def experiment(experiment_name=None):
domain = "domain.pddl"
domain_dir = "blocks"
# benchmark_dir = BENCHMARK_DIR
benchmark_dir = PYPERPLAN_BENCHMARK_DIR
# domain_dir = "gripper"
prob01 = dict(
lp_max_weight=10,
benchmark_dir=benchmark_dir,
instances="task01.pddl",
num_states=500,
num_sampled_states=None,
random_seed=12,
max_concept_size=20,
max_concept_grammar_iterations=3,
concept_generator=None,
parameter_generator=add_domain_parameters,
feature_namer=feature_namer,
)
parameters = {
"task01": prob01,
}.get(experiment_name or "test")
return generate_experiment(domain_dir, domain, **parameters)
def experiment(experiment_name=None):
domain = "domain.pddl"
domain_dir = "childsnacks"
benchmark_dir = BENCHMARK_DIR
childsnacks_1_incremental = dict(
experiment_class=IncrementalExperiment,
lp_max_weight=10,
benchmark_dir=benchmark_dir,
instances=[
'child-snack_pfile01.pddl',
],
test_instances=['child-snack_pfile01-2.pddl'],
test_domain=domain,
#distance_feature_max_complexity=10,
num_states=200,
initial_sample_size=8,
max_concept_grammar_iterations=3,
initial_concept_bound=8,
max_concept_bound=12,
concept_bound_step=2,
batch_refinement_size=5,
clean_workspace=False,
parameter_generator=add_domain_parameters,
feature_namer=feature_namer,
)
parameters = {
"childsnacks_1_incremental": childsnacks_1_incremental,
}.get(experiment_name or "test")
return generate_experiment(domain_dir, domain, **parameters)
def experiment(experiment_name=None):
domain = "domain.pddl"
domain_dir = "storage"
benchmark_dir = DOWNWARD_BENCHMARKS_DIR
storage_incremental = dict(
experiment_class=IncrementalExperiment,
lp_max_weight=10,
benchmark_dir=benchmark_dir,
instances = ['p01.pddl',
'p02.pddl',
'p03.pddl',
'p04.pddl',
'p05.pddl',
'p06.pddl',
'p07.pddl',],
test_instances=['p08.pddl',
'p09.pddl',
'p10.pddl',
'p11.pddl',
'p12.pddl',
'p13.pddl',
'p14.pddl',
'p15.pddl',
'p16.pddl',
'p17.pddl',
'p18.pddl',
'p19.pddl',
'p20.pddl',
'p21.pddl',
'p22.pddl',
'p23.pddl',
'p24.pddl',
'p25.pddl',
'p26.pddl',
'p27.pddl',
'p28.pddl',
'p29.pddl',
'p30.pddl',],
test_domain=domain,
distance_feature_max_complexity=6,
num_states=10000,
initial_sample_size=100,
max_concept_grammar_iterations=3,
initial_concept_bound=8, max_concept_bound=16, concept_bound_step=2,
batch_refinement_size=50,
clean_workspace=False,
parameter_generator=add_domain_parameters,
feature_namer=feature_namer,
)
parameters = {
"storage_incremental": storage_incremental,
}.get(experiment_name or "test")
return generate_experiment(domain_dir, domain, **parameters)
def experiment(experiment_name=None):
domain = "domain.pddl"
domain_dir = "logistics"
benchmark_dir = PYPERPLAN_BENCHMARK_DIR
# Logistics is currently not working
logistics_1 = dict(
lp_max_weight=10,
benchmark_dir=benchmark_dir,
instances=["task01.pddl"],
test_instances=[
"task02.pddl",
],
test_domain=domain,
distance_feature_max_complexity=0,
num_states=20000,
num_sampled_states=None,
random_seed=12,
max_concept_size=10,
max_concept_grammar_iterations=3,
concept_generator=None,
parameter_generator=add_domain_parameters,
feature_namer=feature_namer,
)
logistics_1_incremental = dict(
lp_max_weight=10,
benchmark_dir=benchmark_dir,
instances=["task01.pddl"],
test_instances=[
"task02.pddl",
],
test_domain=domain,
distance_feature_max_complexity=0,
num_states=20000,
initial_sample_size=8,
max_concept_grammar_iterations=3,
