On 09-02-2013, at 21:08, Axel Urbiz <axel.ur...@gmail.com> wrote:
> Dear List,
>
> I'm new in R. I'm trying to solve a simple constrained optimization
> problem.
>
> Essentially, let's say I have a matrix as in the object 'mm' inside the
> function below. My objective function should have a matrix of parameters,
> one parameter for each element 'mm' (4 in this case). The problem is to
> select the value of the parameters to maximize the function 'ff' s.t. (1)
> each parameter being either 0 or 1, and (2) the sum of each row in the
> parameter matrix should equal 1.
>
> I'm using the function constrOptim as shown below, but obviously I'm not
> doing things right. Any help is much appreciated.
>
> ff <- function (x) {
>
> mm <- matrix(c(10, 25, 5, 10), 2, 2)
> matx <- matrix(NA, 2, 2)
>
> for (i in 1:nrow(x)) {
> for (j in 1:ncol(x)) {
>
> matx[i, j] <- x[i, j]
> }
>
> }
>
> -sum(apply(mm ^ matx, 1, prod))
>
> }
>
> constrOptim(theta = c(0, 0, 0, 0), f = ff, ui=rbind(c(1, 1),
> c(1, 1)),
> ci=c(1, 1))
1. your parameter vector has length(4)
2. the ui matrix is 2x2 so ui %*% theta can't be done
3. ui should be something like rbind(c(1,1,0,0),c(0,0,1,1))
4. you haven't specified the gradient function (grad); should be a function or
NULL
5. in your function ff on entry x is a vector. It can be simplified to
ff <- function (x) {
mm <- matrix(c(10, 25, 5, 10), 2, 2)
matx <- matrix(x,2,2)
-sum(apply(mm ^ matx, 1, prod))
}
then
constrOptim(theta = c(1, 1, 1, 1), f = ff, grad=NULL, ui=rbind(c(1,
1,0,0),c(0,0,1, 1)),ci=c(1, 1))
gives:
$par
[1] -63.16203 120.71655 -135.39084 139.50023
$value
[1] -1.797693e+308
$counts
function gradient
501 NA
$convergence
[1] 0
$message
NULL
$outer.iterations
[1] 2
$barrier.value
[1] 0
which is probably not at all what you are looking for.
Look in the CRAN Task View for Optimization for possible options for your
optimization problem.
Berend
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
