def test_change_initial_values(self):
fp = FitParameters(n_sites=2)
fp.change_initial_values(dx=0.1, f_p1=0.1)
initial_values = fp.get_initial_values()
self.assertEqual(initial_values['dx'], 0.1)
self.assertEqual(initial_values['dy'], FitParameters.default_initial_values['dy']) # Default 0
self.assertEqual(initial_values['f_p1'], 0.1)
self.assertEqual(initial_values['f_p2'], FitParameters.default_initial_values['f_px']) # Default 0.15
def test_get_p0_pfix(self):
fp = FitParameters(n_sites=1)
keys = fp.get_keys()
p0, pfix = fp.get_p0_pfix()
self.assertIsInstance(p0, tuple)
self.assertIsInstance(pfix, tuple)
self.assertEqual(len(p0), len(keys))
self.assertEqual(len(pfix), len(keys))
def test_initial_values_after_init(self):
for n_sites in (1, 3, 5):
fitp = FitParameters(n_sites=n_sites)
initial_values = fitp.get_initial_values()
fixed_values = fitp.get_fixed_values()
site_str = f'f_p{n_sites}'
self.assertGreaterEqual(0.15, initial_values[site_str])
self.assertEqual(fixed_values[site_str], False)
self.assertEqual(len(initial_values), 5 + n_sites)
def test_parameter_scale_after_init(self):
parameter_scale = FitParameters.default_scale.copy()
parameter_scale.pop('f_px')
parameter_scale.update({'f_p1': 0.01, 'f_p2': 0.01, 'f_p3': 0.01, 'f_p4': 0.01, 'f_p5': 0.01})
for n_sites in range(5):
fitp = FitParameters(n_sites=n_sites)
scale_dict = fitp.get_step_modifier()
for key in scale_dict:
self.assertEqual(scale_dict[key], parameter_scale[key])
def test_update_bounds_with_datapattern(self):
patt = np.ones((100, 100))
x = np.linspace(-2, 2, 100)
y = np.linspace(-0.5, 3.5, 100)
x_mesh, y_mesh = np.meshgrid(x, y)
dp = pyfdd.DataPattern(pattern_array=patt)
dp.set_xymesh(xmesh=x_mesh, ymesh=y_mesh)
fp = FitParameters(n_sites=1)
fp.update_bounds_with_datapattern(datapattern=dp)
result_bounds = fp.get_bounds()
# Assumes default bounds are (-3, 3)
self.assertEqual(result_bounds['dx'], (-2, 2))
self.assertEqual(result_bounds['dy'], (-0.5, 3))
def test_parameter_bounds_after_init(self):
parameter_bounds = FitParameters.default_bounds.copy()
parameter_bounds.pop('f_px')
parameter_bounds.update({'f_p1': (0, 1), 'f_p2': (0, 1), 'f_p3': (0, 1), 'f_p4': (0, 1), 'f_p5': (0, 1)})
for n_sites in range(5):
fitp = FitParameters(n_sites=n_sites)
bounds_dict = fitp._bounds.copy()
for key in bounds_dict:
self.assertEqual(bounds_dict[key], parameter_bounds[key])
def test_change_fixed_values(self):
fp = FitParameters(n_sites=1)
fixed_values = fp.get_fixed_values()
self.assertEqual(fixed_values['sigma'], True)
self.assertEqual(fixed_values['total_cts'], True)
fp.change_fixed_values(sigma=False, total_cts=False)
fixed_values = fp.get_fixed_values()
self.assertEqual(fixed_values['sigma'], False)
self.assertEqual(fixed_values['total_cts'], False)
fp.change_fixed_values(sigma=0.1, total_cts=1)
fixed_values = fp.get_fixed_values()
self.assertEqual(fixed_values['sigma'], True)
self.assertEqual(fixed_values['total_cts'], True)
initial_values = fp.get_initial_values()
self.assertEqual(initial_values['sigma'], 0.1)
self.assertEqual(initial_values['total_cts'], 1)
def test_update_initial_values_with_datapattern(self):
# Dummy DataPattern
patt = np.ones((100, 100))
x = np.linspace(-2, 2, 100)
y = np.linspace(-2, 2, 100)
x_mesh, y_mesh = np.meshgrid(x, y)
dp = pyfdd.DataPattern(pattern_array=patt)
dp.set_xymesh(xmesh=x_mesh, ymesh=y_mesh)
dp.set_pattern_angular_pos((-0.1, 0.5), 3)
# Test
fitp = FitParameters(n_sites=1)
fitp.update_initial_values_with_datapattern(datapattern=dp)
initial_values = fitp.get_initial_values()
self.assertEqual(initial_values['dx'], -0.1)
self.assertEqual(initial_values['dy'], 0.5)
self.assertEqual(initial_values['phi'], 3)
self.assertEqual(initial_values['total_cts'], dp.pattern_matrix.sum())
def test_get_scale_for_fit(self):
patt = np.ones((100, 100)) * 100 # 10e6
total_cts = patt.sum()
dp = pyfdd.DataPattern(pattern_array=patt)
fp = FitParameters(n_sites=1)
fp.update_initial_values_with_datapattern(datapattern=dp)
keys = fp.get_keys()
scale = fp.get_scale_for_fit(cost_function='chi2')
for k, scale_val in zip(keys, scale):
if k == 'total_cts':
self.assertEqual(scale_val, total_cts * FitParameters.default_scale['total_cts'])
scale = fp.get_scale_for_fit(cost_function='ml')
for k, scale_val in zip(keys, scale):
if k == 'total_cts':
self.assertEqual(scale_val, -1)
def test_parameter_keys_after_init(self):
parameter_keys_5 = ['dx', 'dy', 'phi', 'total_cts', 'sigma', 'f_p1', 'f_p2', 'f_p3', 'f_p4', 'f_p5']
for n_sites in range(5):
fitp = FitParameters(n_sites=n_sites)
self.assertEqual(fitp.get_keys(), parameter_keys_5[0:5 + n_sites])
def test__str__(self):
fp = FitParameters(n_sites=1)
print(fp)
def test_change_step_modifier(self):
fp = FitParameters(n_sites=1)
fp.change_step_modifier(dx=1)
scale = fp.get_step_modifier()
self.assertEqual(scale['dx'], 1)
self.assertEqual(scale['dy'], FitParameters.default_scale['dy']) # .01
def test_change_bounds(self):
fp = FitParameters(n_sites=1)
fp.change_bounds(dx=(-1, 1))
bounds = fp.get_bounds()
self.assertEqual(bounds['dx'], (-1, 1))
self.assertEqual(bounds['dy'], FitParameters.default_bounds['dy']) # (-3, +3)
def n_sites_is_string():
FitParameters(n_sites='1')