def testCDerivative_beta(self):
"""Make sure the cython code returns the same values as python code."""
beta = 0.79 #Start with something small
outputs, timeslots = self.generateRandomTestData(100)
risk_groups = get_risk_groups(outputs, timeslots)
#Now get a new set, that won't match this, so beta doesn't diverge
outputs, rnd_timeslots = self.generateRandomTestData(100)
risk_outputs = [None for i in xrange(len(timeslots))]
beta_risk = [np.zeros(len(risk_groups[i])) for i in xrange(len(risk_groups))]
part_func = np.zeros(len(timeslots))
weighted_avg = np.zeros(len(timeslots))
pyget_slope(beta, risk_outputs, beta_risk, part_func, weighted_avg, outputs, timeslots)
beta_force = get_beta_force(beta, outputs, risk_groups, part_func, weighted_avg)
for output_index in xrange(len(outputs)):
cder = cderivative_beta(beta, part_func, weighted_avg, beta_force, output_index, outputs, timeslots, risk_groups)
pyder = pyderivative_beta(beta, part_func, weighted_avg, beta_force, output_index, outputs, timeslots, risk_groups)
#print(cder, pyder)
assert(isinstance(pyder, cder.__class__))
assert(round(cder, 20) == round(pyder, 20))
assert(cder == pyder)
def testCGet_slope(self):
"""Make sure the cython code returns the same values as python code."""
#outputs, timeslots = self.generateRandomTestData(100)
beta = 0.79 #Start with something small
#risk_groups = get_risk_groups(timeslots)
outputs, timeslots = self.generateRandomTestData(100)
risk_groups = get_risk_groups(outputs, timeslots)
#Now get a new set, that won't match this, so beta doesn't diverge
outputs, rnd_timeslots = self.generateRandomTestData(100)
pybeta_risk = [np.zeros(len(risk_groups[i])) for i in xrange(len(risk_groups))]
cbeta_risk = [np.zeros(len(risk_groups[i])) for i in xrange(len(risk_groups))]
pypart_func = np.zeros(len(timeslots))
cpart_func = np.zeros(len(timeslots))
pyweighted_avg = np.zeros(len(timeslots))
cweighted_avg = np.zeros(len(timeslots))
pyslope = pyget_slope(beta, risk_groups, pybeta_risk, pypart_func, pyweighted_avg, outputs, timeslots)
cslope = cget_slope(beta, risk_groups, cbeta_risk, cpart_func, cweighted_avg, outputs, timeslots)
#Check equality between all returned values
#print(pyslope, pyslope.__class__)
#print(cslope, cslope.__class__)
assert(not np.isnan(pyslope))
assert(not np.isnan(cslope))
#assert(isinstance(pyslope, cslope.__class__)) #Make sure they are of the same type
#print(pyslope, cslope)
assert(round(pyslope, 20) == round(cslope, 20))
assert(pyslope == cslope)
# #Risk outputs
# for i, c in zip(range(len(crisk_outputs)), crisk_outputs):
# p = outputs[risk_groups[i], 0]
# assert(isinstance(p, c.__class__))
# #print(p, p.__class__)
# #print(c, c.__class__)
# for pp, cc in zip(p, c):
# #print(pp, cc)
# assert(isinstance(pp, cc.__class__))
# assert(pp == cc)
#Beta risk
#print("Beta Risk")
for p, c, in zip(pybeta_risk, cbeta_risk):
assert(isinstance(p, c.__class__))
for pp, cc in zip(p, c):
#print(pp, cc)
assert(isinstance(pp, cc.__class__))
#if (pp != cc):
assert(pp == cc)
#part_func
#print("Part func")
for p, c in zip(pypart_func, cpart_func):
#print(p, c)
assert(isinstance(p, c.__class__))
assert(p == c)
#weighted average
#print("Weighted avg")
for p, c in zip(pyweighted_avg, cweighted_avg):
#print(p, c)
assert(isinstance(p, c.__class__))
assert(p == c)