def dictRevenue(v, l):
m = numpy.asarray(v).shape[0]
bundles = listBundles(m)
revenue = listRevenue(bundles, v, l)
d = {}
for b,r in zip(bundles,revenue):
d[tuple(b)] = r
return d
def listRevenue(bundles, v, l):
"""Compute the revenue (valuation) for a given
list of bundles (collection of goods)
Parameters
----------
bundles: array_like, shape (n_bundles, n_goods)
List of collection of goods. Each row is collection, each column a good index.
A 1 in the i^{th} row and j^{th} column implies the good j is contained in the
i^{th} listed bundle.
v: array_like, shape (n_goods)
The value vector described in the Market Scheduling game of YW.
l: int,
The minimal number of goods the agent needs to win to obtain value.
Another parameter of the market schedule game.
Returns
-------
valution: array_like, shape (n_bundles)
valution[i] is the revenue the agent would receive had
he/she been able to procure the collection of goods bundle[i].
"""
if bundles == None:
bundles = listBundles(numpy.atleast_1d(v).shape[0])
else:
bundles = numpy.atleast_2d(bundles)
cs = [numpy.atleast_1d(i) for i in itertools.imap(numpy.cumsum,bundles)]
valuation = []
for bundle in cs:
if bundle[-1] < l:
valuation.append(0)
else:
t = numpy.nonzero(bundle >= l)[0][0]
valuation.append(v[t])
return numpy.atleast_1d(valuation)
def targetPrice64(bundles, revenue, pricePrediction, verbose = False):
samples = pricePrediction.sample(n_samples = 64)
expectedPrices = numpy.mean(samples,0)
return targetPrice(bundles, revenue, expectedPrices, verbose)
def targetPrice256(bundles, revenue, pricePrediction, verbose = False):
samples = pricePrediction.sample(n_samples = 256)
expectedPrices = numpy.mean(samples,0)
return targetPrice(bundles, revenue, expectedPrices, verbose)
if __name__ == "__main__":
from ssapy.util import listBundles
from ssapy.agents.marketSchedule import listRevenue
pp = [5,5]
l = 1
v = [20,10]
bundles = listBundles(2)
rev = listRevenue(bundles, v, l)
print bundles
print rev
bid = targetPrice(bundles,rev,pp,True)
