Source code for silx.gui.fit.FitWidget

# /*##########################################################################
#
# Copyright (c) 2004-2023 European Synchrotron Radiation Facility
#
# This file is part of the PyMca X-ray Fluorescence Toolkit developed at
# the ESRF by the Software group.
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in
# all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
# THE SOFTWARE.
#
# ######################################################################### */
"""This module provides a widget designed to configure and run a fitting
process with constraints on parameters.

The main class is :class:`FitWidget`. It relies on
:mod:`silx.math.fit.fitmanager`, which relies on :func:`silx.math.fit.leastsq`.

The user can choose between functions before running the fit. These function can
be user defined, or by default are loaded from
:mod:`silx.math.fit.fittheories`.
"""

__authors__ = ["V.A. Sole", "P. Knobel"]
__license__ = "MIT"
__date__ = "17/07/2018"

import logging
import sys
import traceback

from silx.math.fit import fittheories
from silx.math.fit import fitmanager, functions
from silx.gui import qt
from .FitWidgets import (
    FitActionsButtons,
    FitStatusLines,
    FitConfigWidget,
    ParametersTab,
)
from .FitConfig import getFitConfigDialog
from .BackgroundWidget import getBgDialog, BackgroundDialog

DEBUG = 0
_logger = logging.getLogger(__name__)


