PAL_PFML_fit uses a brute-force search for initial value.
PAL_PFML_SearchGrid_Demo: PAL_PFML_Fit performs a brute-force search through parameter space in order to find initial values for Nelder-Mead iterative search. This is important to avoid
local maxima in the likelihood function.
Figure 1. When location and slope are the only free parameters, PAL_PFML_Fit searches in 2d parameter space ....
Figure 2. ... but when the guess and/or lapse rate parameters are free, these are included in the brute-force search. Failing to do this might result in finding a local maximum (see our understanding fitting page).
Figure 1. When location and slope are the only free parameters, PAL_PFML_Fit searches in 2d parameter space ....
Figure 2. ... but when the guess and/or lapse rate parameters are free, these are included in the brute-force search. Failing to do this might result in finding a local maximum (see our understanding fitting page).