OptimizerGridSearch
class that implements grid search. The grid is
constructed as a Cartesian product over discretized values per parameter, see
paradox::generate_design_grid()
. The points of the grid are evaluated in a
random order.
In order to support general termination criteria and parallelization, we
evaluate points in a batch-fashion of size batch_size
. Larger batches mean
we can parallelize more, smaller batches imply a more fine-grained checking
of termination criteria.
This Optimizer can be instantiated via the dictionary
mlr_optimizers or with the associated sugar function opt()
:
mlr_optimizers$get("grid_search") opt("grid_search")
resolution
integer(1)
Resolution of the grid, see paradox::generate_design_grid()
.
param_resolutions
named integer()
Resolution per parameter, named by parameter ID, see
paradox::generate_design_grid()
.
batch_size
integer(1)
Maximum number of points to try in a batch.
$optimize()
supports progress bars via the package progressr
combined with a Terminator. Simply wrap the function in
progressr::with_progress()
to enable them. We recommend to use package
progress as backend; enable with progressr::handlers("progress")
.
bbotk::Optimizer
-> OptimizerGridSearch
Inherited methods
new()
Creates a new instance of this R6 class.
OptimizerGridSearch$new()
clone()
The objects of this class are cloneable with this method.
OptimizerGridSearch$clone(deep = FALSE)
deep
Whether to make a deep clone.
library(paradox) domain = ParamSet$new(list(ParamDbl$new("x", lower = -1, upper = 1))) search_space = ParamSet$new(list(ParamDbl$new("x", lower = -1, upper = 1))) codomain = ParamSet$new(list(ParamDbl$new("y", tags = "minimize"))) objective_function = function(xs) { list(y = as.numeric(xs)^2) } objective = ObjectiveRFun$new(fun = objective_function, domain = domain, codomain = codomain) terminator = trm("evals", n_evals = 10) instance = OptimInstanceSingleCrit$new( objective = objective, search_space = search_space, terminator = terminator) optimizer = opt("grid_search") # Modifies the instance by reference optimizer$optimize(instance)#> x x_domain y #> 1: -0.1111111 <list[1]> 0.01234568# Returns best scoring evaluation instance$result#> x x_domain y #> 1: -0.1111111 <list[1]> 0.01234568#> x y x_domain timestamp batch_nr #> 1: -0.3333333 0.11111111 <list[1]> 2021-01-24 15:47:16 1 #> 2: -0.5555556 0.30864198 <list[1]> 2021-01-24 15:47:16 2 #> 3: 1.0000000 1.00000000 <list[1]> 2021-01-24 15:47:16 3 #> 4: -0.1111111 0.01234568 <list[1]> 2021-01-24 15:47:16 4 #> 5: -0.7777778 0.60493827 <list[1]> 2021-01-24 15:47:16 5 #> 6: 0.5555556 0.30864198 <list[1]> 2021-01-24 15:47:16 6 #> 7: 0.1111111 0.01234568 <list[1]> 2021-01-24 15:47:16 7 #> 8: 0.3333333 0.11111111 <list[1]> 2021-01-24 15:47:16 8 #> 9: 0.7777778 0.60493827 <list[1]> 2021-01-24 15:47:16 9 #> 10: -1.0000000 1.00000000 <list[1]> 2021-01-24 15:47:16 10