def gen_pdf(self, state):
dist_x20001 = dist.Normal(mu=0, sigma=1.0)
x20001 = state['x20001']
p10000 = dist_x20001.log_pdf(x20001)
f20004 = -x20001
cond_20003 = state['cond_20003']
dist_y20005 = dist.Normal(mu=1, sigma=1.0)
p10001 = dist_y20005.log_pdf(1) if not cond_20003 else 0
dist_y20006 = dist.Normal(mu=-1, sigma=1.0)
p10002 = dist_y20006.log_pdf(1) if cond_20003 else 0
logp = p10000 + p10001 + p10002
return logp
def gen_prior_samples(self):
dist_x20001 = dist.Normal(mu=1.0, sigma=2.23606797749979)
x20001 = dist_x20001.sample()
dist_x20002 = dist.Normal(mu=1.0, sigma=2.23606797749979)
x20002 = dist_x20002.sample()
dist_y20003 = dist.Normal(mu=x20001, sigma=1.4142135623730951)
y20003 = 7.0
dist_y20004 = dist.Normal(mu=x20002, sigma=1.4142135623730951)
y20004 = 7.0
state = {}
for _gv in self.gen_vars():
state[_gv] = locals()[_gv]
return state # dictionary
def gen_pdf(self, state):
dist_x20001 = dist.Normal(mu=1.0, sigma=2.23606797749979)
x20001 = state['x20001']
p10000 = dist_x20001.log_pdf(x20001)
dist_x20002 = dist.Normal(mu=1.0, sigma=2.23606797749979)
x20002 = state['x20002']
p10001 = dist_x20002.log_pdf(x20002)
dist_y20003 = dist.Normal(mu=x20001, sigma=1.4142135623730951)
p10002 = dist_y20003.log_pdf(7.0)
dist_y20004 = dist.Normal(mu=x20002, sigma=1.4142135623730951)
p10003 = dist_y20004.log_pdf(7.0)
logp = p10000 + p10001 + p10002 + p10003
return logp
def gen_prior_samples(self):
dist_x20001 = dist.Normal(mu=0, sigma=1.0)
x20001 = dist_x20001.sample()
f20004 = -x20001
cond_20003 = (f20004 >= 0).data[0]
dist_y20005 = dist.Normal(mu=1, sigma=1.0)
y20005 = 1
dist_y20006 = dist.Normal(mu=-1, sigma=1.0)
y20006 = 1
c20002 = y20005 if not cond_20003 else y20006
state = {}
for _gv in self.gen_all_keys():
state[_gv] = locals()[_gv]
return state # dictionary
def gen_pdf(self, state):
dist_x20001 = dist.Normal(mu=0, sigma=1.0)
x20001 = state['x20001']
p10000 = dist_x20001.log_pdf(x20001)
f20004 = -x20001
# The line below has been added as cond_ was not previously updating
cond_20003 = (f20004 >= 0).data[0]
# cond_20003 = state['cond_20003']
dist_y20005 = dist.Normal(mu=1, sigma=1.0)
p10001 = dist_y20005.log_pdf(1) if not cond_20003 else 0
dist_y20006 = dist.Normal(mu=-1, sigma=1.0)
p10002 = dist_y20006.log_pdf(1) if cond_20003 else 0
logp = p10000 + p10001 + p10002
# this has changed
for key in state:
if key in locals():
state[key] = locals()[key]
return logp
def gen_prior_samples(self):
dist_x20001 = dist.Normal(mu=0, sigma=1.0)
x20001 = dist_x20001.sample()
f20004 = -x20001
cond_20003 = (f20004 >= 0).data[0]
dist_x20005 = dist.Normal(mu=0, sigma=1.0)
x20005 = dist_x20005.sample()
dist_x20007 = dist.Normal(mu=0, sigma=1.0)
x20007 = dist_x20007.sample()
dist_y20006 = dist.Normal(mu=x20005, sigma=1.4142135623730951)
y20006 = 10
dist_y20008 = dist.Normal(mu=x20007, sigma=1.4142135623730951)
y20008 = 10
c20002 = y20006 if not cond_20003 else y20008
state = {}
for _gv in self.gen_all_keys():
state[_gv] = locals()[_gv]
return state # dictionary
def gen_pdf(self, state):
dist_x20001 = dist.Normal(mu=0, sigma=1.0)
x20001 = state['x20001']
p10000 = dist_x20001.log_pdf(x20001)
f20004 = -x20001
cond_20003 = (f20004 >= 0).data[0]
dist_x20005 = dist.Normal(mu=0, sigma=1.0)
x20005 = state['x20005']
p10001 = dist_x20005.log_pdf(x20005)
dist_x20007 = dist.Normal(mu=0, sigma=1.0)
x20007 = state['x20007']
p10002 = dist_x20007.log_pdf(x20007)
dist_y20006 = dist.Normal(mu=x20005, sigma=1.4142135623730951)
p10003 = dist_y20006.log_pdf(10) if not cond_20003 else 0
dist_y20008 = dist.Normal(mu=x20007, sigma=1.4142135623730951)
p10004 = dist_y20008.log_pdf(10) if cond_20003 else 0
_lcls = locals()
for key in state:
if key in _lcls:
state[key] = _lcls[key]
logp = p10000 + p10001 + p10002 + p10003 + p10004
return logp
def test_pce():
print('\n================ Test 1 ====================\n')
marginals = distr.Marginals([distr.Uniform(), distr.Normal()])
#basis_truncation = {'total_degree':1, 'q_norm':1.0}
basis_truncation = {'max_degree': 1}
model = PCE(marginals, basis_truncation)
print(model)
model.info()
print('\n')
print(dir(PCE))
print('\n')
#print(model._multiIndices)
X = np.array([[1, 2], [3, 7], [9, 10], [10, 12], [14, 19]])
Y = np.array([[1, 3, 2, 5, 7]]).T
model.fit(X, Y)
model.fit(X, Y, method='projection')
predicted_Y = model.predict(X)
model.info()
print('\n')
print(model)
print('predicted_Y =', predicted_Y)
marginals = distr.Marginals([distr.Uniform(a=2, b=10)])
basis_truncation = {'max_degree': 1}
model = PCE(marginals, basis_truncation)
X = np.array([[1], [7], [10], [12], [19]])
Y = np.array([[1, 3, 2, 5, 7]]).T
model.fit(X, Y)
model.info()
err = model.validate(X, Y)
print(err)
print('\n============================================\n')