initial_concept_bound=8,
max_concept_bound=12,
concept_bound_step=2,
batch_refinement_size=5,
clean_workspace=False,
parameter_generator=add_domain_parameters,
feature_namer=feature_namer,
)
parameters = {
"logistics_1": logistics_1,
"logistics_1_incremental": logistics_1_incremental,
}.get(experiment_name or "test")
return generate_experiment(domain_dir, domain, **parameters)
def experiment(experiment_name):
domain = "domain.pddl"
domain_dir = "grid"
benchmark_dir = BENCHMARK_DIR
experiments = dict()
experiments["prob01"] = dict(
lp_max_weight=10,
benchmark_dir=benchmark_dir,
instances="prob01.pddl",
test_instances=["prob02.pddl"],
test_domain=domain,
# instances="task01.pddl",
num_states=300, # num_sampled_states=None, random_seed=12,
max_concept_size=10,
max_concept_grammar_iterations=3,
concept_generator=None,
# concept_generator=generate_chosen_concepts,
parameter_generator=add_domain_parameters,
feature_namer=feature_namer,
)
experiments["grid_inc"] = dict(
# domain_dir="blocks-downward",
lp_max_weight=10,
experiment_class=IncrementalExperiment,
instances=["prob01.pddl", "prob02.pddl"],
test_instances=["prob03.pddl", "prob04.pddl", "prob05.pddl"],
test_domain=domain,
# This is number of sampled states *per training instance*. In an increm. experiment, they will be processed
# in batches, so we can set them high enough.
num_states=500,
initial_sample_size=20,
max_concept_grammar_iterations=3,
initial_concept_bound=8,
max_concept_bound=8,
concept_bound_step=2,
batch_refinement_size=5,
# quiet=True,
clean_workspace=False,
# concept_generator=generate_chosen_concepts,
parameter_generator=add_domain_parameters,
feature_namer=feature_namer,
)
# Select the actual experiment parameters according to the command-line option
parameters = experiments[experiment_name]
parameters["domain_dir"] = parameters.get("domain_dir", domain_dir)
parameters["domain"] = parameters.get("domain", domain)
return generate_experiment(**parameters)
def experiment(experiment_name=None):
domain = "domain.pddl"
domain_dir = "spanner"
benchmark_dir = BENCHMARK_DIR
# Experiment used in the paper
spanner_1_incremental = dict(
experiment_class=IncrementalExperiment,
lp_max_weight=5,
benchmark_dir=benchmark_dir,
instances=[
'training-1-1-5.pddl',
'training-2-1-5.pddl',
'training-2-2-5.pddl',
'training-3-1-5.pddl',
'training-3-3-5.pddl',
'training-4-2-5.pddl',
'training-4-4-5.pddl',
'training-5-3-5.pddl',
'training-5-4-5.pddl',
'training-5-5-5.pddl',
'training-1-1-6.pddl',
'training-1-1-20.pddl',
],
test_instances=[
"prob-3-3-3-1540903410.pddl", "prob-4-3-3-1540907466.pddl",
"prob-4-3-3-1540907466.pddl", "prob-10-10-10-1540903568.pddl",
"prob-15-10-8-1540913795.pddl", "prob-10-10-25.pddl",
"prob-15-15-25.pddl"
],
# "prob-20-15-40.pddl",
# "prob-20-20-50.pddl",],
test_domain=domain,
num_states=12000,
initial_sample_size=100,
max_concept_grammar_iterations=None,
initial_concept_bound=8,
max_concept_bound=12,
concept_bound_step=2,
batch_refinement_size=50,
clean_workspace=False,
parameter_generator=None, # add_domain_parameters,
)
parameters = {
"spanner_1_incremental": spanner_1_incremental,
}.get(experiment_name or "test")
return generate_experiment(domain_dir, domain, **parameters)
def experiment(experiment_name):
domain = "domain.pddl"
domain_dir = "gripper"
benchmark_dir = PYPERPLAN_BENCHMARK_DIR
experiments = dict()
