mmd.nl: MMD objective function for non-linear models

In estsyawo/bayesprdopt: Useful Functions and Algorithms

MMD objective function for non-linear models

Description

mmd.nl computes the objective function of the MMD estimator for a non-linear model

Usage

mmd.nl(theta, U.fun, Z = NULL, Z.m = NULL, sc = 1)

Arguments

theta	parameter vector
U.fun	user-written function of the disturbance function; evaluates to an n x 1 vector
Z	n x pz matrix of instruments; should be supplied in case Z.m is null
Z.m	Euclidean matrix from instrument matrix Z - optional
sc	adjusts the objective function - positive for minimisation, negative for maximisation

Value

function value

Examples

## Generate data and run MMD regression
n=50; set.seed(12); X = rnorm(n); Y = X + (X^2-1)/sqrt(2)
U.fun=function(theta) Y^2-2*Y*(X*theta)-(X*theta)^2
mmd.nl(1,U.fun,Z=X)

estsyawo/bayesprdopt documentation built on April 2, 2024, 2:18 p.m.