init_bounds_optim: Setting of Optimization Bounds and Initial Values

In varycoef: Modeling Spatially Varying Coefficients

Setting of Optimization Bounds and Initial Values

Description

Sets bounds and initial values for optim by extracting potentially given values from SVC_mle_control and checking them, or calculating them from given data. See Details.

Usage

init_bounds_optim(control, p, q, id_obj, med_dist, y_var, OLS_mu)

Arguments

control	(SVC_mle_control output, i.e. list)
p	(numeric(1)) Number of fixed effects
q	(numeric(1)) Number of SVCs
id_obj	(numeric(2*q+1+q)) Index vector to identify the arguments of objective function.
med_dist	(numeric(1)) Median distance between observations
y_var	(numeric(1)) Variance of response y
OLS_mu	(numeric(p)) Coefficient estimates of ordinary least squares (OLS).

Details

If values are not provided, then they are set in the following way. Let d be the median distance med_dist, let s^2_y be the variance of the response y_var, and let b_j be the OLS coefficients of the linear model. The computed values are given in the table below.

Parameter	Lower bound	Initial Value	Upper Bound
Range	d/1000	d/4	10 d
Variance	0	s^2_y/(q+1)	10s^2_y
Nugget	10^{-6}	s^2_y/(q+1)	10s^2_y
Mean j	-Inf	b_j	Inf

Value

A list with three entries: lower, init, and upper.

Author(s)

Jakob Dambon

varycoef documentation built on Sept. 18, 2022, 1:07 a.m.