def calculate(self):
item_splitter = SurplusMaximizer()
result = item_splitter.split( \
[i.name for i in self.items.all()], \
[p.name for p in self.participants.all()], \
self.splitter_bids() \
)
return result
class TestSurplusMaximizer(unittest.TestCase):
""" Test the Surplus Maximizer splitter """
def setUp(self):
self.splitter = SurplusMaximizer()
def test_single_identical_bid(self):
""" Test when people bid the same thing on a single item
This should theoretically not happen very frequently, but if it did we
would expect the algo to cope with it
"""
bids = [Cost(ITEM1, ACTOR1, 1640),
Cost(ITEM2, ACTOR1, 1540),
Cost(ITEM3, ACTOR1, 1140),
Cost(ITEM4, ACTOR1, 1640),
Cost(ITEM5, ACTOR1, 1740),
Cost(ITEM1, ACTOR2, 1540),
Cost(ITEM2, ACTOR2, 1240),
Cost(ITEM3, ACTOR2, 1340),
Cost(ITEM4, ACTOR2, 1640),
Cost(ITEM5, ACTOR2, 1940),
Cost(ITEM1, ACTOR3, 1740),
Cost(ITEM2, ACTOR3, 1540),
Cost(ITEM3, ACTOR3, 1040),
Cost(ITEM4, ACTOR3, 1640),
Cost(ITEM5, ACTOR3, 1740),
Cost(ITEM1, ACTOR4, 1440),
Cost(ITEM2, ACTOR4, 1640),
Cost(ITEM3, ACTOR4, 1240),
Cost(ITEM4, ACTOR4, 1640),
Cost(ITEM5, ACTOR4, 1740),
Cost(ITEM1, ACTOR5, 1640),
Cost(ITEM2, ACTOR5, 1640),
Cost(ITEM3, ACTOR5, 1140),
Cost(ITEM4, ACTOR5, 1640),
Cost(ITEM5, ACTOR5, 1640)]
result = self.splitter.split(ITEMS[:5], ACTORS[:5], bids)
expected = [(ITEM1, ACTOR3, None),
(ITEM2, ACTOR5, None),
(ITEM3, ACTOR4, None),
(ITEM4, ACTOR1, None),
(ITEM5, ACTOR2, None)]
item_assignments_present(self, result, expected)
def test_obvious_auction(self):
""" Test when people bid such that there is a "right" answer """
bids = [Cost(ITEM1, ACTOR1, 1000),
Cost(ITEM2, ACTOR1, 1000),
Cost(ITEM3, ACTOR1, 1000),
Cost(ITEM4, ACTOR1, 5000),
Cost(ITEM1, ACTOR2, 1000),
Cost(ITEM2, ACTOR2, 1000),
Cost(ITEM3, ACTOR2, 5000),
Cost(ITEM4, ACTOR2, 1000),
Cost(ITEM1, ACTOR3, 1000),
Cost(ITEM2, ACTOR3, 5000),
Cost(ITEM3, ACTOR3, 1000),
Cost(ITEM4, ACTOR3, 1000),
Cost(ITEM1, ACTOR4, 5000),
Cost(ITEM2, ACTOR4, 1000),
Cost(ITEM3, ACTOR4, 1000),
Cost(ITEM4, ACTOR4, 1000)]
result = self.splitter.split(ITEMS[:4], ACTORS[:4], bids)
expected = [(ITEM1, ACTOR4, None),
(ITEM2, ACTOR3, None),
(ITEM3, ACTOR2, None),
(ITEM4, ACTOR1, None)]
item_assignments_present(self, result, expected)
def test_reasonable_auction(self):
""" Test a "real world" auction
Each person has 5000 to use as they wish, on 5 items
"""
# I have no preferences
bids = [Cost(ITEM1, ACTOR1, 1000),
Cost(ITEM2, ACTOR1, 1000),
Cost(ITEM3, ACTOR1, 1000),
Cost(ITEM4, ACTOR1, 1000),
Cost(ITEM5, ACTOR1, 1000),
