class LMFitTab(LMFitBaseTab, LMFitTab_UI):
fit_finished = pyqtSignal(list, list, dict)
def __init__(self, sliced_data, orbitals):
self.model = LMFitModel(sliced_data, orbitals)
# Setup GUI
super(LMFitTab, self).__init__()
self.setupUi(self)
self._setup()
self._connect()
self.refresh_sliced_plot()
self.refresh_selected_plot()
kmap = self.refresh_sum_plot()
self.refresh_residual_plot(weight_sum_data=kmap)
def get_title(self):
return 'LM-Fit'
def trigger_fit(self):
results = self.model.fit()
settings = self.model.get_settings()
data = [self.model.sliced_data, self.model.orbitals]
self.fit_finished.emit(data, results, settings)
def change_slice(self):
axis_index = self.slider.get_axis()
slice_policy = self.lmfit_options.get_slice_policy()
combined = True if slice_policy == 'all combined' else False
slice_indices = (self.slider.get_index()
if slice_policy == 'only one' else 'all')
self.model.set_slices(slice_indices,
axis_index=axis_index,
combined=combined)
self.refresh_sliced_plot()
self.refresh_residual_plot()
def change_axis(self):
axis = self.interpolation.get_axis()
self.model.set_axis(axis)
self.refresh_sliced_plot()
self.refresh_selected_plot()
kmap = self.refresh_sum_plot()
self.refresh_residual_plot(weight_sum_data=kmap)
def _change_slice_policy(self, slice_policy):
axis = self.slider.get_axis()
if slice_policy == 'all':
self.model.set_slices('all', axis_index=axis, combined=False)
elif slice_policy == 'only one':
index = self.slider.get_index()
self.model.set_slices([index], axis_index=axis, combined=False)
else:
self.model.set_slices('all', axis_index=axis, combined=True)
def _change_region(self, *args):
self.model.set_region(*args)
self.refresh_residual_plot()
def _change_background(self, *args):
new_variables = self.model.set_background_equation(*args)
for variable in new_variables:
self.tree.add_equation_parameter(variable)
self.refresh_sum_plot()
self.refresh_residual_plot()
def _setup(self):
LMFitBaseTab._setup(self)
self.orbital_options = LMFitOrbitalOptions()
self.tree = LMFitTree(self.model.orbitals, self.model.parameters)
self.interpolation = LMFitInterpolation()
self.lmfit_options = LMFitOptions(self)
self.change_axis()
self.model.set_crosshair(self.crosshair.model)
self._change_background(self.lmfit_options.get_background())
layout = QVBoxLayout()
layout.setContentsMargins(3, 3, 3, 3)
layout.setSpacing(3)
self.scroll_area.widget().setLayout(layout)
layout.insertWidget(0, self.slider)
layout.insertWidget(1, self.orbital_options)
layout.insertWidget(2, self.interpolation)
layout.insertWidget(3, self.lmfit_options)
layout.insertWidget(4, self.colormap)
layout.insertWidget(5, self.crosshair)
self.layout.insertWidget(1, self.tree)
def _connect(self):
LMFitBaseTab._connect(self)
self.interpolation.interpolation_changed.connect(self.change_axis)
self.tree.value_changed.connect(self._refresh_orbital_plots)
self.tree.vary_changed.connect(self.update_chi2_label)
self.lmfit_options.background_changed.connect(self._change_background)
self.lmfit_options.fit_triggered.connect(self.trigger_fit)
self.lmfit_options.method_changed.connect(self.model.set_fit_method)
self.lmfit_options.slice_policy_changed.connect(
self._change_slice_policy)
self.orbital_options.symmetrization_changed.connect(
self.model.set_symmetrization)
self.orbital_options.symmetrization_changed.connect(
self._refresh_orbital_plots)
self.orbital_options.polarization_changed.connect(
self.model.set_polarization)
self.orbital_options.polarization_changed.connect(
self._refresh_orbital_plots)
range_, dk = [-3.0, 3.0], 0.04
lmfit.set_axis_by_step_size(range_, dk)
lmfit.set_polarization('toroid', 'p')
lmfit.set_background_equation('c')
# Set parameters not intended for fitting to desired value
lmfit.edit_parameter('E_kin', value=27.2, vary=False)
lmfit.edit_parameter('alpha', value=40, vary=False)
# Activate fitting for background ('c') and all orbital weights
lmfit.edit_parameter('c', value=250, vary=True) # constant background
for i in [0, 1, 2, 3]:
lmfit.edit_parameter('w_' + str(i), vary=True)
# Set slices to be used and perform fit
lmfit.set_slices('all', combined=False)
lmfit.set_fit_method(method='matrix_inversion')
results = lmfit.fit()
