R: Alternating Optimization.

ao {ao}

R Documentation

Alternating Optimization.

Description

This function performs alternating optimization on the function f.

Usage

ao(
  f,
  partition,
  initial = 0,
  iterations = 10,
  tolerance = 1e-06,
  minimize = TRUE,
  progress = FALSE,
  plot = TRUE
)

Arguments

`f`	An object of class `ao_f`, i.e. the output of `set_f`.
`partition`	A list of vectors of parameter indices 1,...,n of the function. For example, choosing `partition = list(1, 2)` as in the example optimizes each parameter separately, while choosing `partition = list(1:2)` leads to joint optimization. Parameter indices can be members of multiple subsets.
`initial`	A vector of length `f$npar` of initial parameter values. Per default, the algorithm is initialized at the origin.
`iterations`	The number of iterations through all subsets.
`tolerance`	A non-negative numeric value. The function terminates prematurely if the euclidean distance between the current solution and the one from the last iteration is smaller than `tolerance`.
`minimize`	If `TRUE`, minimization, if `FALSE`, maximization.
`progress`	If `TRUE`, progress is printed.
`plot`	If `TRUE`, the parameter updates are plotted.

Details

This function depends on optimx.

Value

An object of class ao, which is a list of

optimum, the optimal value,
estimate, the parameter vector that yields the optimum,
sequence, a data frame of the estimates in the single iterations,
time, the total estimation time in seconds.

Examples

himmelblau <- function(x) (x[1]^2 + x[2] - 11)^2 + (x[1] + x[2]^2 - 7)^2
f <- set_f(f = himmelblau, npar = 2, lower = -5, upper = 5)
ao(f = f, partition = list(1, 2))

[Package ao version 0.2.1 Index]