def load_data(self, prog_reporter):
# Load the data and clean from Nans
raw_xvals = self.getProperty('InputWorkspace').value.readX(0).copy()
raw_yvals = self.getProperty('InputWorkspace').value.readY(0).copy()
prog_reporter.report('Loaded data')
# If the data does not have errors use poisson statistics create an workspace with added errors
raw_error = self.getProperty('InputWorkspace').value.readE(0).copy()
if len(np.argwhere(raw_error > 0)) == 0:
raw_error = np.sqrt(raw_yvals)
error_ws = CreateWorkspace(DataX=raw_xvals,
DataY=raw_yvals,
DataE=raw_error,
StoreInADS=False)
self.setPropertyValue(
'OutputWorkspace', '{}_with_errors'.format(
self.getPropertyValue('InputWorkspace')))
self.setProperty('OutputWorkspace', error_ws)
else:
error_ws = self.getProperty('InputWorkspace').value
self.setPropertyValue('OutputWorkspace', error_ws.getName())
return raw_xvals, raw_yvals, raw_error, error_ws
class FindPeaksAutomaticTest(unittest.TestCase):
data_ws = None
peak_guess_table = None
peak_table_header = [
'centre', 'error centre', 'height', 'error height', 'sigma',
'error sigma', 'area', 'error area'
]
alg_instance = None
x_values = None
y_values = None
def setUp(self):
# Creating two peaks on an exponential background with gaussian noise
self.x_values = np.linspace(0, 100, 1001)
self.x_values = np.linspace(0, 100, 1001)
self.centre = [25, 75]
self.height = [35, 20]
self.width = [10, 5]
self.y_values = self.gaussian(self.x_values, self.centre[0],
self.height[0], self.width[0])
self.y_values += self.gaussian(self.x_values, self.centre[1],
self.height[1], self.width[1])
self.background = 10 * np.ones(len(self.x_values))
self.y_values += self.background
# Generating a table with a guess of the position of the centre of the peaks
peak_table = CreateEmptyTableWorkspace()
peak_table.addColumn(type='float', name='Approximated Centre')
peak_table.addRow([self.centre[0] + 2])
peak_table.addRow([self.centre[1] - 3])
self.peakids = [
np.argwhere(self.x_values == self.centre[0])[0, 0],
np.argwhere(self.x_values == self.centre[1])[0, 0]
]
# Generating a workspace with the data and a flat background
self.raw_ws = CreateWorkspace(DataX=self.x_values,
DataY=self.y_values,
OutputWorkspace='raw_ws')
self.data_ws = CreateWorkspace(
DataX=np.concatenate((self.x_values, self.x_values)),
DataY=np.concatenate((self.y_values, self.background)),
DataE=np.sqrt(np.concatenate((self.y_values, self.background))),
NSpec=2,
OutputWorkspace='data_ws')
self.peak_guess_table = peak_table
self.alg_instance = _FindPeaksAutomatic.FindPeaksAutomatic()
def tearDown(self):
self.delete_if_present('data_ws')
self.delete_if_present('peak_guess_table')
self.delete_if_present('peak_table')
self.delete_if_present('refit_peak_table')
self.delete_if_present('fit_cost')
self.delete_if_present('fit_result_NormalisedCovarianceMatrix')
self.delete_if_present('fit_result_Parameters')
self.delete_if_present('fit_result_Workspace')
self.delete_if_present('fit_table')
self.delete_if_present('data_table')
self.delete_if_present('refit_data_table')
self.delete_if_present('tmp_table')
self.alg_instance = None
self.peak_guess_table = None
self.data_ws = None
@staticmethod
def gaussian(xvals, centre, height, sigma):
exponent = (xvals - centre) / (np.sqrt(2) * sigma)
return height * np.exp(-exponent * exponent)
@staticmethod
def delete_if_present(workspace):
if workspace in mtd:
DeleteWorkspace(workspace)
def assertTableEqual(self, expected, actual):
self.assertEqual(expected.columnCount(), actual.columnCount())
self.assertEqual(expected.rowCount(), actual.rowCount())
for i in range(expected.rowCount()):
self.assertEqual(expected.row(i), actual.row(i))
def assertPeakFound(self,
peak_params,
centre,
height,
sigma,
tolerance=0.01):
if not np.isclose(peak_params['centre'], centre, rtol=tolerance):
raise Exception(
'Expected {}, got {}. Difference greater than tolerance {}'.