# Extract fitting results
weights = np.array([[
result[1].params['w_0'].value, result[1].params['w_1'].value,
result[1].params['w_2'].value, result[1].params['w_3'].value,
result[1].params['c'].value
] for result in results])
print(weights)
# Plot results: weights of orbitals (pDOS) vs. kinetic energy
names = ['PTCDA_C', 'PTCDA_D', 'PTCDA_E', 'PTCDA_F', 'background']
styles = ['.r-', 'k-', 'r--', '^g-', 'k:']
# Set parameters not intended for fitting to desired value
lmfit.edit_parameter('E_kin', value=28, vary=False)
lmfit.edit_parameter('alpha', value=40, vary=False)
lmfit.set_background_equation('c')
# Activate fitting for background ('c') and all orbital weights and theta
lmfit.edit_parameter('c', value=1, vary=True) # constant background
for i in [0]:
lmfit.edit_parameter('w_' + str(i), vary=True)
lmfit.edit_parameter('theta_' + str(i), min=0, value=5, vary=True)
lmfit.edit_parameter('phi_' + str(i), value=90, vary=False)
lmfit.edit_parameter('psi_' + str(i), value=90, vary=False)
# Set slices to be used and perform fit
lmfit.set_slices([2], combined=False)
lmfit.set_fit_method(method='leastsq', xtol=1e-12)
best_fit = lmfit.fit()[0][1]
# Print results of best fit
print('reduced chi^2 = ', best_fit.redchi)
print(best_fit.params['theta_0'])
print(best_fit.params['w_0'])
print(best_fit.params['c'])
# Now make plot of chi^2 vs. theta by looping over a list of
# theta-values, but setting fixing all variables (vary=False) in the fit
lmfit.edit_parameter('w_0', value=best_fit.params['w_0'].value, vary=False)
lmfit.edit_parameter('c', value=best_fit.params['c'].value, vary=False)
theta_values = np.linspace(0, 60, 61)
class TestLMFitModel(unittest.TestCase):
@classmethod
def setUpClass(cls):
file_path = os.path.dirname(os.path.realpath(__file__))
input_path = file_path + '/../../example/data/'
sliced_path = input_path + 'example5_6584.hdf5'
cls.sliced_data = SlicedData.init_from_hdf5(sliced_path)
orbital_paths = [
'PTCDA_C.cube', 'PTCDA_D.cube', 'PTCDA_E.cube', 'PTCDA_F.cube'
]
cls.orbitals = [
OrbitalData.init_from_file(input_path + path, ID)
for ID, path in enumerate(orbital_paths)
]
cls.expected = np.loadtxt(file_path + '/output/weights_PTCDA')
cls.background_expected = np.loadtxt(file_path +
'/output/background_expected')
def setUp(self):
self.lmfit = LMFitModel(TestLMFitModel.sliced_data,
TestLMFitModel.orbitals)
self.crosshair = CrosshairAnnulusModel()
def test_set_crosshair(self):
self.lmfit.set_crosshair(self.crosshair)
self.assertEqual(self.lmfit.crosshair, self.crosshair)
def test_set_axis(self):
step_size = 0.24
range_ = [-3, 3]
axis = np.linspace(*range_,
num=step_size_to_num(range_, step_size),
endpoint=True)
self.lmfit.set_axis_by_step_size(range_, step_size)
npt.assert_almost_equal(self.lmfit.get_sliced_kmap(0).x_axis, axis)
def test_set_slices(self):
self.lmfit.set_slices([1, 2, 3])
self.lmfit.set_slices([0, 1, 2, 3], combined=True)
def test_set_background_equation(self):
self.lmfit.set_background_equation('np.exp(a)')
npt.assert_equal(self.lmfit.background_equation, ['np.exp(a)', ['a']])
self.assertRaises(ValueError, self.lmfit.set_background_equation,
'np.exp(a')
def test_parameters(self):
self.lmfit.set_background_equation('np.exp(a)')
self.assertEqual(self.lmfit.parameters['w_1'].min, 0)
self.assertEqual(self.lmfit.parameters['a'].value, 0)
self.assertEqual(self.lmfit.parameters['E_kin'].max, 150)
def test_edit_parameter(self):
self.lmfit.edit_parameter('w_1', value=1.5)
self.assertEqual(self.lmfit.parameters['w_1'].value, 1.5)
def test_background(self):
range_, dk = [-3.0, 3.0], 0.025
self.lmfit.set_axis_by_step_size(range_, dk)
self.lmfit.set_background_equation(
'(np.exp(-x**2-y**2)-np.exp(-(x-1)**2-(y-1)**2))/2')
background = self.lmfit._get_background(variables={})
npt.assert_almost_equal(background, TestLMFitModel.background_expected)
def test_settings(self):
lmfit_new = LMFitModel(TestLMFitModel.sliced_data,
TestLMFitModel.orbitals)
lmfit_new.set_settings(self.lmfit.get_settings())
def test_PTCDA(self):
if float(config.get_key('orbital', 'dk3D')) != 0.12:
print('WARNING: Test \'test_PTCDA\' from the ' +
'\'test_lmfit\' module has not been run. It requires ' +
'\'dk3D\' setting from the \'cube\' category to be ' +
'to 0.12.')