experiments["incremental"] = dict(
lp_max_weight=10,
experiment_class=IncrementalExperiment,
benchmark_dir=benchmark_dir,
instances=['task01.pddl',
'task02.pddl',
'task03.pddl',
'task04.pddl',],
test_instances=['task05.pddl',
'task06.pddl',
'task07.pddl',
'task08.pddl',
'task09.pddl',
'task10.pddl',
'task11.pddl',
'task12.pddl',
'task13.pddl',
'task14.pddl',
'task15.pddl',
'task16.pddl',
'task17.pddl',
'task18.pddl',
'task19.pddl',
'task20.pddl',],
test_domain=domain,
# This is number of sampled states *per training instance*. In an increm. experiment, they will be processed
# in batches, so we can set them high enough.
num_states=3000,
initial_sample_size=100,
max_concept_grammar_iterations=3,
initial_concept_bound=8, max_concept_bound=12, concept_bound_step=2,
batch_refinement_size=10,
clean_workspace=False,
parameter_generator=None,
feature_namer=feature_namer,
)
# Select the actual experiment parameters according to the command-line option
parameters = experiments[experiment_name]
parameters["domain_dir"] = parameters.get("domain_dir", domain_dir)
parameters["domain"] = parameters.get("domain", domain)
return generate_experiment(**parameters)
def experiment():
benchmark_dir = DOWNWARD_BENCHMARKS_DIR
experiments = dict()
list_of_experiments = []
list_of_domains = domains #[x for x in os.listdir(benchmark_dir)]
for domain in list_of_domains:
domain_dir = os.path.join(DOWNWARD_BENCHMARKS_DIR, domain)
if not os.path.isdir(domain_dir):
continue
if "domain.pddl" not in os.listdir(domain_dir):
continue
instances = []
experiment_name = "experiment_" + domain
for problem in os.listdir(domain_dir):
if problem == "domain.pddl":
continue
instances.append(problem)
experiments[experiment_name] = dict(
lp_max_weight=5,
experiment_class=IncrementalExperiment,
benchmark_dir=benchmark_dir,
instances=[instances[1], instances[2]],
test_instances=[instances[1]],
test_domain='domain.pddl',
# This is number of sampled states *per training instance*. In an increm. experiment, they will be processed
# in batches, so we can set them high enough.
num_states=2000,
initial_sample_size=50,
max_concept_grammar_iterations=3,
initial_concept_bound=7,
max_concept_bound=12,
concept_bound_step=2,
batch_refinement_size=10,
clean_workspace=False,
parameter_generator=None,
)
# Select the actual experiment parameters according to the command-line option
parameters = experiments[experiment_name]
parameters["domain_dir"] = parameters.get("domain_dir", domain_dir)
parameters["domain"] = parameters.get("domain", 'domain.pddl')
list_of_experiments.append(generate_experiment(**parameters))
return list_of_experiments
def experiment(experiment_name):
domain = "domain.pddl"
domain_dir = "gripper-m"
benchmark_dir = BENCHMARK_DIR
experiments = dict()
