nlmixr2Hess {nlmixr2est}

Calculate Hessian

Description

Unlike 'stats::optimHess' which assumes the gradient is accurate, nlmixr2Hess does not make as strong an assumption that the gradient is accurate but takes more function evaluations to calculate the Hessian. In addition, this procedures optimizes the forward difference interval by nlmixr2Gill83

Usage

nlmixr2Hess(par, fn, ..., envir = parent.frame())

Arguments

par

Initial values for the parameters to be optimized over.

fn

A function to be minimized (or maximized), with first argument the vector of parameters over which minimization is to take place. It should return a scalar result.

...

Extra arguments sent to nlmixr2Gill83

envir

an environment within which to evaluate the call. This will be most useful if what is a character string and the arguments are symbols or quoted expressions.

Details

If you have an analytical gradient function, you should use 'stats::optimHess'

Value

Hessian matrix based on Gill83

Author(s)

Matthew Fidler

See Also

nlmixr2Gill83, optimHess

Examples

 func0 <- function(x){ sum(sin(x))  }
 x <- (0:10)*2*pi/10
 nlmixr2Hess(x, func0)

fr <- function(x) {   ## Rosenbrock Banana function
    x1 <- x[1]
    x2 <- x[2]
    100 * (x2 - x1 * x1)^2 + (1 - x1)^2
}
grr <- function(x) { ## Gradient of 'fr'
    x1 <- x[1]
    x2 <- x[2]
    c(-400 * x1 * (x2 - x1 * x1) - 2 * (1 - x1),
       200 *      (x2 - x1 * x1))
}

h1 <- optimHess(c(1.2,1.2), fr, grr)

h2 <- optimHess(c(1.2,1.2), fr)

## in this case h3 is closer to h1 where the gradient is known

h3 <- nlmixr2Hess(c(1.2,1.2), fr)