OptimizerBatchLocalSearch class that implements a simple Local Search.
Local Search starts by determining the n_initial_points initial best points present in the Archive of the OptimInstance.
If fewer points than n_initial_points are present, additional initial_random_sample_size points sampled uniformly at random are evaluated and the best n_initial_points initial points are determined.
In each iteration, for each of the n_initial_points initial best points, neighbors_per_point neighbors are generated by local mutation.
Local mutation generates a neighbor by sampling a single parameter that is to be mutated and then proceeds as follows: Double parameters (paradox::p_dbl()) are mutated via Gaussian mutation (with a prior standardization to [0, 1] and retransformation after mutation).
Integer parameters (paradox::p_int()) undergo the same mutation but are rounded to the closest integer after mutation.
Categorical parameters (paradox::p_fct() and paradox::p_lgl()) are mutated via uniform mutation.
Note that parameters that are conditioned on (i.e., they are parents of a paradox::Condition, see the dependencies of the search space) are not mutated.
Dictionary
This Optimizer can be instantiated via the dictionary
mlr_optimizers or with the associated sugar function opt():
Parameters
n_initial_pointsinteger(1)
Size of the set of initial best points which are used as starting points for the Local Search. Default is10.initial_random_sample_sizeinteger(1)
Number of points that are sampled uniformly at random before the bestn_initial_pointsinitial points are determined, if fewer points thann_initial_pointsare present in the Archive of the OptimInstance. Default is100.neighbors_per_pointinteger(1)
Number of neighboring points to generate for each of then_initial_pointsbest starting points in each iteration. Default is100.mutation_sdnumeric(1)
Standard deviation used to create neighbors during mutation of numeric parameters on the standardized[0, 1]scale. Default is0.1.
Archive
The Archive holds the following additional column that is specific to the algorithm:
.point_id(integer(1))
The id (1, ..., n_initial_points) indicating from which of then_initial_pointsbest points the evaluated point was generated from.
Progress Bars
$optimize() supports progress bars via the package progressr
combined with a Terminator. Simply wrap the function in
progressr::with_progress() to enable them. We recommend to use package
progress as backend; enable with progressr::handlers("progress").
Super classes
bbotk::Optimizer -> bbotk::OptimizerBatch -> OptimizerBatchLocalSearch
Examples
search_space = domain = ps(x = p_dbl(lower = -1, upper = 1))
codomain = ps(y = p_dbl(tags = "minimize"))
objective_function = function(xs) {
list(y = as.numeric(xs)^2)
}
objective = ObjectiveRFun$new(
fun = objective_function,
domain = domain,
codomain = codomain)
instance = OptimInstanceBatchSingleCrit$new(
objective = objective,
search_space = search_space,
terminator = trm("evals", n_evals = 100))
# evaluate an initial sample of 10 points uniformly at random
# choose the best 3 points as the initial points
# for each of these points generate 10 neighbors
# repeat this process
optimizer = opt("local_search",
n_initial_points = 3,
initial_random_sample_size = 10,
neighbors_per_point = 10)
# modifies the instance by reference
optimizer$optimize(instance)
#> x x_domain y
#> <num> <list> <num>
#> 1: 0.002401566 <list[1]> 5.76752e-06
# returns best scoring evaluation
instance$result