format(centre, peak_params['centre'], tolerance))
if not np.isclose(peak_params['height'], height, rtol=tolerance):
raise Exception(
'Expected {}, got {}. Difference greater than tolerance {}'.
format(height, peak_params['height'], tolerance))
if not np.isclose(peak_params['sigma'], sigma, rtol=tolerance):
raise Exception(
'Expected {}, got {}. Difference greater than tolerance {}'.
format(sigma, peak_params['sigma'], tolerance))
def test_algorithm_with_no_input_workspace_raises_exception(self):
with self.assertRaises(RuntimeError):
FindPeaksAutomatic()
def test_algorithm_with_negative_acceptance_threshold_throws(self):
with self.assertRaises(ValueError):
FindPeaksAutomatic(InputWorkspace=self.data_ws,
AcceptanceThreshold=-0.1,
PlotPeaks=False)
def test_algorithm_with_negative_smooth_window_throws(self):
with self.assertRaises(ValueError):
FindPeaksAutomatic(InputWorkspace=self.data_ws,
SmoothWindow=-5,
PlotPeaks=False)
def test_algorithm_with_negative_num_bad_peaks_to_consider_throws(self):
with self.assertRaises(ValueError):
FindPeaksAutomatic(InputWorkspace=self.data_ws,
BadPeaksToConsider=-3,
PlotPeaks=False)
def test_algorithm_with_negative_estimate_of_peak_sigma_throws(self):
with self.assertRaises(ValueError):
FindPeaksAutomatic(InputWorkspace=self.data_ws,
EstimatePeakSigma=-3,
PlotPeaks=False)
def test_algorithm_with_negative_min_peak_sigma_throws(self):
with self.assertRaises(ValueError):
FindPeaksAutomatic(InputWorkspace=self.data_ws,
MinPeakSigma=-0.1,
PlotPeaks=False)
def test_algorithm_with_negative_max_peak_sigma_throws(self):
with self.assertRaises(ValueError):
FindPeaksAutomatic(InputWorkspace=self.data_ws,
MaxPeakSigma=-0.1,
PlotPeaks=False)
def test_algorithm_creates_all_output_workspaces(self):
ws_name = self.raw_ws.getName()
FindPeaksAutomatic(self.raw_ws)
self.assertIn('{}_with_errors'.format(ws_name), mtd)
self.assertIn('{}_{}'.format(self.raw_ws.getName(), 'properties'), mtd)
self.assertIn(
'{}_{}'.format(self.raw_ws.getName(), 'refit_properties'), mtd)
def test_algorithm_does_not_create_temporary_workspaces(self):
FindPeaksAutomatic(self.raw_ws)
self.assertNotIn('ret', mtd)
self.assertNotIn('raw_data_ws', mtd)
self.assertNotIn('flat_ws', mtd)
self.assertNotIn('fit_result_NormalisedCovarianceMatrix', mtd)
self.assertNotIn('fit_result_Parameters', mtd)
self.assertNotIn('fit_result_Workspace', mtd)
self.assertNotIn('fit_cost', mtd)
def test_output_tables_are_correctly_formatted(self):
FindPeaksAutomatic(self.raw_ws, FitToBaseline=True)
peak_table = mtd['{}_{}'.format(self.raw_ws.getName(), 'properties')]
refit_peak_table = mtd['{}_{}'.format(self.raw_ws.getName(),
'refit_properties')]
self.assertEqual(self.peak_table_header, peak_table.getColumnNames())
self.assertEqual(self.peak_table_header,
refit_peak_table.getColumnNames())
self.assertEqual(2, peak_table.rowCount())
self.assertEqual(0, refit_peak_table.rowCount())
def test_single_erosion_returns_correct_result(self):
yvals = np.array([-2, 3, 1, 0, 4])
self.assertEqual(-2, self.alg_instance._single_erosion(yvals, 2, 2))
def test_single_erosion_checks_extremes_of_list_correctly(self):
yvals = np.array([-5, -3, 0, 1, -2, 2, 9])
self.assertEqual(-2, self.alg_instance._single_erosion(yvals, 3, 1))
self.assertEqual(-3, self.alg_instance._single_erosion(yvals, 3, 2))
def test_single_erosion_with_zero_window_does_nothing(self):
yvals = np.array([-5, -3, 0, 1, -2, 2, 9])
self.assertEqual(0, self.alg_instance._single_erosion(yvals, 2, 0))
