R help - Oct 2003 - 'optim' and extra argument to the objective function

Hello,


I'd like to use optim, and give extra arguments to the objective
function. The man page says that the '...' should let one do it,
but I have a hard time to understand how.

Example:

x <- 1:10
y <- rnorm(10)
cost.f <- function(par, x, y) {
 A <- par[1]
 cost <- sum( (log(A*x) - log(y))^2)
 return(cost)
}

optim(3, cost.f, x, y)
## returns:
Error in pmatch(x, table, duplicates.ok) : 
	argument is not of mode character
## uh, uh... may the problem is with the argument matching ?
optim(3, cost.f, method="BFGS", x, y)
## returns:
Error in log(A * x) : Argument "x" is missing, with no default
In addition: Warning message: 
bounds can only be used with method L-BFGS-B in: optim(3, cost.f, method =
"BFGS", x, y)

Any suggestion ?


L.


PS: this was done with both R-1.7.1-patched and R-1.8.0-patched for Linux.

Have you tried the following: 

optim(3, cost.f, x=x, y=y)

      In the form given below, R interprets you call as equivalent to 
the following: 

optim(par=3, fn=cost.f, gr=x, method=y)

      Arguments will get passed via "..." when they have names that
are
different from names in the official argument list. 

hope this helps.  spencer graves

Laurent Gautier wrote:
>Hello,
>
>
>I'd like to use optim, and give extra arguments to the objective
>function. The man page says that the '...' should let one do it,
>but I have a hard time to understand how.
>
>Example:
>
>x <- 1:10
>y <- rnorm(10)
>cost.f <- function(par, x, y) {
> A <- par[1]
> cost <- sum( (log(A*x) - log(y))^2)
> return(cost)
>}
>
>optim(3, cost.f, x, y)
>## returns:
>Error in pmatch(x, table, duplicates.ok) : 
>	argument is not of mode character
>## uh, uh... may the problem is with the argument matching ?
>optim(3, cost.f, method="BFGS", x, y)
>## returns:
>Error in log(A * x) : Argument "x" is missing, with no default
>In addition: Warning message: 
>bounds can only be used with method L-BFGS-B in: optim(3, cost.f, method =
"BFGS", x, y)
>
>Any suggestion ?
>
>
>L.
>
>
>PS: this was done with both R-1.7.1-patched and R-1.8.0-patched for Linux.
>
>______________________________________________
>R-help at stat.math.ethz.ch mailing list
>https://www.stat.math.ethz.ch/mailman/listinfo/r-help
>  
>

On Mon, 20 Oct 2003, Laurent Gautier wrote:
>
> Hello,
>
>
> I'd like to use optim, and give extra arguments to the objective
> function. The man page says that the '...' should let one do it,
> but I have a hard time to understand how.
>
> Example:
>
> x <- 1:10
> y <- rnorm(10)
> cost.f <- function(par, x, y) {
>  A <- par[1]
>  cost <- sum( (log(A*x) - log(y))^2)
>  return(cost)
> }
>
> optim(3, cost.f, x, y)

You need
  optim(3, cost.f, x=x, y=y)

As always, if you don't specify a tag for an argument it gets matched
positionally, in this case to the gr and method arguments of optim().

	-thomas

On Mon, 20 Oct 2003, Laurent Gautier wrote:
> I'd like to use optim, and give extra arguments to the objective
> function. The man page says that the '...' should let one do it,
> but I have a hard time to understand how.
Try reading up on argument matching, then.  E.g. in `S Programming'
(see the R FAQ).
> Example:
> 
> x <- 1:10
> y <- rnorm(10)
> cost.f <- function(par, x, y) {
>  A <- par[1]
>  cost <- sum( (log(A*x) - log(y))^2)
>  return(cost)
> }
> 
> optim(3, cost.f, x, y)
> ## returns:
> Error in pmatch(x, table, duplicates.ok) : 
> 	argument is not of mode character
> ## uh, uh... may the problem is with the argument matching ?
> optim(3, cost.f, method="BFGS", x, y)
> ## returns:
> Error in log(A * x) : Argument "x" is missing, with no default
> In addition: Warning message: 
> bounds can only be used with method L-BFGS-B in: optim(3, cost.f, method =
"BFGS", x, y)
> 
> Any suggestion ?
optim(3, cost.f, method="BFGS", x=x, y=y)

except that some component of y is negative with high probability ....

You do have to *name* arguments in ... -- that's standard S/R syntax.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595