Clash between additional function and optimizer arguments in `apply_optimizer()`

[loelschlaeger]

In

f <- function(x, control, for_f) {
  print(for_f)
  sum(x^2)
}
apply_optimizer(optimizer_optim(), f, 1:3, control = list("something" = 1), for_f = "hi")
#> Warning message:
#> In (function (par, fn, gr = NULL, ..., method = c("Nelder-Mead",  :
#>   unbekannte Namen in control: something

both arguments control and for_f should be passed to f, but only for_f actually is. That is because control instead is passed to optimizer_optim(). However, additional arguments for the optimizer should not be passed via the ... argument in apply_optimizer() but rather in define_optimizer().

Check implementation of apply_optimizer().

[loelschlaeger]

It turned out, sometimes one still may want to supply arguments to the optimizer via apply_optimizer() (EM example). That was no longer possible with the above fix, so I reverted.

[loelschlaeger]

But this argument clash is a problem when optimizing with optim() directly, anyway. So it is not a bug of {optimizeR}, rather solving this clash would be a feature. Maybe try defining function arguments directly via setting formals(objective)?