def test_single_dilation_returns_correct_result(self):
yvals = np.array([-2, 3, 1, 0, 4])
self.assertEqual(4, self.alg_instance._single_dilation(yvals, 2, 2))
def test_single_dilation_checks_extremes_of_list_correctly(self):
yvals = np.array([-5, 3, 0, -7, 2, -2, 9])
self.assertEqual(2, self.alg_instance._single_dilation(yvals, 3, 1))
self.assertEqual(3, self.alg_instance._single_dilation(yvals, 3, 2))
def test_single_dilation_with_zero_window_does_nothing(self):
yvals = np.array([-5, -3, 0, 1, -2, 2, 9])
self.assertEqual(0, self.alg_instance._single_dilation(yvals, 2, 0))
def test_erosion_with_zero_window_is_an_invariant(self):
np.testing.assert_equal(self.y_values,
self.alg_instance.erosion(self.y_values, 0))
def test_erosion_calls_single_erosion_the_correct_number_of_times(self, ):
with mock.patch(
'plugins.algorithms.WorkflowAlgorithms.FindPeaksAutomatic.FindPeaksAutomatic._single_erosion'
) as mock_single_erosion:
times = len(self.y_values)
win_size = 2
call_list = []
for i in range(times):
call_list.append(mock.call(self.y_values, i, win_size))
self.alg_instance.erosion(self.y_values, win_size)
self.assertEqual(times, mock_single_erosion.call_count)
mock_single_erosion.assert_has_calls(call_list, any_order=True)
def test_dilation_with_zero_window_is_an_invariant(self):
np.testing.assert_equal(self.y_values,
self.alg_instance.dilation(self.y_values, 0))
def test_dilation_calls_single_erosion_the_correct_number_of_times(self):
with mock.patch(
'plugins.algorithms.WorkflowAlgorithms.FindPeaksAutomatic.FindPeaksAutomatic._single_dilation'
) as mock_single_dilation:
times = len(self.y_values)
win_size = 2
call_list = []
for i in range(times):
call_list.append(mock.call(self.y_values, i, win_size))
self.alg_instance.dilation(self.y_values, win_size)
self.assertEqual(times, mock_single_dilation.call_count)
mock_single_dilation.assert_has_calls(call_list, any_order=True)
@mock.patch(
'plugins.algorithms.WorkflowAlgorithms.FindPeaksAutomatic.FindPeaksAutomatic.erosion'
)
@mock.patch(
'plugins.algorithms.WorkflowAlgorithms.FindPeaksAutomatic.FindPeaksAutomatic.dilation'
)
def test_opening_calls_correct_functions_in_correct_order(
self, mock_dilation, mock_erosion):
win_size = 3
self.alg_instance.opening(self.y_values, win_size)
self.assertEqual(mock_erosion.call_count, 1)
self.assertEqual(mock_dilation.call_count, 1)
erosion_ret = self.alg_instance.erosion(self.y_values, win_size)
mock_erosion.assert_called_with(self.y_values, win_size)
mock_dilation.assert_called_with(erosion_ret, win_size)
@mock.patch(
'plugins.algorithms.WorkflowAlgorithms.FindPeaksAutomatic.FindPeaksAutomatic.opening'
)
@mock.patch(
'plugins.algorithms.WorkflowAlgorithms.FindPeaksAutomatic.FindPeaksAutomatic.dilation'
)
@mock.patch(
'plugins.algorithms.WorkflowAlgorithms.FindPeaksAutomatic.FindPeaksAutomatic.erosion'
)
def test_average_calls_right_functions_in_right_order(
self, mock_erosion, mock_dilation, mock_opening):
win_size = 3
self.alg_instance.average(self.y_values, win_size)
self.assertEqual(mock_erosion.call_count, 1)
self.assertEqual(mock_dilation.call_count, 1)
self.assertEqual(mock_opening.call_count, 2)
op_ret = self.alg_instance.opening(self.y_values, win_size)
mock_opening.assert_called_with(self.y_values, win_size)
mock_dilation.assert_called_with(op_ret, win_size)
mock_erosion.assert_called_with(op_ret, win_size)
def test_generate_peak_guess_table_correctly_formats_table(self):
peakids = [2, 4, 10, 34]
peak_guess_table = self.alg_instance.generate_peak_guess_table(
