numerical_deriv: Compute numerical derivatives

In xzhaopsy/MIRT: Multidimensional Item Response Theory

Description

Compute numerical derivatives using forward/backword difference, central difference, or Richardson extropolation.

Usage

numerical_deriv(par, f, ..., delta = 1e-05, gradient = TRUE,
  type = "forward")

Arguments

par	a vector of parameters
f	the objective function being evaluated
...	additional arguments to be passed to f and the numDeriv package when the Richardson type is used
delta	the term used to perturb the f function. Default is 1e-5
gradient	logical; compute the gradient terms? If FALSE then the Hessian is computed instead
type	type of difference to compute. Can be either 'forward' for the forward difference, 'central' for the central difference, or 'Richardson' for the Richardson extropolation. Backword difference is acheived by supplying a negative delta value

Author(s)

Phil Chalmers [e-mail]

Examples

## Not run: 
f <- function(x) 3*x[1]^3 - 4*x[2]^2
par <- c(3,8)

# grad = 9 * x^2 , -8 * y
(actual <- c(9 * par[1]^2, -8 * par[2]))
numerical_deriv(par, f, type = 'forward')
numerical_deriv(par, f, type = 'central')
numerical_deriv(par, f, type = 'Richardson')

# hessian = h11 -> 18 * x, h22 -> -8, h12 -> h21 -> 0
(actual <- matrix(c(18 * par[1], 0, 0, -8), 2, 2))
numerical_deriv(par, f, type = 'forward', gradient = FALSE)
numerical_deriv(par, f, type = 'central', gradient = FALSE)
numerical_deriv(par, f, type = 'Richardson', gradient = FALSE)


## End(Not run)

xzhaopsy/MIRT documentation built on May 29, 2019, 12:42 p.m.