def test_alternating_offers_eliciting_mechanism_voi_optimal():
config = SAOElicitingMechanism.generate_config(
cost=0.001,
n_outcomes=10,
opponent_model_adaptive=False,
opponent_type="limited_outcomes",
conflict=1.0,
n_steps=500,
time_limit=100000.0,
own_utility_uncertainty=0.5,
own_reserved_value=0.1,
)
p = SAOElicitingMechanism(**config, elicitor_type="voi_optimal")
p.run()
assert len(p.history) > 0
# pprint(p.elicitation_state)
# print(p.agents[1].reserved_value)
assert p.elicitation_state["elicitation_cost"] >= 0.0
assert (
p.elicitation_state["elicitor_utility"]
+ p.elicitation_state["elicitation_cost"]
>= p.negotiators[1].reserved_value
)
assert (
p.elicitation_state["total_voi"] is not None
and p.elicitation_state["total_voi"] >= 0
)
def test_alternating_offers_eliciting_mechanism_voi():
for strategy, dynamic in [(None, False), (None, True)]:
config = SAOElicitingMechanism.generate_config(
cost=0.001,
n_outcomes=10,
opponent_model_adaptive=False,
opponent_type="limited_outcomes",
conflict=1.0,
n_steps=500,
time_limit=100000.0,
own_utility_uncertainty=0.1,
own_reserved_value=0.1,
)
p = SAOElicitingMechanism(
**config,
elicitation_strategy=strategy,
elicitor_type="voi",
dynamic_queries=dynamic,
)
p.run()
assert len(p.history) > 0
# pprint(p.elicitation_state)
# print(p.agents[1].reserved_value)
assert p.elicitation_state["elicitation_cost"] >= 0.0
assert (
p.elicitation_state["elicitor_utility"] >= p.negotiators[1].reserved_value
)
def test_a_small_elicitation_session():
import os
config = SAOElicitingMechanism.generate_config(cost=0.2, n_outcomes=5, n_steps=10)
p = SAOElicitingMechanism(
**config,
history_file_name=f"{os.path.expanduser('~/logs/negmas/tmp/elicit.log')}'",
)
p.run()
def test_a_typical_elicitation_session():
import random
import os
n_outcomes = 10
accepted_outcomes = [int(random.random() <= 0.5) for _ in range(n_outcomes)]
config = SAOElicitingMechanism.generate_config(
cost=0.2, n_outcomes=n_outcomes, n_steps=100, own_reserved_value=0.25
)
p = SAOElicitingMechanism(
**config,
history_file_name=f"{os.path.expanduser('~/logs/negmas/tmp/elicit.log')}'",
)
p.run()
def test_alternating_offers_eliciting_mechanism_full_knowledge():
config = SAOElicitingMechanism.generate_config(
cost=0.001,
n_outcomes=10,
opponent_model_adaptive=False,
opponent_type="limited_outcomes",
conflict=1.0,
n_steps=500,
time_limit=100000.0,
own_utility_uncertainty=0.1,
own_reserved_value=0.1,
)
p = SAOElicitingMechanism(
**config, elicitation_strategy="bisection", elicitor_type="full_knowledge"
)
p.run()
assert len(p.history) > 0
# pprint(p.elicitation_state)
# print(p.agents[1].reserved_value)
assert p.elicitation_state["elicitation_cost"] == 0.0
assert p.elicitation_state["elicitor_utility"] >= p.negotiators[1].reserved_value
def test_elicitation_run_with_no_conflict():
n_outcomes = 50
n_steps = 100
config = SAOElicitingMechanism.generate_config(
cost=0.05,
n_outcomes=n_outcomes,
conflict=0.0,
winwin=1.0,
n_steps=n_steps,
own_reserved_value=0.1,
opponent_type="tough",
opponent_model_uncertainty=0.0,
own_utility_uncertainty=0.0,
)
neg = SAOElicitingMechanism(**config, elicitor_type="full_knowledge")
frontier, frontier_outcomes = neg.pareto_frontier(sort_by_welfare=True)
assert len(frontier) > 0
neg.run()