self.x_values, peakids)
self.assertEqual(peak_guess_table.getColumnNames(), ['centre'])
def test_generate_peak_guess_table_with_no_peaks_generates_empty_table(
self):
peak_guess_table = self.alg_instance.generate_peak_guess_table(
self.x_values, [])
self.assertEqual(peak_guess_table.rowCount(), 0)
def test_generate_peak_guess_table_adds_correct_values_of_peak_centre(
self):
peakids = [2, 23, 19, 34, 25, 149, 234]
peak_guess_table = self.alg_instance.generate_peak_guess_table(
self.x_values, peakids)
for i, pid in enumerate(sorted(peakids)):
self.assertAlmostEqual(
peak_guess_table.row(i)['centre'], self.x_values[pid], 5)
def test_find_good_peaks_calls_fit_gaussian_peaks_twice_if_no_peaks_given(
self):
with mock.patch(
'plugins.algorithms.WorkflowAlgorithms.FindPeaksAutomatic.FitGaussianPeaks'
) as mock_fit:
tmp_table = CreateEmptyTableWorkspace()
tmp_table.addColumn(type='float', name='chi2')
tmp_table.addColumn(type='float', name='poisson')
tmp_table.addRow([10, 20])
mock_fit.return_value = (mock.MagicMock(), mock.MagicMock(),
tmp_table)
self.alg_instance.min_sigma = 1
self.alg_instance.max_sigma = 10
self.alg_instance.find_good_peaks(self.x_values, [], 0.1, 5, False,
self.data_ws, 5)
self.assertEqual(2, mock_fit.call_count)
def _table_side_effect(self, idx):
raise ValueError('Index = %d' % idx)
def test_find_good_peaks_selects_correct_column_for_error(self):
with mock.patch(
'plugins.algorithms.WorkflowAlgorithms.FindPeaksAutomatic.FitGaussianPeaks'
) as mock_fit:
mock_table = mock.Mock()
mock_table.column.side_effect = self._table_side_effect
mock_fit.return_value = None, None, mock_table
# chi2 cost
with self.assertRaises(ValueError) as chi2:
self.alg_instance.find_good_peaks(self.x_values, [], 0.1, 5,
False, self.data_ws, 5)
# poisson cost
with self.assertRaises(ValueError) as poisson:
self.alg_instance.find_good_peaks(self.x_values, [], 0.1, 5,
True, self.data_ws, 5)
self.assertIn('Index = 0', chi2.exception.args)
self.assertNotIn('Index = 1', chi2.exception.args)
self.assertNotIn('Index = 0', poisson.exception.args)
self.assertIn('Index = 1', poisson.exception.args)
def test_find_good_peaks_returns_correct_peaks(self):
self.alg_instance._min_sigma = 1
self.alg_instance._max_sigma = 10
actual_peaks, peak_table, refit_peak_table = self.alg_instance.find_good_peaks(
self.x_values, self.peakids, 0, 5, False, self.data_ws, 5)
peak1 = peak_table.row(0)
peak2 = peak_table.row(1)
self.assertEquals(self.peakids, actual_peaks)
self.assertEqual(0, refit_peak_table.rowCount())
self.assertEqual(refit_peak_table.getColumnNames(),
peak_table.getColumnNames())
self.assertPeakFound(peak1, self.centre[0], self.height[0] + 10,
self.width[0], 0.05)
self.assertPeakFound(peak2, self.centre[1], self.height[1] + 10,
self.width[1], 0.05)
def test_find_peaks_is_called_if_scipy_version_higher_1_1_0(self):
mock_scipy = mock.MagicMock()
mock_scipy.__version__ = '1.1.0'
mock_scipy.signal.find_peaks.return_value = (self.peakids, {
'prominences':
self.peakids
})
with mock.patch.dict('sys.modules', scipy=mock_scipy):
self.alg_instance.process(self.x_values,
self.y_values,
raw_error=np.sqrt(self.y_values),
acceptance=0,
average_window=50,
bad_peak_to_consider=2,
use_poisson=False,
peak_width_estimate=5,
fit_to_baseline=False,
prog_reporter=mock.Mock())
self.assertEqual(2, mock_scipy.signal.find_peaks.call_count)
self.assertEqual(0, mock_scipy.signal.find_peaks_cwt.call_count)
def test_find_peaks_cwt_is_called_if_scipy_version_lower_1_1_0(self):