# I have linear preferences
Cost(ITEM1, ACTOR2, 700),
Cost(ITEM2, ACTOR2, 800),
Cost(ITEM3, ACTOR2, 1000),
Cost(ITEM4, ACTOR2, 1200),
Cost(ITEM5, ACTOR2, 1300),
# I have non-linear preferences
Cost(ITEM1, ACTOR3, 400),
Cost(ITEM2, ACTOR3, 800),
Cost(ITEM3, ACTOR3, 1000),
Cost(ITEM4, ACTOR3, 1200),
Cost(ITEM5, ACTOR3, 1600),
# I have arbitrary preference
Cost(ITEM1, ACTOR4, 2435),
Cost(ITEM2, ACTOR4, 305),
Cost(ITEM3, ACTOR4, 310),
Cost(ITEM4, ACTOR4, 1725),
Cost(ITEM5, ACTOR4, 225),
# I have strong preferences
Cost(ITEM1, ACTOR5, 0),
Cost(ITEM2, ACTOR5, 0),
Cost(ITEM3, ACTOR5, 0),
Cost(ITEM4, ACTOR5, 0),
Cost(ITEM5, ACTOR5, 5000)]
result = self.splitter.split(ITEMS[:5], ACTORS[:5], bids)
expected = [(ITEM1, ACTOR4, None),
(ITEM2, ACTOR1, None),
(ITEM3, ACTOR3, None),
(ITEM4, ACTOR2, None),
(ITEM5, ACTOR5, None)]
item_assignments_present(self, result, expected)
def test_odd_items(self):
""" Test an auction where we have more items than people
"""
# I have some preferences
bids = [Cost(ITEM1, ACTOR1, 1000),
Cost(ITEM2, ACTOR1, 1000),
Cost(ITEM3, ACTOR1, 1000),
Cost(ITEM4, ACTOR1, 0),
Cost(ITEM5, ACTOR1, 2000),
# I have other preferences
Cost(ITEM1, ACTOR2, 0),
Cost(ITEM2, ACTOR2, 1500),
Cost(ITEM3, ACTOR2, 1500),
Cost(ITEM4, ACTOR2, 500),
Cost(ITEM5, ACTOR2, 1500)]
result = self.splitter.split(ITEMS[:5], ACTORS[:2], bids)
expected = [(ITEM1, ACTOR1, None),
(ITEM2, ACTOR2, None),
(ITEM3, ACTOR2, None),
(ITEM4, ACTOR2, None),
(ITEM5, ACTOR1, None)]
item_assignments_present(self, result, expected)
def test_normalize_equal(self):
test_dict = {
ITEM1: (ACTOR1, 1000),
ITEM2: (ACTOR2, 500),
ITEM3: (ACTOR3, 0),
ITEM4: (ACTOR4, 2000)
}
total = 3500
expected = [(ITEM1, ACTOR1, 1000),
(ITEM2, ACTOR2, 500),
(ITEM3, ACTOR3, 0),
(ITEM4, ACTOR4, 2000)]
result = self.splitter.normalize(test_dict, total)
item_assignments_present(self, result, expected)
def test_normalize_high(self):
test_dict = {
ITEM1: (ACTOR1, 1000),
ITEM2: (ACTOR2, 500),
ITEM3: (ACTOR3, 0),
ITEM4: (ACTOR4, 2000)
}
total = 1000
expected = [(ITEM1, ACTOR1, 286),
(ITEM2, ACTOR2, 143),
(ITEM3, ACTOR3, 0),
(ITEM4, ACTOR4, 571)]
result = self.splitter.normalize(test_dict, total)
item_assignments_present(self, result, expected)
def test_normalize_low(self):
test_dict = {
ITEM1: (ACTOR1, 1000),
ITEM2: (ACTOR2, 500),
ITEM3: (ACTOR3, 0),
ITEM4: (ACTOR4, 2000)
}
total = 5000
expected = [(ITEM1, ACTOR1, 1429),
(ITEM2, ACTOR2, 714),
(ITEM3, ACTOR3, 0),
(ITEM4, ACTOR4, 2857)]
result = self.splitter.normalize(test_dict, total)
item_assignments_present(self, result, expected)