src/library/stats/man/nlminb.Rd

% File src/library/stats/man/nlminb.Rd
% Part of the R package, https://www.R-project.org
% Copyright 1995-2017 R Core Team
% Distributed under GPL 2 or later

\name{nlminb}
\alias{nlminb}
\title{Optimization using PORT routines }
\description{
  Unconstrained and box-constrained optimization using PORT routines.

  For historical compatibility.
}
\usage{
nlminb(start, objective, gradient = NULL, hessian = NULL, \dots,
       scale = 1, control = list(), lower = -Inf, upper = Inf)
}
\arguments{
  \item{start}{
    numeric vector, initial values for the parameters to be optimized.
  }
  \item{objective}{
    Function to be minimized.  Must return a scalar value.  The first
    argument to \code{objective} is the vector of parameters to be
    optimized, whose initial values are supplied through \code{start}.
    Further arguments (fixed during the course of the optimization) to
    \code{objective} may be specified as well (see \code{\dots}).
  }
  \item{gradient}{
    Optional function that takes the same arguments as \code{objective} and
    evaluates the gradient of \code{objective} at its first argument.  Must
    return a vector as long as \code{start}.
  }
  \item{hessian}{
    Optional function that takes the same arguments as \code{objective} and
    evaluates the hessian of \code{objective} at its first argument.  Must
    return a square matrix of order \code{length(start)}.  Only the
    lower triangle is used.
  }
  \item{\dots}{Further arguments to be supplied to \code{objective}.}
  \item{scale}{See PORT documentation (or leave alone).}
  \item{control}{A list of control parameters. See below for details.}
  \item{lower, upper}{
    vectors of lower and upper bounds, replicated to be as long as
    \code{start}.  If unspecified, all parameters are assumed to be
    unconstrained.
  }
}
\details{
  Any names of \code{start} are passed on to \code{objective} and where
  applicable, \code{gradient} and \code{hessian}.  The parameter vector
  will be coerced to double.

  %% The PORT documentation is at
  %% \url{http://netlib.bell-labs.com/cm/cs/cstr/153.pdf}.

  If any of the functions returns \code{NA} or \code{NaN} this is an
  error for the gradient and Hessian, and such values for function
  evaluation are replaced by \code{+Inf} with a warning.
}
% see PR#15052.
\value{
  A list with components:
  \item{par}{The best set of parameters found.}
  \item{objective}{The value of \code{objective} corresponding to \code{par}.}
  \item{convergence}{An integer code. \code{0} indicates successful
    convergence.
  }
  \item{message}{
    A character string giving any additional information returned by the
    optimizer, or \code{NULL}. For details, see PORT documentation.
  }
  \item{iterations}{Number of iterations performed.}
  \item{evaluations}{Number of objective function and gradient function evaluations}
}
\section{Control parameters}{
  Possible names in the \code{control} list and their default values
  are:
  \describe{
    \item{\code{eval.max}}{Maximum number of evaluations of the objective
      function allowed.  Defaults to 200.}% MXFCAL
    \item{\code{iter.max}}{Maximum number of iterations allowed.
      Defaults to 150.}% MXITER
    \item{\code{trace}}{The value of the objective function and the parameters
      is printed every trace'th iteration.  Defaults to 0 which
      indicates no trace information is to be printed.}
    \item{\code{abs.tol}}{Absolute tolerance.  Defaults
      to 0 so the absolute convergence test is not used.  If the objective
      function is known to be non-negative, the previous default of
      \code{1e-20} would be more appropriate.}% AFCTOL  31
    \item{\code{rel.tol}}{Relative tolerance.  Defaults to
      \code{1e-10}.}% RFCTOL  32
    \item{\code{x.tol}}{X tolerance.  Defaults to \code{1.5e-8}.}% XCTOL  33
    \item{\code{xf.tol}}{false convergence tolerance.  Defaults to
      \code{2.2e-14}.}% XFTOL  34
    \item{\code{step.min, step.max}}{Minimum and maximum step size.  Both
      default to \code{1.}.}% LMAX0 35  /  LMAXS 36
    \item{sing.tol}{singular convergence tolerance; defaults to
      \code{rel.tol}.}% SCTOL  37
    \item{scale.init}{...}% DINIT  38
    \item{diff.g}{an estimated bound on the relative error in the
      objective function value.}% ETA0  42
  }
}
\source{
  \url{http://www.netlib.org/port/}
}
\references{
  David M. Gay (1990),
  Usage summary for selected optimization routines.
  Computing Science Technical Report 153, AT&T Bell Laboratories, Murray
  Hill.
}  
\author{
  \R port: Douglas Bates and Deepayan Sarkar.

  Underlying Fortran code by David M. Gay
}
\seealso{
  \code{\link{optim}} (which is preferred) and \code{\link{nlm}}.

  \code{\link{optimize}} for one-dimensional minimization and
  \code{\link{constrOptim}} for constrained optimization.
}
% Lots of near-zeros, platform-specific results.
\examples{\donttest{
x <- rnbinom(100, mu = 10, size = 10)
hdev <- function(par)
    -sum(dnbinom(x, mu = par[1], size = par[2], log = TRUE))
nlminb(c(9, 12), hdev)
nlminb(c(20, 20), hdev, lower = 0, upper = Inf)
nlminb(c(20, 20), hdev, lower = 0.001, upper = Inf)

## slightly modified from the S-PLUS help page for nlminb
# this example minimizes a sum of squares with known solution y
sumsq <- function( x, y) {sum((x-y)^2)}
y <- rep(1,5)
x0 <- rnorm(length(y))
nlminb(start = x0, sumsq, y = y)
# now use bounds with a y that has some components outside the bounds
y <- c( 0, 2, 0, -2, 0)
nlminb(start = x0, sumsq, lower = -1, upper = 1, y = y)
# try using the gradient
sumsq.g <- function(x, y) 2*(x-y)
nlminb(start = x0, sumsq, sumsq.g,
       lower = -1, upper = 1, y = y)
# now use the hessian, too
sumsq.h <- function(x, y) diag(2, nrow = length(x))
nlminb(start = x0, sumsq, sumsq.g, sumsq.h,
       lower = -1, upper = 1, y = y)

## Rest lifted from optim help page

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))
}
nlminb(c(-1.2,1), fr)
nlminb(c(-1.2,1), fr, grr)


flb <- function(x)
    { p <- length(x); sum(c(1, rep(4, p-1)) * (x - c(1, x[-p])^2)^2) }
## 25-dimensional box constrained
## par[24] is *not* at boundary
nlminb(rep(3, 25), flb, lower = rep(2, 25), upper = rep(4, 25))
## trying to use a too small tolerance:
r <- nlminb(rep(3, 25), flb, control = list(rel.tol = 1e-16))
stopifnot(grepl("rel.tol", r$message))
}}
\keyword{optimize}