mock_scipy = mock.MagicMock()
mock_scipy.__version__ = '1.0.0'
mock_scipy.signal.find_peaks.return_value = (self.peakids, {
'prominences':
self.peakids
})
with mock.patch.dict('sys.modules', scipy=mock_scipy):
self.alg_instance.process(self.x_values,
self.y_values,
raw_error=np.sqrt(self.y_values),
acceptance=0,
average_window=50,
bad_peak_to_consider=2,
use_poisson=False,
peak_width_estimate=5,
fit_to_baseline=False,
prog_reporter=mock.Mock())
self.assertEqual(0, mock_scipy.signal.find_peaks.call_count)
self.assertEqual(1, mock_scipy.signal.find_peaks_cwt.call_count)
def test_process_calls_find_good_peaks(self):
with mock.patch(
'plugins.algorithms.WorkflowAlgorithms.FindPeaksAutomatic.CreateWorkspace'
) as mock_create_ws:
mock_create_ws.return_value = self.data_ws
self.alg_instance.find_good_peaks = mock.Mock()
self.alg_instance.process(self.x_values,
self.y_values,
raw_error=np.sqrt(self.y_values),
acceptance=0,
average_window=50,
bad_peak_to_consider=2,
use_poisson=False,
peak_width_estimate=5,
fit_to_baseline=False,
prog_reporter=mock.Mock())
base = self.alg_instance.average(self.y_values, 50)
base += self.alg_instance.average(self.y_values - base, 50)
flat = self.y_values - base
self.assertEqual(1, self.alg_instance.find_good_peaks.call_count)
self.alg_instance.find_good_peaks.asser_called_with(
self.x_values,
flat,
acceptance=0,
bad_peak_to_consider=2,
use_poisson=False,
fit_ws=self.data_ws,
peak_width_estimate=5)
def test_process_returns_the_return_value_of_find_good_peaks(self):
with mock.patch(
'plugins.algorithms.WorkflowAlgorithms.FindPeaksAutomatic.CreateWorkspace'
) as mock_create_ws:
mock_create_ws.return_value = self.data_ws
win_size = 500
actual_return = self.alg_instance.process(
self.x_values,
self.y_values,
raw_error=np.sqrt(self.y_values),
acceptance=0,
average_window=win_size,
bad_peak_to_consider=2,
use_poisson=False,
peak_width_estimate=5,
fit_to_baseline=False,
prog_reporter=mock.Mock())
import copy
actual_return = copy.deepcopy(actual_return)
base = self.alg_instance.average(self.y_values, win_size)
base += self.alg_instance.average(self.y_values - base, win_size)
expected_return = self.alg_instance.find_good_peaks(
self.x_values,
self.peakids,
acceptance=0,
bad_peak_to_consider=2,
use_poisson=False,
fit_ws=self.data_ws,
peak_width_estimate=5), base
self.assertEqual(expected_return[0][0], actual_return[0][0])
self.assertTableEqual(expected_return[0][1], actual_return[0][1])
np.testing.assert_almost_equal(expected_return[1],
actual_return[1])
def _assert_matplotlib_not_present(self, *args):
import sys
self.assertNotIn('matplotlib.pyplot', sys.modules)
# If matplotlib.pyplot is imported other tests fail on windows and ubuntu
def test_matplotlib_pyplot_is_not_imported(self):
self.alg_instance.dilation = mock.Mock(
side_effect=self._assert_matplotlib_not_present)
self.alg_instance.opening(self.y_values, 0)
def test_that_algorithm_finds_peaks_correctly(self):
FindPeaksAutomatic(
InputWorkspace=self.raw_ws,
SmoothWindow=500,
EstimatePeakSigma=5,
MinPeakSigma=3,
MaxPeakSigma=15,
)
peak_table = mtd['{}_{}'.format(self.raw_ws.getName(), 'properties')]
refit_peak_table = mtd['{}_{}'.format(self.raw_ws.getName(),
'refit_properties')]
self.assertEqual(2, peak_table.rowCount())
self.assertEqual(0, refit_peak_table.rowCount())
self.assertPeakFound(peak_table.row(0), self.centre[0], self.height[0],
self.width[0], 0.05)
self.assertPeakFound(peak_table.row(1), self.centre[1], self.height[1],
self.width[1], 0.05)