# Experiment used in the paper
experiments["gripper_std_inc"] = dict(
lp_max_weight=5,
experiment_class=IncrementalExperiment,
instances=['test01.pddl',
'test02.pddl',
'test03.pddl',
'test04.pddl',
'test05.pddl',
'test06.pddl',
'prob03.pddl',
'prob_3balls_3rooms_1rob.pddl'],
test_instances=["prob01.pddl",
"prob02.pddl",
"prob03.pddl",
"prob04.pddl",
"prob05.pddl",
"prob06.pddl",
'prob_3balls_3rooms_1rob.pddl',
"prob_4balls_4rooms_1rob.pddl",
"prob_4balls_4rooms_2rob.pddl",
"prob_10balls_4rooms_1rob.pddl",],
test_domain=domain,
# This is number of sampled states *per training instance*. In an increm. experiment, they will be processed
# in batches, so we can set them high enough.
num_states=12000,
initial_sample_size=100,
max_concept_grammar_iterations=None, random_seed=19,
initial_concept_bound=8, max_concept_bound=12, concept_bound_step=2,
batch_refinement_size=50,
clean_workspace=False,
parameter_generator=None, # add_domain_parameters,
feature_namer=feature_namer,
)
# Select the actual experiment parameters according to the command-line option
parameters = experiments[experiment_name]
parameters["domain_dir"] = parameters.get("domain_dir", domain_dir)
parameters["domain"] = parameters.get("domain", domain)
return generate_experiment(**parameters)
def experiment(experiment_name=None):
domain = "domain.pddl"
domain_dir = "blocks"
# benchmark_dir = BENCHMARK_DIR
benchmark_dir = PYPERPLAN_BENCHMARK_DIR
# domain_dir = "gripper"
experiments = dict()
experiments["task01_inc"] = dict(
lp_max_weight=5,
benchmark_dir=benchmark_dir,
experiment_class=IncrementalExperiment,
instances=['task01.pddl',
'task02.pddl',
'task03.pddl',
],
test_instances=['task04.pddl',
'task05.pddl',
'task06.pddl',
'task07.pddl',
'task08.pddl',
],
test_domain=domain,
# This is number of sampled states *per training instance*. In an increm. experiment, they will be processed
# in batches, so we can set them high enough.
num_states=12000,
initial_sample_size=125,
max_concept_grammar_iterations=3,
initial_concept_bound=12, max_concept_bound=12, concept_bound_step=2,
batch_refinement_size=10,
clean_workspace=False,
parameter_generator=None,
concept_generator=generate_chosen_concepts,
feature_namer=feature_namer,
)
parameters = experiments[experiment_name]
parameters["domain_dir"] = parameters.get("domain_dir", domain_dir)
parameters["domain"] = parameters.get("domain", domain)
return generate_experiment(**parameters)
def experiment(experiment_name=None):
domain = "domain.pddl"
domain_dir = "blocks"
exps = dict()
exps["clear_5"] = dict(
instances="instance_5_clear_x_1.pddl",
test_domain=domain,
test_instances=[
"instance_5_clear_x_2.pddl",
"instance_8_clear_x_0.pddl"
],
lp_max_weight=5,
num_states=2000, max_width=[-1],
num_sampled_states=300,
max_concept_size=6,
max_concept_grammar_iterations=None,
)
exps["on_x_y_big"] = dict(
instances=[
"inst_on_x_y_10.pddl",
"inst_on_x_y_14.pddl",
"holding_a_b_unclear.pddl",
],
test_domain=domain,
test_instances=[
"inst_on_x_y_5.pddl",
"inst_on_x_y_6.pddl",
"inst_on_x_y_7.pddl",
"inst_on_x_y_7_2.pddl",
],
lp_max_weight=5,
num_states=2000, max_width=[-1],
num_sampled_states=50,
max_concept_size=6,
max_concept_grammar_iterations=None,
)