#> x x_domain y
#> <num> <list> <num>
#> 1: 0.002401566 <list[1]> 5.76752e-06
# allows access of data.table of full path of all evaluations
as.data.table(instance$archive$data)
#> x y x_domain timestamp batch_nr .point_id
#> <num> <num> <list> <POSc> <int> <int>
#> 1: 0.434562679 1.888447e-01 <list[1]> 2025-02-11 15:24:18 1 NA
#> 2: -0.125425003 1.573143e-02 <list[1]> 2025-02-11 15:24:18 1 NA
#> 3: 0.033470919 1.120302e-03 <list[1]> 2025-02-11 15:24:18 1 NA
#> 4: 0.353672964 1.250846e-01 <list[1]> 2025-02-11 15:24:18 1 NA
#> 5: 0.177525542 3.151532e-02 <list[1]> 2025-02-11 15:24:18 1 NA
#> 6: 0.099881455 9.976305e-03 <list[1]> 2025-02-11 15:24:18 1 NA
#> 7: 0.041221941 1.699248e-03 <list[1]> 2025-02-11 15:24:18 1 NA
#> 8: -0.320341957 1.026190e-01 <list[1]> 2025-02-11 15:24:18 1 NA
#> 9: -0.931926232 8.684865e-01 <list[1]> 2025-02-11 15:24:18 1 NA
#> 10: -0.622936200 3.880495e-01 <list[1]> 2025-02-11 15:24:18 1 NA
#> 11: 0.240396995 5.779072e-02 <list[1]> 2025-02-11 15:24:18 2 1
#> 12: 0.078773803 6.205312e-03 <list[1]> 2025-02-11 15:24:18 2 1
#> 13: -0.264627964 7.002796e-02 <list[1]> 2025-02-11 15:24:18 2 1
#> 14: 0.289058224 8.355466e-02 <list[1]> 2025-02-11 15:24:18 2 1
#> 15: -0.264506043 6.996345e-02 <list[1]> 2025-02-11 15:24:18 2 1
#> 16: -0.529376677 2.802397e-01 <list[1]> 2025-02-11 15:24:18 2 1
#> 17: 0.478036678 2.285191e-01 <list[1]> 2025-02-11 15:24:18 2 1
#> 18: -0.430212293 1.850826e-01 <list[1]> 2025-02-11 15:24:18 2 1
#> 19: -0.122211896 1.493575e-02 <list[1]> 2025-02-11 15:24:18 2 1
#> 20: 0.247608977 6.131021e-02 <list[1]> 2025-02-11 15:24:18 2 1
#> 21: 0.223349981 4.988521e-02 <list[1]> 2025-02-11 15:24:18 2 2
#> 22: 0.188007406 3.534678e-02 <list[1]> 2025-02-11 15:24:18 2 2
#> 23: 0.098430576 9.688578e-03 <list[1]> 2025-02-11 15:24:18 2 2
#> 24: -0.103020881 1.061330e-02 <list[1]> 2025-02-11 15:24:18 2 2
#> 25: 0.223285963 4.985662e-02 <list[1]> 2025-02-11 15:24:18 2 2
#> 26: 0.101562463 1.031493e-02 <list[1]> 2025-02-11 15:24:18 2 2
#> 27: 0.567367025 3.219053e-01 <list[1]> 2025-02-11 15:24:18 2 2
#> 28: -0.267905320 7.177326e-02 <list[1]> 2025-02-11 15:24:18 2 2
#> 29: 0.102661949 1.053948e-02 <list[1]> 2025-02-11 15:24:18 2 2
#> 30: 0.172578100 2.978320e-02 <list[1]> 2025-02-11 15:24:18 2 2
#> 31: 0.078924628 6.229097e-03 <list[1]> 2025-02-11 15:24:18 2 3
#> 32: 0.457976118 2.097421e-01 <list[1]> 2025-02-11 15:24:18 2 3
#> 33: -0.084205498 7.090566e-03 <list[1]> 2025-02-11 15:24:18 2 3
#> 34: -0.302660706 9.160350e-02 <list[1]> 2025-02-11 15:24:18 2 3
#> 35: 0.143066197 2.046794e-02 <list[1]> 2025-02-11 15:24:18 2 3
#> 36: -0.280027320 7.841530e-02 <list[1]> 2025-02-11 15:24:18 2 3
#> 37: 0.102836571 1.057536e-02 <list[1]> 2025-02-11 15:24:18 2 3
#> 38: -0.239177842 5.720604e-02 <list[1]> 2025-02-11 15:24:18 2 3
#> 39: 0.102444799 1.049494e-02 <list[1]> 2025-02-11 15:24:18 2 3
#> 40: 0.228273249 5.210868e-02 <list[1]> 2025-02-11 15:24:18 2 3
#> 41: -0.019443213 3.780385e-04 <list[1]> 2025-02-11 15:24:18 3 1
#> 42: 0.351599501 1.236222e-01 <list[1]> 2025-02-11 15:24:18 3 1
#> 43: 0.093679678 8.775882e-03 <list[1]> 2025-02-11 15:24:18 3 1
#> 44: -0.183109416 3.352906e-02 <list[1]> 2025-02-11 15:24:18 3 1
#> 45: 0.117927886 1.390699e-02 <list[1]> 2025-02-11 15:24:18 3 1
#> 46: -0.031940662 1.020206e-03 <list[1]> 2025-02-11 15:24:18 3 1
#> 47: -0.344077844 1.183896e-01 <list[1]> 2025-02-11 15:24:18 3 1
#> 48: 0.240515781 5.784784e-02 <list[1]> 2025-02-11 15:24:18 3 1