__authors__ = ["V.A. Sole", "P. Knobel"]
__license__ = "MIT"
__date__ = "30/11/2016"


[docs] class FitWidget(qt.QWidget): """This widget can be used to configure, run and display results of a fitting process. The standard steps for using this widget is to initialize it, then load the data to be fitted. Optionally, you can also load user defined fit theories. If you skip this step, a series of default fit functions will be presented (gaussian-like functions), and you can later load your custom fit theories from an external file using the GUI. A fit theory is a fit function and its associated features: - estimation function, - list of parameter names - numerical derivative algorithm - configuration widget Once the widget is up and running, the user may select a fit theory and a background theory, change configuration parameters specific to the theory run the estimation, set constraints on parameters and run the actual fit. The results are displayed in a table. .. image:: img/FitWidget.png """ sigFitWidgetSignal = qt.Signal(object) """This signal is emitted by the estimation and fit methods. It carries a dictionary with two items: - *event*: one of the following strings - *EstimateStarted*, - *FitStarted* - *EstimateFinished*, - *FitFinished* - *EstimateFailed* - *FitFailed* - *data*: None, or fit/estimate results (see documentation for :attr:`silx.math.fit.fitmanager.FitManager.fit_results`) """
[docs] def __init__( self, parent=None, title=None, fitmngr=None, enableconfig=True, enablestatus=True, enablebuttons=True, ): """ :param parent: Parent widget :param title: Window title :param fitmngr: User defined instance of :class:`silx.math.fit.fitmanager.FitManager`, or ``None`` :param enableconfig: If ``True``, activate widgets to modify the fit configuration (select between several fit functions or background functions, apply global constraints, peak search parameters…) :param enablestatus: If ``True``, add a fit status widget, to display a message when fit estimation is available and when fit results are available, as well as a measure of the fit error. :param enablebuttons: If ``True``, add buttons to run estimation and fitting. """ if title is None: title = "FitWidget" qt.QWidget.__init__(self, parent) self.setWindowTitle(title) layout = qt.QVBoxLayout(self) self.fitmanager = self._setFitManager(fitmngr) """Instance of :class:`FitManager`. This is the underlying data model of this FitWidget. If no custom theories are defined, the default ones from :mod:`silx.math.fit.fittheories` are imported. """ # reference fitmanager.configure method for direct access self.configure = self.fitmanager.configure self.fitconfig = self.fitmanager.fitconfig self.configdialogs = {} """This dictionary defines the fit configuration widgets associated with the fit theories in :attr:`fitmanager.theories` Keys must correspond to existing theory names, i.e. existing keys in :attr:`fitmanager.theories`. Values must be instances of QDialog widgets with an additional *output* attribute, a dictionary storing configuration parameters interpreted by the corresponding fit theory. The dialog can also define a *setDefault* method to initialize the widget values with values in a dictionary passed as a parameter. This will be executed first. In case the widget does not actually inherit :class:`QDialog`, it must at least implement the following methods (executed in this particular order): - :meth:`show`: should cause the widget to become visible to the user) - :meth:`exec`: should run while the user is interacting with the widget, interrupting the rest of the program. It should typically end (*return*) when the user clicks an *OK* or a *Cancel* button. - :meth:`result`: must return ``True`` if the new configuration in attribute :attr:`output` is to be accepted (user clicked *OK*), or return ``False`` if :attr:`output` is to be rejected (user clicked *Cancel*) To associate a custom configuration widget with a fit theory, use :meth:`associateConfigDialog`. E.g.:: fw = FitWidget() my_config_widget = MyGaussianConfigWidget(parent=fw) fw.associateConfigDialog(theory_name="Gaussians", config_widget=my_config_widget) """ self.bgconfigdialogs = {} """Same as :attr:`configdialogs`, except that the widget is associated with a background theory in :attr:`fitmanager.bgtheories`""" self._associateDefaultConfigDialogs() self.guiConfig = None """Configuration widget at the top of FitWidget, to select fit function, background function, and open an advanced configuration dialog.""" self.guiParameters = ParametersTab(self) """Table widget for display of fit parameters and constraints""" if enableconfig: self.guiConfig = FitConfigWidget(self) """Function selector and configuration widget""" self.guiConfig.FunConfigureButton.clicked.connect( self.__funConfigureGuiSlot ) self.guiConfig.BgConfigureButton.clicked.connect(self.__bgConfigureGuiSlot) self.guiConfig.WeightCheckBox.setChecked( self.fitconfig.get("WeightFlag", False) ) self.guiConfig.WeightCheckBox.stateChanged[int].connect(self.weightEvent) if qt.BINDING == "PyQt5": self.guiConfig.BkgComBox.activated[str].connect(self.bkgEvent) self.guiConfig.FunComBox.activated[str].connect(self.funEvent) else: # Qt6 self.guiConfig.BkgComBox.textActivated.connect(self.bkgEvent) self.guiConfig.FunComBox.textActivated.connect(self.funEvent) self._populateFunctions() layout.addWidget(self.guiConfig) layout.addWidget(self.guiParameters) if enablestatus: self.guistatus = FitStatusLines(self) """Status bar""" layout.addWidget(self.guistatus) if enablebuttons: self.guibuttons = FitActionsButtons(self) """Widget with estimate, start fit and dismiss buttons""" self.guibuttons.EstimateButton.clicked.connect(self.estimate) self.guibuttons.EstimateButton.setEnabled(False) self.guibuttons.StartFitButton.clicked.connect(self.startFit) self.guibuttons.StartFitButton.setEnabled(False) self.guibuttons.DismissButton.clicked.connect(self.dismiss) layout.addWidget(self.guibuttons)
