def __fit_zero_loss_peak(self, spectrum_xdata: DataAndMetadata.DataAndMetadata) -> DataAndMetadata.DataAndMetadata:
reference_frame = Calibration.ReferenceFrameAxis(spectrum_xdata.datum_dimensional_calibrations[0], spectrum_xdata.datum_dimension_shape[0])
z = reference_frame.convert_to_pixel(Calibration.Coordinate(Calibration.CoordinateType.CALIBRATED, 0)).int_value
calibration = copy.deepcopy(spectrum_xdata.datum_dimensional_calibrations[0])
ys = spectrum_xdata.data
if spectrum_xdata.is_navigable:
calibrations = list(copy.deepcopy(spectrum_xdata.navigation_dimensional_calibrations)) + [calibration]
yss = numpy.reshape(ys, (numpy.product(ys.shape[:-1]),) + (ys.shape[-1],))
fit_data = self._perform_fits(yss, z)
data_descriptor = DataAndMetadata.DataDescriptor(False, spectrum_xdata.navigation_dimension_count,
spectrum_xdata.datum_dimension_count)
model_xdata = DataAndMetadata.new_data_and_metadata(
numpy.reshape(fit_data, ys.shape[:-1] + (ys.shape[-1],)),
data_descriptor=data_descriptor,
dimensional_calibrations=calibrations,
intensity_calibration=spectrum_xdata.intensity_calibration)
else:
poly_data = self._perform_fit(ys, z)
model_xdata = DataAndMetadata.new_data_and_metadata(poly_data, dimensional_calibrations=[calibration],
intensity_calibration=spectrum_xdata.intensity_calibration)
return model_xdata
def execute(self, eels_spectrum_data_item, background_model,
fit_interval_graphics, **kwargs) -> None:
try:
spectrum_xdata = eels_spectrum_data_item.xdata
assert spectrum_xdata.is_datum_1d
assert spectrum_xdata.datum_dimensional_calibrations[
0].units == "eV"
eels_spectrum_xdata = spectrum_xdata
# fit_interval_graphics.interval returns normalized coordinates. create calibrated intervals.
fit_intervals = list()
for fit_interval_graphic in fit_interval_graphics:
fit_interval = Calibration.CalibratedInterval(
Calibration.Coordinate(
Calibration.CoordinateType.NORMALIZED,
fit_interval_graphic.interval[0]),
Calibration.Coordinate(
Calibration.CoordinateType.NORMALIZED,
fit_interval_graphic.interval[1]))
fit_intervals.append(fit_interval)
fit_minimum = min(
[fit_interval.start.value for fit_interval in fit_intervals])
signal_interval = Calibration.CalibratedInterval(
Calibration.Coordinate(Calibration.CoordinateType.NORMALIZED,
fit_minimum),
Calibration.Coordinate(Calibration.CoordinateType.NORMALIZED,
1.0))
reference_frame = Calibration.ReferenceFrameAxis(
eels_spectrum_xdata.datum_dimensional_calibrations[0],
eels_spectrum_xdata.datum_dimension_shape[0])
signal_xdata = Core.get_calibrated_interval_slice(
eels_spectrum_xdata, reference_frame, signal_interval)
background_xdata = None
subtracted_xdata = None
if background_model._data_structure.entity:
entity_id = background_model._data_structure.entity.entity_type.entity_id
for component in Registry.get_components_by_type(
"background-model"):
if entity_id == component.background_model_id:
fit_result = component.fit_background(
spectrum_xdata=spectrum_xdata,
fit_intervals=fit_intervals,
background_interval=signal_interval)
background_xdata = fit_result["background_model"]
# use 'or' to avoid doing subtraction if subtracted_spectrum already present
subtracted_xdata = fit_result.get(
"subtracted_spectrum",
None) or Core.calibrated_subtract_spectrum(
spectrum_xdata, background_xdata)
if background_xdata is None:
background_xdata = DataAndMetadata.new_data_and_metadata(
numpy.zeros_like(signal_xdata.data),
intensity_calibration=signal_xdata.intensity_calibration,
dimensional_calibrations=signal_xdata.
dimensional_calibrations)
if subtracted_xdata is None:
subtracted_xdata = DataAndMetadata.new_data_and_metadata(
signal_xdata.data,
intensity_calibration=signal_xdata.intensity_calibration,
dimensional_calibrations=signal_xdata.
dimensional_calibrations)
self.__background_xdata = background_xdata
self.__subtracted_xdata = subtracted_xdata
except Exception as e:
import traceback
print(traceback.format_exc())
print(e)
raise
def __fit_background(
self, spectrum_xdata: DataAndMetadata.DataAndMetadata,
fit_intervals: typing.Sequence[Calibration.CalibratedInterval],
background_interval: Calibration.CalibratedInterval
) -> DataAndMetadata.DataAndMetadata:
reference_frame = Calibration.ReferenceFrameAxis(
spectrum_xdata.datum_dimensional_calibrations[0],
spectrum_xdata.datum_dimension_shape[0])
# fit polynomial to the data
xs = numpy.concatenate([
Core.get_calibrated_interval_domain(reference_frame, fit_interval)
for fit_interval in fit_intervals
])
if len(fit_intervals) > 1:
ys = numpy.concatenate([
Core.get_calibrated_interval_slice(spectrum_xdata,
reference_frame,
fit_interval).data
for fit_interval in fit_intervals
])
else:
ys = Core.get_calibrated_interval_slice(spectrum_xdata,
reference_frame,
fit_intervals[0]).data
# generate background model data from the series
n = reference_frame.convert_to_pixel(
background_interval.end
).int_value - reference_frame.convert_to_pixel(
background_interval.start).int_value
interval_start = reference_frame.convert_to_calibrated(
background_interval.start).value
interval_end = reference_frame.convert_to_calibrated(
background_interval.end).value
interval_end -= (interval_end - interval_start
) / n # n samples at the left edges of each pixel
calibration = copy.deepcopy(
spectrum_xdata.datum_dimensional_calibrations[0])
calibration.offset = reference_frame.convert_to_calibrated(
background_interval.start).value
fs = numpy.linspace(interval_start, interval_end, n)
if spectrum_xdata.is_navigable:
calibrations = list(
copy.deepcopy(
spectrum_xdata.navigation_dimensional_calibrations)) + [
calibration
]
yss = numpy.reshape(ys, (numpy.product(ys.shape[:-1]), ) +
(ys.shape[-1], ))
fit_data = self._perform_fits(xs, yss, fs)
data_descriptor = DataAndMetadata.DataDescriptor(
False, spectrum_xdata.navigation_dimension_count,
spectrum_xdata.datum_dimension_count)
background_xdata = DataAndMetadata.new_data_and_metadata(
numpy.reshape(fit_data, ys.shape[:-1] + (n, )),
data_descriptor=data_descriptor,
dimensional_calibrations=calibrations,
intensity_calibration=spectrum_xdata.intensity_calibration)
else:
poly_data = self._perform_fit(xs, ys, fs)
background_xdata = DataAndMetadata.new_data_and_metadata(
poly_data,
dimensional_calibrations=[calibration],
intensity_calibration=spectrum_xdata.intensity_calibration)
return background_xdata