exps["on_x_y_inc"] = dict(
lp_max_weight=5,
experiment_class=IncrementalExperiment,
instances=[
"inst_on_x_y_10.pddl",
],
test_instances=[
"inst_on_x_y_11.pddl",
],
test_domain=domain,
# This is number of sampled states *per training instance*. In an increm. experiment, they will be processed
# in batches, so we can set them high enough.
num_states=12000,
initial_sample_size=50,
initial_concept_bound=6, max_concept_bound=12, concept_bound_step=2,
batch_refinement_size=10,
clean_workspace=False,
)
parameters = exps[experiment_name]
parameters["domain_dir"] = parameters.get("domain_dir", domain_dir)
parameters["domain"] = parameters.get("domain", domain)
return generate_experiment(**parameters)
def experiment(experiment_name=None):
domain = "domain.pddl"
domain_dir = "miconic"
benchmark_dir = BENCHMARK_DIR
miconic_1_simple = dict(
experiment_class=IncrementalExperiment,
lp_max_weight=10,
benchmark_dir=benchmark_dir,
instances=[
's1-0.pddl',
's1-1.pddl',
's1-2.pddl',
's1-3.pddl',
# 's1-4.pddl',
's2-0.pddl',
's2-1.pddl',
's2-2.pddl',
's2-3.pddl',
# 's2-4.pddl',
# 's3-0.pddl',
# 's3-1.pddl',
# 's3-2.pddl',
# 's3-3.pddl',
# 's3-4.pddl',
# 's4-0.pddl',
# 's4-1.pddl',
# 's4-2.pddl',
# 's4-3.pddl',
# 's4-4.pddl',
],
test_instances=['s1-4.pddl', 's2-4.pddl', 's3-4.pddl'],
test_domain=domain,
# distance_feature_max_complexity=10,
num_states=20000,
initial_sample_size=100,
max_concept_grammar_iterations=10,
initial_concept_bound=8,
max_concept_bound=16,
concept_bound_step=2,
batch_refinement_size=50,
clean_workspace=False,
parameter_generator=add_domain_parameters,
feature_namer=feature_namer,
concept_generator=generate_chosen_concepts,
)
# Experiment used in the paper
miconic_1_incremental = dict(
experiment_class=IncrementalExperiment,
lp_max_weight=4,
benchmark_dir=benchmark_dir,
instances=[
's1-0.pddl',
's1-1.pddl',
's1-2.pddl',
's1-3.pddl',
's1-4.pddl',
's2-0.pddl',
's2-1.pddl',
's2-2.pddl',
's2-3.pddl',
's2-4.pddl',
's3-0.pddl',
's3-1.pddl',
's3-2.pddl',
's3-3.pddl',
],
test_instances=[
's5-0.pddl',
's5-1.pddl',
's5-2.pddl',
's5-3.pddl',
's5-4.pddl',
's6-0.pddl',
's6-1.pddl',
's6-2.pddl',
's6-3.pddl',
's6-4.pddl',
's7-0.pddl',
's7-1.pddl',
's7-2.pddl',
's7-3.pddl',
's7-4.pddl',
's8-0.pddl',
's8-1.pddl',
's8-2.pddl',
's8-3.pddl',
's8-4.pddl',
's9-0.pddl',
's9-1.pddl',
's9-2.pddl',
's9-3.pddl',
's9-4.pddl',
's10-0.pddl',
's10-1.pddl',
's10-2.pddl',
's10-3.pddl',
's10-4.pddl',
's1-0.pddl',
's11-0.pddl',
's11-1.pddl',
's11-2.pddl',
's11-3.pddl',
's11-4.pddl',
's1-1.pddl',
's12-0.pddl',
's12-1.pddl',
's12-2.pddl',
's12-3.pddl',
's12-4.pddl',
's1-2.pddl',
's13-0.pddl',
's13-1.pddl',
's13-2.pddl',
's13-3.pddl',
's13-4.pddl',
's1-3.pddl',
's14-0.pddl',
's14-1.pddl',
's14-2.pddl',
's14-3.pddl',
's14-4.pddl',
's1-4.pddl',
's15-0.pddl',
's15-1.pddl',
's15-2.pddl',
's15-3.pddl',
