def test_get_probability_many(self):
"""Test NormalSigmaPCRestraint.get_probability"""
m = IMP.Model()
n = 20
for i in range(n):
sval = random.uniform(0, 10)
su = random.uniform(5, 20)
s = create_scale(m, sval)
a = random.uniform(0, .5)
r = NormalSigmaPCRestraint(s, su, a)
exp_dens = math.exp(-get_score(sval, su, a))
self.assertAlmostEqual(r.get_probability(), exp_dens, delta=1e-6)
r.set_was_used(True)
def test_evaluate_score_sigma_0(self):
"""Test NormalSigmaPCRestraint.unprotected_evaluate for sigma=0"""
m = IMP.Model()
n = 20
for i in range(n):
su = random.uniform(5, 20)
s = create_scale(m, 0)
a = random.uniform(0, .5)
r = NormalSigmaPCRestraint(s, su, a)
exp_score = math.log(su / (-math.log(a)))
self.assertAlmostEqual(r.unprotected_evaluate(None), exp_score,
delta=1e-6)
r.set_was_used(True)
def test_evaluate_score_many(self):
"""Test NormalSigmaPCRestraint.unprotected_evaluate"""
m = IMP.Model()
n = 20
for i in range(n):
sval = random.uniform(0, 10)
su = random.uniform(5, 20)
s = create_scale(m, sval)
a = random.uniform(0, .5)
r = NormalSigmaPCRestraint(s, su, a)
exp_score = get_score(sval, su, a)
self.assertAlmostEqual(r.unprotected_evaluate(None), exp_score,
delta=1e-6)
r.set_was_used(True)
def test_create_restraint2(self):
m = IMP.Model()
s = create_scale(m)
r = NormalSigmaPCRestraint(m, s.get_particle_index(), 20, .001)
self.assertEqual(s, r.get_sigma())
self.assertAlmostEqual(r.get_sigma_upper(), 20)
self.assertAlmostEqual(r.get_alpha(), .001)
r.set_was_used(True)
def test_create_restraint1(self):
m = IMP.Model()
s = create_scale(m)
r = NormalSigmaPCRestraint(s, 10, .01)
self.assertEqual(s, r.get_sigma())
self.assertAlmostEqual(r.get_sigma_upper(), 10)
self.assertAlmostEqual(r.get_alpha(), .01)
r.set_was_used(True)
def test_get_inputs(self):
"""Test NormalSigmaPCRestraint.get_inputs"""
m = IMP.Model()
s = create_scale(m)
r = NormalSigmaPCRestraint(s, 50, .01)
self.assertEqual(len(r.get_inputs()), 1)
self.assertEqual(r.get_inputs()[0], s.get_particle())
r.set_was_used(True)
def test_evaluate_derivative_wrt_sigma(self):
"""Test NormalSigmaPCRestraint.evaluate_derivative_Dxi"""
m = IMP.Model()
n = 10
for i in range(n):
su = random.uniform(5, 20)
s = create_scale(m)
a = random.uniform(0, .5)
r = NormalSigmaPCRestraint(s, su, a)
exp_deriv = -math.log(a) / su
da = IMP.DerivativeAccumulator()
r.unprotected_evaluate(da)
self.assertAlmostEqual(s.get_scale_derivative(), exp_deriv,
delta=1e-6)
r.set_was_used(True)