def interpolateAndReturnLogBeta(self, N, gamma, currSmallest=0):
#TODO: redesign API so that only this method and a few other similar ones are exposed.
if not self.eta:
raise ValueError(
"Eta must be defined prior to interpolating and returning log beta"
)
#check against currSmallest
#test case: N=150, gamma=0.01
leftNVal = iP.find_ge(self.NList, N)[0] #200
if leftNVal is self.NList[-1]:
rightNVal = leftNVal
else:
rightNVal = iP.find_gt(self.NList, leftNVal)[0] #300
rightNIndex = self.NFirstIndexHash[rightNVal] #20
leftNIndex = self.NFirstIndexHash[leftNVal] #40
clampedGamma = max(
gamma, self.betasByAscendingNAscendingGamma['gamma'][leftNIndex])
clampedGamma = min(
clampedGamma,
self.betasByAscendingNAscendingGamma['gamma'][rightNIndex])
candidateSmallerLogBeta = np.log(
self.betasByAscendingNAscendingGamma['beta'][iP.find_le_withBounds(
self.betasByAscendingNAscendingGamma['gamma'], clampedGamma,
leftNIndex, rightNIndex)[1]])
if candidateSmallerLogBeta > currSmallest:
return currSmallest
return min(
currSmallest,
self.interpolateAndReturnLogBetaWithoutChecking(
N, gamma)) #TODO: write wrapper function that mins this with 0
def interpolateAndReturnLogBeta(self, N, gamma, currSmallest=0):
#TODO: redesign API so that only this method and a few other similar ones are exposed.
if not self.eta:
raise ValueError("Eta must be defined prior to interpolating and returning log beta")
#check against currSmallest
#test case: N=150, gamma=0.01
leftNVal = iP.find_ge(self.NList,N)[0] #200
if leftNVal is self.NList[-1]:
rightNVal = leftNVal
else:
rightNVal = iP.find_gt(self.NList, leftNVal)[0] #300
rightNIndex = self.NFirstIndexHash[rightNVal] #20
leftNIndex = self.NFirstIndexHash[leftNVal] #40
clampedGamma = max(gamma, self.betasByAscendingNAscendingGamma['gamma'][leftNIndex])
clampedGamma = min(clampedGamma, self.betasByAscendingNAscendingGamma['gamma'][rightNIndex])
candidateSmallerLogBeta = np.log(self.betasByAscendingNAscendingGamma['beta'][iP.find_le_withBounds(
self.betasByAscendingNAscendingGamma['gamma'], clampedGamma, leftNIndex, rightNIndex)[1]])
if candidateSmallerLogBeta > currSmallest:
return currSmallest
return min(currSmallest, self.interpolateAndReturnLogBetaWithoutChecking(N, gamma)) #TODO: write wrapper function that mins this with 0
def test_fine_le_withBounds(self): a=[1,2,3,4,1,2,3,4,1,2,3,4] x=2.6 self.failUnlessEqual(iP.find_le_withBounds(a, x,4,8),(2,6))
def test_fine_le_withBounds(self):
a = [1, 2, 3, 4, 1, 2, 3, 4, 1, 2, 3, 4]
x = 2.6
self.failUnlessEqual(iP.find_le_withBounds(a, x, 4, 8), (2, 6))