's15-4.pddl',
's16-0.pddl',
's16-1.pddl',
's16-2.pddl',
's16-3.pddl',
's16-4.pddl',
's17-0.pddl',
's17-1.pddl',
's17-2.pddl',
's17-3.pddl',
's17-4.pddl',
's18-0.pddl',
's18-1.pddl',
's18-2.pddl',
's18-3.pddl',
's18-4.pddl',
's19-0.pddl',
's19-1.pddl',
's19-2.pddl',
's19-3.pddl',
's19-4.pddl',
's20-0.pddl',
's20-1.pddl',
's20-2.pddl',
's20-3.pddl',
's20-4.pddl',
's2-0.pddl',
's21-0.pddl',
's21-1.pddl',
's21-2.pddl',
's21-3.pddl',
's21-4.pddl',
's2-1.pddl',
's22-0.pddl',
's22-1.pddl',
's22-2.pddl',
's22-3.pddl',
's22-4.pddl',
's2-2.pddl',
's23-0.pddl',
's23-1.pddl',
's23-2.pddl',
's23-3.pddl',
's23-4.pddl',
's2-3.pddl',
's24-0.pddl',
's24-1.pddl',
's24-2.pddl',
's24-3.pddl',
's24-4.pddl',
's2-4.pddl',
's25-0.pddl',
's25-1.pddl',
's25-2.pddl',
's25-3.pddl',
's25-4.pddl',
's26-0.pddl',
's26-1.pddl',
's26-2.pddl',
's26-3.pddl',
's26-4.pddl',
's27-0.pddl',
's27-1.pddl',
's27-2.pddl',
's27-3.pddl',
's27-4.pddl',
's28-0.pddl',
's28-1.pddl',
's28-2.pddl',
's28-3.pddl',
's28-4.pddl',
's29-0.pddl',
's29-1.pddl',
's29-2.pddl',
's29-3.pddl',
's29-4.pddl',
's30-0.pddl',
's30-1.pddl',
's30-2.pddl',
's30-3.pddl',
's30-4.pddl',
],
test_domain=domain,
#distance_feature_max_complexity=5,
num_states=12000,
initial_sample_size=50,
max_concept_grammar_iterations=10,
initial_concept_bound=8,
max_concept_bound=16,
concept_bound_step=2,
batch_refinement_size=50,
clean_workspace=False,
parameter_generator=None,
feature_namer=feature_namer,
#concept_generator=generate_chosen_concepts,
)
parameters = {
"miconic_1_incremental": miconic_1_incremental,
"miconic_1_simple": miconic_1_simple,
}.get(experiment_name or "test")
return generate_experiment(domain_dir, domain, **parameters)
def experiment(experiment_name=None):
domain = "domain.pddl"
domain_dir = "spanner-ipc2011"
benchmark_dir = BENCHMARK_DIR
spanner_1 = dict(
lp_max_weight=10,
benchmark_dir=benchmark_dir,
#instances=["pfile01-001-training-inst.pddl"],
instances=["prob-3-3-3-1540903410.pddl"],
#"prob-4-4-3-1540907456.pddl",
#"prob-4-3-3-1540907466.pddl",],
#"prob-10-10-10-1540903568.pddl",
#"prob-15-10-8-1540913795.pddl"],
test_instances=[
"pfile01-001.pddl",
"pfile01-002.pddl",
"pfile01-003.pddl",
"pfile01-004.pddl",
"pfile01-005.pddl",
"pfile02-006.pddl",
"pfile02-007.pddl",
"pfile02-008.pddl",
"pfile02-009.pddl",
"pfile02-010.pddl",
"pfile03-011.pddl",
"pfile03-012.pddl",
"pfile03-013.pddl",
"pfile03-014.pddl",
"pfile03-015.pddl",
"pfile04-016.pddl",
"pfile04-017.pddl",
"pfile04-018.pddl",
"pfile04-019.pddl",
"pfile04-020.pddl",
"pfile05-021.pddl",
"pfile05-022.pddl",
"pfile05-023.pddl",
"pfile05-024.pddl",
"pfile05-025.pddl",
"pfile06-026.pddl",
"pfile06-027.pddl",
"pfile06-028.pddl",
"pfile06-029.pddl",
"pfile06-030.pddl",
],
test_domain=domain,
distance_feature_max_complexity=0,
num_states=300,
num_sampled_states=None,