return
# Set certain parameters not being fitted but desired to be changed
range_, dk = [-3.0, 3.0], 0.04
self.lmfit.set_axis_by_step_size(range_, dk)
self.lmfit.set_polarization('toroid', 'p')
self.lmfit.set_background_equation('c')
# Set certain fit parameter to desired value
self.lmfit.edit_parameter('E_kin', value=27.2)
self.lmfit.edit_parameter('alpha', value=40)
self.lmfit.edit_parameter('c', value=1, vary=True)
# Activate fitting for all weights (i is a dummy ID used in setUpClass)
for i in [0, 1, 2, 3]:
self.lmfit.edit_parameter('w_' + str(i), vary=True)
# Set slices to be used
self.lmfit.set_slices('all', combined=False)
results = self.lmfit.fit()
# Test results
weights = np.array([[
result[1].params['w_0'].value, result[1].params['w_1'].value,
result[1].params['w_2'].value, result[1].params['w_3'].value
] for result in results]).T
npt.assert_almost_equal(weights, TestLMFitModel.expected, decimal=5)
class LMFitTab(LMFitBaseTab, LMFitTab_UI):
fit_finished = pyqtSignal(list, dict)
def __init__(self, sliced_tab, orbital_tab, max_orbitals=-1):
self.sliced_tab = sliced_tab
self.orbital_tab = orbital_tab
if max_orbitals != -1:
self.model = LMFitModel(sliced_tab.get_data(),
orbital_tab.get_orbitals()[:max_orbitals])
else:
self.model = LMFitModel(sliced_tab.get_data(),
orbital_tab.get_orbitals())
# Setup GUI
super(LMFitTab, self).__init__()
self.setupUi(self)
self._setup()
self._connect()
self.refresh_all()
@classmethod
def init_from_save(cls, save, dependencies, tab_widget):
sliced_tab = tab_widget.get_tab_by_ID(dependencies['sliced_tab'])
orbital_tab = tab_widget.get_tab_by_ID(dependencies['orbital_tab'])
# max orbitals: If orbitals were loaded after lmfit was created they
# will not be reflected in the results -> not load them
self = cls(sliced_tab, orbital_tab, max_orbitals=len(save['tree'])-2)
self.locked_tabs = [sliced_tab, orbital_tab]
self.title = save['title']
self.tree.restore_state(save['tree'])
self.interpolation.restore_state(save['interpolation']),
self.lmfit_options.restore_state(save['lmfit_options'])
self.orbital_options.restore_state(save['orbital_options']),
self.slider.restore_state(save['slider'])
self.colormap.restore_state(save['colormap'])
self.crosshair.restore_state(save['crosshair'])
self.sliced_plot.set_colormap(save['colorscales']['sliced'])
self.sliced_plot.set_levels(save['levels']['sliced'])
self.sum_plot.set_colormap(save['colorscales']['sum'])
self.sum_plot.set_levels(save['levels']['sum'])
self.residual_plot.set_levels(save['levels']['residual'])
self.residual_plot.set_colormap(save['colorscales']['residual'])
self.selected_plot.set_levels(save['levels']['selected'])
self.selected_plot.set_colormap(save['colorscales']['selected'])
return self
def get_title(self):
return self.title
def get_data(self):
return [self.model.sliced_data, self.model.orbitals]
def trigger_fit(self):
try:
new_results = self.model.fit()
except ValueError as e:
logging.getLogger('kmap').warning(str(e))
self.lmfit_options.update_fit_button()
return
if not hasattr(self, 'results'):
self.results = new_results
else:
results_dict = dict(self.results)
new_results_dict = dict(new_results)
results_dict.update(new_results_dict)
self.results = sorted(list(results_dict.items()), key=lambda x: x[0])
settings = self.model.get_settings()
self.fit_finished.emit(self.results, settings)
def change_slice(self):
axis_index = self.slider.get_axis()
slice_policy = self.lmfit_options.get_slice_policy()
combined = True if slice_policy == 'all combined' else False
