acp: Optimization using an iterative hill-climbing algorithm
In matsukik/mrsat: Multiple Response Speed Accuracy Tradeoff

Description Usage Arguments Details Value Author(s) References See Also Examples

Box-constrained optimization using an iterative hill-climbing algorithm

1	acp(start, objective, ..., control = list(), lower, upper)

`start`	Initial staring values for the parameters to be optimized
`objective`	A function to be minimized. The fist argument to the function should be the vector of parameters to be optimized. The function should return a scalar value.
`...`	Further arguments to be supplied to `objective`.
`control`	A list of control parameters. See 'Details' for more information.
`lower, upper`	Vectors, replicated to be as long as start. Lower and upper bounds of the parameter space.

The control argument is a list that can supply any of the following components:

trace: logical. If true, tracing information on the progress of the optimization is produced.
maxit: The maximum number of iterations. Defaults to 1000
lmax: The number of stepping cycles per iteration. Defaults to 20
strict: logical. If TRUE, iteration continues until all the values of stepsize are less than the value of abs.tol. Defaults to FALSE
abs.tol: The absolute convergence tolerance. Defaults to .Machine$double.eps. Relevant only when strict=TURE
auto.correct: logical indicating wheather to adjust the starting value with warning when value is equal to upper or lower bound (which may lead to convergence error). Defaults to TRUE

A list containing:

`par`	The best set of parameters found through iterative searching.
`convergence`	An integer indicating success or possible error. 0 indicates successful completion.
`iterations`	The number of iterations took took to converge.
`value`	The value of `objective` with `par`.
`trace`	If argument `control(trace=TRUE)`, a dataframe containing the values and pars at each iteration is returened. Otherwise, `NULL`

R port: Kazunaga Matsuki

original Fortran code by Unknown

Chandler, J.P. (1969). Subroutine STEPIT – finds local minimum of a smooth function of several parameters. Behavioral Science, 14, 81-82.

Judd, C. M., & McClelland, G. H. (1989). Data analysis: A model-comparison approach. San Diego: Harcourt Brace Jovanovich.

optim, nlm, nlminb

## example taken from nlmimb
x <- rnbinom(100, mu = 10, size = 10)
hdev <- function(par)
    -sum(dnbinom(x, mu = par[1], size = par[2], log = TRUE))
    
nlminb(c(20, 20), hdev, lower = 0.001, upper = Inf) 

## acp produce comparable results
acp(c(20, 20), hdev, lower = 0.001, upper = Inf)