#> 49: -0.003064378 9.390416e-06 <list[1]> 2025-02-11 15:24:18 3 1
#> 50: -0.090463875 8.183713e-03 <list[1]> 2025-02-11 15:24:18 3 1
#> 51: 0.267099959 7.134239e-02 <list[1]> 2025-02-11 15:24:18 3 2
#> 52: 0.067776339 4.593632e-03 <list[1]> 2025-02-11 15:24:18 3 2
#> 53: 0.002401566 5.767520e-06 <list[1]> 2025-02-11 15:24:18 3 2
#> 54: -0.099498981 9.900047e-03 <list[1]> 2025-02-11 15:24:18 3 2
#> 55: 0.053078882 2.817368e-03 <list[1]> 2025-02-11 15:24:18 3 2
#> 56: 0.362144847 1.311489e-01 <list[1]> 2025-02-11 15:24:18 3 2
#> 57: 0.378641165 1.433691e-01 <list[1]> 2025-02-11 15:24:18 3 2
#> 58: 0.014201558 2.016843e-04 <list[1]> 2025-02-11 15:24:18 3 2
#> 59: -0.192897827 3.720957e-02 <list[1]> 2025-02-11 15:24:18 3 2
#> 60: -0.071325349 5.087305e-03 <list[1]> 2025-02-11 15:24:18 3 2
#> 61: 0.073715337 5.433951e-03 <list[1]> 2025-02-11 15:24:18 3 3
#> 62: 0.182270847 3.322266e-02 <list[1]> 2025-02-11 15:24:18 3 3
#> 63: 0.135811673 1.844481e-02 <list[1]> 2025-02-11 15:24:18 3 3
#> 64: 0.113078756 1.278681e-02 <list[1]> 2025-02-11 15:24:18 3 3
#> 65: 0.330411781 1.091719e-01 <list[1]> 2025-02-11 15:24:18 3 3
#> 66: -0.096564927 9.324785e-03 <list[1]> 2025-02-11 15:24:18 3 3
#> 67: -0.094851444 8.996796e-03 <list[1]> 2025-02-11 15:24:18 3 3
#> 68: 0.416035569 1.730856e-01 <list[1]> 2025-02-11 15:24:18 3 3
#> 69: -0.172022315 2.959168e-02 <list[1]> 2025-02-11 15:24:18 3 3
#> 70: 0.101083439 1.021786e-02 <list[1]> 2025-02-11 15:24:18 3 3
#> 71: 0.255765908 6.541620e-02 <list[1]> 2025-02-11 15:24:19 4 1
#> 72: -0.179159708 3.209820e-02 <list[1]> 2025-02-11 15:24:19 4 1
#> 73: -0.044734783 2.001201e-03 <list[1]> 2025-02-11 15:24:19 4 1
#> 74: -0.170547961 2.908661e-02 <list[1]> 2025-02-11 15:24:19 4 1
#> 75: -0.459601149 2.112332e-01 <list[1]> 2025-02-11 15:24:19 4 1
#> 76: 0.108248374 1.171771e-02 <list[1]> 2025-02-11 15:24:19 4 1
#> 77: -0.097656151 9.536724e-03 <list[1]> 2025-02-11 15:24:19 4 1
#> 78: 0.083178488 6.918661e-03 <list[1]> 2025-02-11 15:24:19 4 1
#> 79: -0.007977024 6.363291e-05 <list[1]> 2025-02-11 15:24:19 4 1
#> 80: 0.249340862 6.217087e-02 <list[1]> 2025-02-11 15:24:19 4 1
#> 81: -0.308352315 9.508115e-02 <list[1]> 2025-02-11 15:24:19 4 2
#> 82: -0.136337555 1.858793e-02 <list[1]> 2025-02-11 15:24:19 4 2
#> 83: 0.068922042 4.750248e-03 <list[1]> 2025-02-11 15:24:19 4 2
#> 84: -0.119960333 1.439048e-02 <list[1]> 2025-02-11 15:24:19 4 2
#> 85: -0.375316340 1.408624e-01 <list[1]> 2025-02-11 15:24:19 4 2
#> 86: 0.051073021 2.608454e-03 <list[1]> 2025-02-11 15:24:19 4 2
#> 87: 0.052424317 2.748309e-03 <list[1]> 2025-02-11 15:24:19 4 2
#> 88: 0.234961920 5.520710e-02 <list[1]> 2025-02-11 15:24:19 4 2
#> 89: -0.003919949 1.536600e-05 <list[1]> 2025-02-11 15:24:19 4 2
#> 90: 0.231008659 5.336500e-02 <list[1]> 2025-02-11 15:24:19 4 2
#> 91: -0.085801556 7.361907e-03 <list[1]> 2025-02-11 15:24:19 4 3
#> 92: 0.106501326 1.134253e-02 <list[1]> 2025-02-11 15:24:19 4 3
#> 93: 0.208232389 4.336073e-02 <list[1]> 2025-02-11 15:24:19 4 3
#> 94: 0.201602434 4.064354e-02 <list[1]> 2025-02-11 15:24:19 4 3
#> 95: 0.015728904 2.473984e-04 <list[1]> 2025-02-11 15:24:19 4 3
#> 96: 0.116287249 1.352272e-02 <list[1]> 2025-02-11 15:24:19 4 3
#> 97: 0.114082568 1.301483e-02 <list[1]> 2025-02-11 15:24:19 4 3
#> 98: 0.192165550 3.692760e-02 <list[1]> 2025-02-11 15:24:19 4 3
#> 99: 0.072009012 5.185298e-03 <list[1]> 2025-02-11 15:24:19 4 3
#> 100: -0.264301091 6.985507e-02 <list[1]> 2025-02-11 15:24:19 4 3
#> x y x_domain timestamp batch_nr .point_id