This package is an implementation of the variable neighborhood trust region search algorithm Bierlaire et al. (2009) "A Heuristic for Nonlinear Global Optimization".
-
Install the latest package version via running
install.packages("vntrs")
in your R console. -
Specify an R function
f
that computes value, gradient, and Hessian of the function to be optimized and returns them as a named list with elementsvalue
,gradient
, andhessian
. -
Call
vntrs::vntrs(f = f, npar = npar, minimize = minimize)
, where
-
npar
is the number of parameters off
and -
minimize
is a boolean, determining whetherf
should be minimized (minimize = TRUE
, the default) or maximized (minimize = FALSE
).
Optionally, you can tune the algorithm by specifying the named list controls
and passing it to search
. See the help file of help("vntrs")
for details.