Function to calculate du/dx in the gradient of d(x, Xi), will be used in ForLion_MLM_func() function, details see Appendix C in Huang, Li, Mandal, Yang (2024)
dprime_func_self(
xi,
bvec,
h.func,
h.prime,
inv.F.mat,
Ux,
link = "continuation",
k.continuous
)
dU/dx in the gradient of sensitivity function d(x, Xi)
a vector of design point
parameter of the multinomial logistic regression model
function, is used to transfer xi to model matrix (e.g. add interaction term, add intercept)
function, is used to find dX/dx
inverse of F_Xi matrix, inverse of fisher information of current design w/o new point
U_x matrix in the algorithm, get from Fi_MLM_func() function
multinomial link function, default is"continuation", other choices "baseline", "cumulative", and "adjacent"
number of continuous factors