def test_no_dof(self):
TOL = 1E-10
v = ureal(4.999, 0.0032, independent=False)
i = ureal(0.019661, 0.0000095, independent=False)
phi = ureal(1.04446, 0.00075, independent=False)
set_correlation(-0.36, v, i)
set_correlation(0.86, v, phi)
set_correlation(-0.65, i, phi)
r = v * cos(phi) / i
x = v * sin(phi) / i
z = v / i
equivalent(uncertainty(r), 0.0699787279884, TOL)
equivalent(uncertainty(x), 0.295716826846, TOL)
equivalent(uncertainty(z), 0.236602971835, TOL)
equivalent(math.sqrt(welch_satterthwaite(r)[0]), uncertainty(r), TOL)
equivalent(math.sqrt(welch_satterthwaite(x)[0]), uncertainty(x), TOL)
equivalent(math.sqrt(welch_satterthwaite(z)[0]), uncertainty(z), TOL)
equivalent(get_correlation(r, x), -0.591484610819, TOL)
equivalent(get_correlation(x, z), 0.992797472722, TOL)
equivalent(get_correlation(r, z), -0.490623905441, TOL)
def test_with_dof_2(self):
"""
Same test as above, but now ensure that the
ensemble calculation can deal with influences
that are not in sequence
"""
TOL = 1E-10
# The dummy variables mean that those included
# in the ensemble will not be a consecutive
# series of IDs, provided we include the
# dummies in the equations, as is done below
# with zero weight.
v = ureal(4.999, 0.0032, 5, independent=False)
dummy_1 = ureal(1, 1)
i = ureal(0.019661, 0.0000095, 5, independent=False)
dummy_2 = ureal(1, 1)
phi = ureal(1.04446, 0.00075, 5, independent=False)
dummy_3 = ureal(1, 1)
real_ensemble([v, i, phi], 5)
set_correlation(-0.36, v, i)
set_correlation(0.86, v, phi)
set_correlation(-0.65, i, phi)
r = v * cos(phi) / i + (0 * dummy_1)
x = v * sin(phi) / i + (0 * dummy_2)
z = v / i + (0 * dummy_3)
# The presence of finite DoF should not alter previous results
equivalent(uncertainty(r), 0.0699787279884, TOL)
equivalent(uncertainty(x), 0.295716826846, TOL)
equivalent(uncertainty(z), 0.236602971835, TOL)
equivalent(math.sqrt(welch_satterthwaite(r)[0]), uncertainty(r), TOL)
equivalent(math.sqrt(welch_satterthwaite(x)[0]), uncertainty(x), TOL)
equivalent(math.sqrt(welch_satterthwaite(z)[0]), uncertainty(z), TOL)
equivalent(get_correlation(r, x), -0.591484610819, TOL)
equivalent(get_correlation(x, z), 0.992797472722, TOL)
equivalent(get_correlation(r, z), -0.490623905441, TOL)
equivalent(dof(r), 5, TOL)
equivalent(dof(x), 5, TOL)
equivalent(dof(z), 5, TOL)
def test_with_dof(self):
TOL = 1E-10
# The old way!
# NB, I think there are now multiple instances of this test
# v = ureal(4.999,0.0032,5,independent=False)
# i = ureal(0.019661,0.0000095,5,independent=False)
# phi = ureal(1.04446,0.00075,5,independent=False)
v, i, phi = multiple_ureal([4.999, 0.019661, 1.04446],
[0.0032, 0.0000095, 0.00075], 5)
set_correlation(-0.36, v, i)
set_correlation(0.86, v, phi)
set_correlation(-0.65, i, phi)
r = v * cos(phi) / i
x = v * sin(phi) / i
z = v / i
# The presence of finite DoF should not alter previous results
equivalent(uncertainty(r), 0.0699787279884, TOL)
equivalent(uncertainty(x), 0.295716826846, TOL)
equivalent(uncertainty(z), 0.236602971835, TOL)
equivalent(math.sqrt(welch_satterthwaite(r)[0]), uncertainty(r), TOL)
equivalent(math.sqrt(welch_satterthwaite(x)[0]), uncertainty(x), TOL)
equivalent(math.sqrt(welch_satterthwaite(z)[0]), uncertainty(z), TOL)
equivalent(get_correlation(r, x), -0.591484610819, TOL)
equivalent(get_correlation(x, z), 0.992797472722, TOL)
equivalent(get_correlation(r, z), -0.490623905441, TOL)
equivalent(dof(r), 5, TOL)
equivalent(dof(x), 5, TOL)
equivalent(dof(z), 5, TOL)
# Add another influence and it breaks again, but earlier results OK
f = ureal(1, 0.0032, 5, independent=False)
self.assertRaises(RuntimeError, set_correlation, 0.3, f, i)
equivalent(dof(r), 5, TOL)
equivalent(dof(x), 5, TOL)
equivalent(dof(z), 5, TOL)
def test_with_dof_multiple_ureal(self):
TOL = 1E-10
x = [4.999, 0.019661, 1.04446]
u = [0.0032, 0.0000095, 0.00075]
labels = ['v', 'i', 'phi']
v, i, phi = multiple_ureal(x, u, 5, labels)
set_correlation(-0.36, v, i)
set_correlation(0.86, v, phi)
set_correlation(-0.65, i, phi)
r = v * cos(phi) / i
x = v * sin(phi) / i
z = v / i
# The presence of finite DoF should not alter previous results
equivalent(uncertainty(r), 0.0699787279884, TOL)
equivalent(uncertainty(x), 0.295716826846, TOL)
equivalent(uncertainty(z), 0.236602971835, TOL)
equivalent(math.sqrt(welch_satterthwaite(r)[0]), uncertainty(r), TOL)
equivalent(math.sqrt(welch_satterthwaite(x)[0]), uncertainty(x), TOL)
equivalent(math.sqrt(welch_satterthwaite(z)[0]), uncertainty(z), TOL)
equivalent(get_correlation(r, x), -0.591484610819, TOL)
equivalent(get_correlation(x, z), 0.992797472722, TOL)
equivalent(get_correlation(r, z), -0.490623905441, TOL)
# Dof calculation should be legal
equivalent(dof(r), 5, TOL)
equivalent(dof(x), 5, TOL)
equivalent(dof(z), 5, TOL)
# Add another influence and it breaks again, but earlier results OK
f = ureal(1, 0.0032, 5, independent=False)
self.assertRaises(RuntimeError, set_correlation, 0.3, f, i)
equivalent(dof(r), 5, TOL)
equivalent(dof(x), 5, TOL)
equivalent(dof(z), 5, TOL)
def test(self):
x = ureal(1, 1, 4)
self.assertEqual(4, x.df)
self.assertEqual(4, welch_satterthwaite(x)[1])
# product with zero values
x1 = ureal(0, 1, 4)
x2 = ureal(0, 1, 3)
y = x1 * x2
self.assertEqual(0, y.u)
self.assertTrue(nan is y.df)
# Pathological case - not sure it can be created in practice
x1 = ureal(1, 1)
x2 = UncertainReal._elementary(1, 0, 4, label=None, independent=True)
x3 = UncertainReal._elementary(1, 0, 3, label=None, independent=True)
y = x1 + x2 + x3
self.assertEqual(len(y._u_components), 3)
self.assertTrue(inf is y.df)