. Because of the presence of the term at in Eq. (5.16) the analysis in the present case is not so immediate, but still all that one needs are local means. The least squares estimates are obtained by regression of Y on the cPnk(t/T) and on I made orthogonal to the cPnk· Further details on the fitting are given in the Appendix to this chapter.
