Add configuration files for the alt-div version of regexp-search

config/regexp_search_alt_div/default.json

@@ -0,0 +1,98 @@
+{
+ "pop_size": 1000,
+ "growth_factor": 2.0,
+ "crossover_ratio": 0.5,
+ "mutation_ratio": 0.45,
+ "remove_duplicates": true,
+
+ "plugin_dir":"_build/plugins/regexp_search/",
+ "plugins":
+ [
+ "../selection/Tournament.cmxs",
+ "../parent_chooser/UniformParentChooser.cmxs",
+ "creation/RandomCreation.cmxs",
+ "creation/CreationFromExample.cmxs",
+ "mutation/RebuildSubtreeMutation.cmxs",
+ "mutation/ReduceMutation.cmxs",
+ "crossover/TakeSubtreeCrossover.cmxs",
+ "crossover/CombineCrossover.cmxs",
+ "fitness/MultiobjectiveFitness.cmxs",
+ "simplification/RepeatSimplification.cmxs",
+ "simplification/GroupChars.cmxs"
+ ],
+
+ "creation":
+ [
+ {
+ "proba":0.1, "method":"from_example",
+ "params":
+ {
+ "replacement_patterns":
+ [
+ { "proba":0.5, "regexp":"[A-Z]" },
+ { "proba":0.5, "regexp":"[a-z]" },
+ { "proba":0.7, "regexp":"[0-9]" },
+ { "proba":0.5, "regexp":"[ \t]" },
+ { "proba":0.8, "regexp":"[A-Za-z]" },
+ { "proba":0.2, "regexp":"." }
+ ]
+ }
+ },
+ { "proba":0.4, "method":"fill", "params":"&random_gen" },
+ { "proba":0.5, "method":"grow", "params":"&random_gen" }
+ ],
+ "mutation":
+ [
+ { "proba":0.7, "method":"rebuild_subtree", "params":"&random_gen" },
+ { "proba":0.3, "method":"reduce" }
+ ],
+ "crossover":
+ [
+ { "proba":0.9, "method":"take_subtree" },
+ { "proba":0.1, "method":"combine", "params":{"alt_proba":0.0} }
+ ],
+ "fitness":
+ {
+ "method":"multiobjective",
+ "params":
+ {
+ "full_example_score":10.0,
+ "full_counter_example_score":250.0,
+ "partial_example_score":5.0,
+ "partial_counter_example_score":0.0,
+ "size_score":1.0,
+ "elim_size":150
+ }
+ },
+ "simplifications":
+ [
+ { "schedule":100, "method":"repeat_simplification" },
+ { "schedule":100, "method":"group_chars" }
+ ],
+ "selection" : { "method":"tournament" },
+ "parent_choice": { "method":"uniform" },
+
+ "&random_gen":
+ {
+ "min_init_depth": 3,
+ "max_init_depth": 7,
+ "max_depth": 10,
+
+ "concat_proba": 0.55,
+ "alt_proba": 0.0,
+ "opt_proba": 0.05,
+ "plus_proba": 0.05,
+ "star_proba": 0.05,
+ "rand_char_proba": 0.25,
+ "predefined_proba": 0.05,
+ "predefined_list":
+ [
+ { "proba":0.2, "regexp":"." },
+ { "proba":0.3, "regexp":"[A-Za-z]" },
+ { "proba":0.1, "regexp":"[A-Z]" },
+ { "proba":0.1, "regexp":"[a-z]" },
+ { "proba":0.2, "regexp":"[0-9]" },
+ { "proba":0.1, "regexp":"[ \t]" }
+ ]
+ }
+}

config/regexp_search_alt_div/strict.json

@@ -0,0 +1,98 @@
+{
+ "pop_size": 1000,
+ "growth_factor": 2.0,
+ "crossover_ratio": 0.5,
+ "mutation_ratio": 0.45,
+ "remove_duplicates": true,
+
+ "plugin_dir":"_build/plugins/regexp_search/",
+ "plugins":
+ [
+ "../selection/Tournament.cmxs",
+ "../parent_chooser/UniformParentChooser.cmxs",
+ "creation/RandomCreation.cmxs",
+ "creation/CreationFromExample.cmxs",
+ "mutation/RebuildSubtreeMutation.cmxs",
+ "mutation/ReduceMutation.cmxs",
+ "crossover/TakeSubtreeCrossover.cmxs",
+ "crossover/CombineCrossover.cmxs",
+ "fitness/MultiobjectiveFitness.cmxs",
+ "simplification/RepeatSimplification.cmxs",
+ "simplification/GroupChars.cmxs"
+ ],
+
+ "creation":
+ [
+ {
+ "proba":0.1, "method":"from_example",
+ "params":
+ {
+ "replacement_patterns":
+ [
+ { "proba":0.5, "regexp":"[A-Z]" },
+ { "proba":0.5, "regexp":"[a-z]" },
+ { "proba":0.7, "regexp":"[0-9]" },
+ { "proba":0.5, "regexp":"[ \t]" },
+ { "proba":0.8, "regexp":"[A-Za-z]" },
+ { "proba":0.2, "regexp":"." }
+ ]
+ }
+ },
+ { "proba":0.4, "method":"fill", "params":"&random_gen" },
+ { "proba":0.5, "method":"grow", "params":"&random_gen" }
+ ],
+ "mutation":
+ [
+ { "proba":0.7, "method":"rebuild_subtree", "params":"&random_gen" },
+ { "proba":0.3, "method":"reduce" }
+ ],
+ "crossover":
+ [
+ { "proba":0.9, "method":"take_subtree" },
+ { "proba":0.1, "method":"combine", "params":{"alt_proba":0.0} }
+ ],
+ "fitness":
+ {
+ "method":"multiobjective",
+ "params":
+ {
+ "full_example_score":10.0,
+ "full_counter_example_score":1000.0,
+ "partial_example_score":5.0,
+ "partial_counter_example_score":1.0,
+ "size_score":1.0,
+ "elim_size":60
+ }
+ },
+ "simplifications":
+ [
+ { "schedule":100, "method":"repeat_simplification" },
+ { "schedule":100, "method":"group_chars" }
+ ],
+ "selection" : { "method":"tournament" },
+ "parent_choice": { "method":"uniform" },
+
+ "&random_gen":
+ {
+ "min_init_depth": 3,
+ "max_init_depth": 7,
+ "max_depth": 10,
+
+ "concat_proba": 0.55,
+ "alt_proba": 0.0,
+ "opt_proba": 0.05,
+ "plus_proba": 0.05,
+ "star_proba": 0.05,
+ "rand_char_proba": 0.25,
+ "predefined_proba": 0.05,
+ "predefined_list":
+ [
+ { "proba":0.2, "regexp":"." },
+ { "proba":0.3, "regexp":"[A-Za-z]" },
+ { "proba":0.1, "regexp":"[A-Z]" },
+ { "proba":0.1, "regexp":"[a-z]" },
+ { "proba":0.2, "regexp":"[0-9]" },
+ { "proba":0.1, "regexp":"[ \t]" }
+ ]
+ }
+}

implementations/regexp_search/regexp_search_alt_div.ml

@@ -36,7 +36,7 @@ let () =
  It creates and evolve a population of candidate regular expressions with a genetic algorithm.\n\
  All the genetic operators on the population must be provided through plugins loaded dynamically.\n\
  You need to provide a configuration file specifying the required plugins and the parameters of the evolution process.\n\
- Usage: regexp-search-alt-d&c [options] configFilename\n\
+ Usage: regexp-search-alt-div [options] configFilename\n\
  \n\
  Options available:"
  in