random_seed=12,
max_concept_size=10,
max_concept_grammar_iterations=3,
concept_generator=None,
parameter_generator=add_domain_parameters,
feature_namer=feature_namer,
)
spanner_1_incremental = dict(
lp_max_weight=10,
benchmark_dir=benchmark_dir,
instances=["pfile01-001-training-inst.pddl"],
test_instances=[
"pfile01-001.pddl",
"pfile01-002.pddl",
"pfile01-003.pddl",
"pfile01-004.pddl",
"pfile01-005.pddl",
"pfile02-006.pddl",
"pfile02-007.pddl",
"pfile02-008.pddl",
"pfile02-009.pddl",
"pfile02-010.pddl",
"pfile03-011.pddl",
"pfile03-012.pddl",
"pfile03-013.pddl",
"pfile03-014.pddl",
"pfile03-015.pddl",
"pfile04-016.pddl",
"pfile04-017.pddl",
"pfile04-018.pddl",
"pfile04-019.pddl",
"pfile04-020.pddl",
"pfile05-021.pddl",
"pfile05-022.pddl",
"pfile05-023.pddl",
"pfile05-024.pddl",
"pfile05-025.pddl",
"pfile06-026.pddl",
"pfile06-027.pddl",
"pfile06-028.pddl",
"pfile06-029.pddl",
"pfile06-030.pddl",
],
test_domain=domain,
distance_feature_max_complexity=0,
num_states=1000,
initial_sample_size=8,
max_concept_grammar_iterations=3,
initial_concept_bound=8,
max_concept_bound=12,
concept_bound_step=2,
batch_refinement_size=5,
clean_workspace=False,
parameter_generator=add_domain_parameters,
feature_namer=feature_namer,
)
parameters = {
"spanner_1": spanner_1,
"spanner_1_incremental": spanner_1_incremental,
}.get(experiment_name or "test")
return generate_experiment(domain_dir, domain, **parameters)
def experiment(experiment_name=None):
domain = "domain.pddl"
domain_dir = "visitall"
# Incremental version
# Experiment used in the paper
problem03full_incremental = dict(
benchmark_dir=BENCHMARK_DIR,
lp_max_weight=5,
experiment_class=IncrementalExperiment,
test_domain=domain,
instances=[
'training01.pddl',
'training02.pddl',
'training03.pddl',
'training04.pddl',
'training05.pddl',
'training06.pddl',
'training09.pddl',
'training10.pddl',
'training11.pddl',
],
test_instances=[
"problem02-full.pddl",
"problem03-full.pddl",
"problem04-full.pddl",
"problem05-full.pddl",
"problem06-full.pddl",
"problem07-full.pddl",
"problem08-full.pddl",
"problem09-full.pddl",
"problem10-full.pddl",
"problem11-full.pddl",
'p-1-10.pddl',
'p-1-11.pddl',
'p-1-12.pddl',
'p-1-13.pddl',
'p-1-14.pddl',
'p-1-15.pddl',
'p-1-16.pddl',
'p-1-17.pddl',
'p-1-18.pddl',
'p-1-5.pddl',
'p-1-6.pddl',
'p-1-7.pddl',
'p-1-8.pddl',
'p-1-9.pddl',
],
num_states=10000,
initial_sample_size=100,
distance_feature_max_complexity=5,
max_concept_grammar_iterations=10,
initial_concept_bound=8,
max_concept_bound=12,
concept_bound_step=2,
batch_refinement_size=50,
clean_workspace=False,
parameter_generator=add_domain_parameters,
feature_namer=feature_namer,
)
parameters = {
"problem03full_incremental": problem03full_incremental,
}.get(experiment_name or "test")
return generate_experiment(domain_dir, domain, **parameters)