change default optim.method in km (#285)

* change default optim.method in km

Author: ja-thomas

R/makeMboLearner.R

@@ -10,6 +10,9 @@
 #' The nugget effect is set to 10^-8 * Var(y) on each model training.
 #' For mixed parameter spaces the function returns a random forest regression learner.
 #' The method to estimate the variance is the standard deviation of the bagged predictions.
+#' Instead of the default \code{"BFGS"} optimization method we use rgenoud (\code{"gen"}),
+#' which is a hybrid algorithm, to combine global search based on genetic algorithms and local search based on gradients.
+#' This may improve the model fit and will produce a constant surrogate model much less frequent.  
 #' You can override this setting in \code{...}.
 #' 
 #' @template arg_control
@@ -23,7 +26,7 @@
 makeMboLearner = function(control, fun, ...) {
   par.vals = list(...)
   if (!hasDiscrete(getParamSet(fun), include.logical = TRUE)) {
-    par.vals = insert(list(covtype = "matern5_2"), par.vals)
+    par.vals = insert(list(covtype = "matern5_2", optim.method = "gen"), par.vals)
     base.learner = makeLearner("regr.km", predict.type = "se")
     if (!control$filter.proposed.points) {
       warningf("filter.proposed.points is not set in the control object. This might lead to the 'leading minor of order ...' error during model fit.")