Skip to contents

Uses the hybrid approach: takes an analytical (or AD) gradient function and computes its Jacobian using forward-mode AD to get the Hessian.

Usage

ad_hessian(score_fn, par)

Arguments

score_fn

A function score_fn(par) returning the gradient vector

par

Numeric vector of parameters

Value

Hessian matrix (p x p)