slice_indices = (self.slider.get_index()
if slice_policy == 'only one' else 'all')
self.model.set_slices(
slice_indices, axis_index=axis_index, combined=combined)
self.refresh_sliced_plot()
self.refresh_residual_plot()
def change_axis(self):
axis = self.interpolation.get_axis()
self.model.set_axis(axis)
self.refresh_all()
def save_state(self):
save, dependencies = super().save_state()
save.update({'orbital_options': self.orbital_options.save_state(),
'interpolation': self.interpolation.save_state(),
'lmfit_options': self.lmfit_options.save_state(),
'tree': self.tree.save_state()})
dependencies.update({'sliced_tab': self.sliced_tab.ID,
'orbital_tab': self.orbital_tab.ID})
return save, dependencies
def _change_slice_policy(self, slice_policy):
axis = self.slider.get_axis()
if slice_policy == 'all':
self.model.set_slices('all', axis_index=axis, combined=False)
elif slice_policy == 'only one':
index = self.slider.get_index()
self.model.set_slices([index], axis_index=axis, combined=False)
elif slice_policy == 'all combined':
self.model.set_slices('all', axis_index=axis, combined=True)
else:
indices = [int(e) for e in slice_policy.split(' ')]
self.model.set_slices(indices, axis_index=axis, combined=False)
def _change_method(self, method):
self._change_to_matrix_state(method == 'matrix_inversion')
self.model.set_fit_method(method)
def _change_to_matrix_state(self, state):
if state:
variables = self.model.background_equation[1]
if 'c' not in variables:
self.lmfit_options._pre_factor_background()
self.tree._change_to_matrix_state(state)
def _change_region(self, *args):
self.model.set_region(*args)
self.refresh_all()
def _change_background(self, *args):
new_variables = self.model.set_background_equation(*args)
for variable in new_variables:
self.tree.add_equation_parameter(variable)
self.refresh_sum_plot()
self.refresh_residual_plot()
def _setup(self):
LMFitBaseTab._setup(self)
self.orbital_options = LMFitOrbitalOptions()
self.tree = LMFitTree(self.model.orbitals, self.model.parameters)
self.interpolation = LMFitInterpolation()
self.lmfit_options = LMFitOptions(self)
self.change_axis()
self.model.set_crosshair(self.crosshair.model)
self._change_background(self.lmfit_options.get_background())
layout = QVBoxLayout()
layout.setContentsMargins(3, 3, 3, 3)
layout.setSpacing(3)
self.scroll_area.widget().setLayout(layout)
layout.insertWidget(0, self.slider)
layout.insertWidget(1, self.orbital_options)
layout.insertWidget(2, self.interpolation)
layout.insertWidget(3, self.lmfit_options)
layout.insertWidget(4, self.colormap)
layout.insertWidget(5, self.crosshair)
self.layout.insertWidget(1, self.tree)
def _connect(self):
LMFitBaseTab._connect(self)
self.interpolation.interpolation_changed.connect(self.change_axis)
self.tree.value_changed.connect(self._refresh_orbital_plots)
self.tree.vary_changed.connect(self.update_chi2_label)
self.lmfit_options.background_changed.connect(self._change_background)
self.lmfit_options.fit_triggered.connect(self.trigger_fit)
self.lmfit_options.method_changed.connect(self._change_method)
self.lmfit_options.slice_policy_changed.connect(
self._change_slice_policy)
self.lmfit_options.region_changed.connect(self._change_region)
self.orbital_options.symmetrization_changed.connect(
self.model.set_symmetrization)
self.orbital_options.symmetrization_changed.connect(
self._refresh_orbital_plots)
self.orbital_options.polarization_changed.connect(
self.model.set_polarization)
self.orbital_options.polarization_changed.connect(
self._refresh_orbital_plots)