def _setFitManager(self, fitinstance): """Initialize a :class:`FitManager` instance, to be assigned to :attr:`fitmanager`, or use a custom FitManager instance. :param fitinstance: Existing instance of FitManager, possibly customized by the user, or None to load a default instance.""" if isinstance(fitinstance, fitmanager.FitManager): # customized fitmngr = fitinstance else: # initialize default instance fitmngr = fitmanager.FitManager() # initialize the default fitting functions in case # none is present if not len(fitmngr.theories): fitmngr.loadtheories(fittheories) return fitmngr def _associateDefaultConfigDialogs(self): """Fill :attr:`bgconfigdialogs` and :attr:`configdialogs` by calling :meth:`associateConfigDialog` with default config dialog widgets. """ # associate silx.gui.fit.FitConfig with all theories # Users can later associate their own custom dialogs to # replace the default. configdialog = getFitConfigDialog(parent=self, default=self.fitconfig) for theory in self.fitmanager.theories: self.associateConfigDialog(theory, configdialog) for bgtheory in self.fitmanager.bgtheories: self.associateConfigDialog( bgtheory, configdialog, theory_is_background=True ) # associate silx.gui.fit.BackgroundWidget with Strip and Snip bgdialog = getBgDialog(parent=self, default=self.fitconfig) for bgtheory in ["Strip", "Snip"]: if bgtheory in self.fitmanager.bgtheories: self.associateConfigDialog( bgtheory, bgdialog, theory_is_background=True ) def _populateFunctions(self): """Fill combo-boxes with fit theories and background theories loaded by :attr:`fitmanager`. Run :meth:`fitmanager.configure` to ensure the custom configuration of the selected theory has been loaded into :attr:`fitconfig`""" for theory_name in self.fitmanager.bgtheories: self.guiConfig.BkgComBox.addItem(theory_name) self.guiConfig.BkgComBox.setItemData( self.guiConfig.BkgComBox.findText(theory_name), self.fitmanager.bgtheories[theory_name].description, qt.Qt.ToolTipRole, ) for theory_name in self.fitmanager.theories: self.guiConfig.FunComBox.addItem(theory_name) self.guiConfig.FunComBox.setItemData( self.guiConfig.FunComBox.findText(theory_name), self.fitmanager.theories[theory_name].description, qt.Qt.ToolTipRole, ) # - activate selected fit theory (if any) # - activate selected bg theory (if any) configuration = self.fitmanager.configure() if self.fitmanager.selectedtheory is None: # take the first one by default self.guiConfig.FunComBox.setCurrentIndex(1) self.funEvent(list(self.fitmanager.theories.keys())[0]) else: idx = list(self.fitmanager.theories).index(self.fitmanager.selectedtheory) self.guiConfig.FunComBox.setCurrentIndex(idx + 1) self.funEvent(self.fitmanager.selectedtheory) if self.fitmanager.selectedbg is None: self.guiConfig.BkgComBox.setCurrentIndex(1) self.bkgEvent(list(self.fitmanager.bgtheories.keys())[0]) else: idx = list(self.fitmanager.bgtheories).index(self.fitmanager.selectedbg) self.guiConfig.BkgComBox.setCurrentIndex(idx + 1) self.bkgEvent(self.fitmanager.selectedbg) configuration.update(self.configure())
[docs] def setData(self, x=None, y=None, sigmay=None, xmin=None, xmax=None): """Set data to be fitted. :param x: Abscissa data. If ``None``, :attr:`xdata`` is set to ``numpy.array([0.0, 1.0, 2.0, ..., len(y)-1])`` :type x: Sequence or numpy array or None :param y: The dependant data ``y = f(x)``. ``y`` must have the same shape as ``x`` if ``x`` is not ``None``. :type y: Sequence or numpy array or None :param sigmay: The uncertainties in the ``ydata`` array. These are used as weights in the least-squares problem. If ``None``, the uncertainties are assumed to be 1. :type sigmay: Sequence or numpy array or None :param xmin: Lower value of x values to use for fitting :param xmax: Upper value of x values to use for fitting """ if y is None: self.guibuttons.EstimateButton.setEnabled(False) self.guibuttons.StartFitButton.setEnabled(False) else: self.guibuttons.EstimateButton.setEnabled(True) self.guibuttons.StartFitButton.setEnabled(True) self.fitmanager.setdata(x=x, y=y, sigmay=sigmay, xmin=xmin, xmax=xmax) for config_dialog in self.bgconfigdialogs.values(): if isinstance(config_dialog, BackgroundDialog): config_dialog.setData(x, y, xmin=xmin, xmax=xmax)
[docs] def associateConfigDialog( self, theory_name, config_widget, theory_is_background=False ): """Associate an instance of custom configuration dialog widget to a fit theory or to a background theory. This adds or modifies an item in the correspondence table :attr:`configdialogs` or :attr:`bgconfigdialogs`. :param str theory_name: Name of fit theory. This must be a key of dict :attr:`fitmanager.theories` :param config_widget: Custom configuration widget. See documentation for :attr:`configdialogs` :param bool theory_is_background: If flag is *True*, add dialog to :attr:`bgconfigdialogs` rather than :attr:`configdialogs` (default). :raise: KeyError if parameter ``theory_name`` does not match an existing fit theory or background theory in :attr:`fitmanager`. :raise: AttributeError if the widget does not implement the mandatory methods (*show*, *exec*, *result*, *setDefault*) or the mandatory attribute (*output*). """ theories = ( self.fitmanager.bgtheories if theory_is_background else self.fitmanager.theories ) if theory_name not in theories: raise KeyError("%s does not match an existing fitmanager theory") if config_widget is not None: if not hasattr(config_widget, "exec") and not hasattr( config_widget, "exec_" ): raise AttributeError( "Custom configuration widget must define exec or exec_" ) for mandatory_attr in ["show", "result", "output"]: if not hasattr(config_widget, mandatory_attr): raise AttributeError( "Custom configuration widget must define " + "attribute or method " + mandatory_attr ) if theory_is_background: self.bgconfigdialogs[theory_name] = config_widget else: self.configdialogs[theory_name] = config_widget
def _emitSignal(self, ddict): """Emit pyqtSignal after estimation completed (``ddict = {'event': 'EstimateFinished', 'data': fit_results}``) and after fit completed (``ddict = {'event': 'FitFinished', 'data': fit_results}``)""" self.sigFitWidgetSignal.emit(ddict) def __funConfigureGuiSlot(self): """Open an advanced configuration dialog widget""" self.__configureGui(dialog_type="function") def __bgConfigureGuiSlot(self): """Open an advanced configuration dialog widget""" self.__configureGui(dialog_type="background") def __configureGui(self, newconfiguration=None, dialog_type="function"): """Open an advanced configuration dialog widget to get a configuration dictionary, or use a supplied configuration dictionary. Call :meth:`configure` with this dictionary as a parameter. Update the gui accordingly. Reinitialize the fit results in the table and in :attr:`fitmanager`. :param newconfiguration: User supplied configuration dictionary. If ``None``, open a dialog widget that returns a dictionary.""" configuration = self.configure() # get new dictionary if newconfiguration is None: newconfiguration = self.configureDialog(configuration, dialog_type) # update configuration configuration.update(self.configure(**newconfiguration)) # set fit function theory try: i = 1 + list(self.fitmanager.theories.keys()).index( self.fitmanager.selectedtheory ) self.guiConfig.FunComBox.setCurrentIndex(i) self.funEvent(self.fitmanager.selectedtheory) except ValueError: _logger.error("Function not in list %s", self.fitmanager.selectedtheory) self.funEvent(list(self.fitmanager.theories.keys())[0]) # current background try: i = 1 + list(self.fitmanager.bgtheories.keys()).index( self.fitmanager.selectedbg ) self.guiConfig.BkgComBox.setCurrentIndex(i) self.bkgEvent(self.fitmanager.selectedbg) except ValueError: _logger.error("Background not in list %s", self.fitmanager.selectedbg) self.bkgEvent(list(self.fitmanager.bgtheories.keys())[0]) # update the Gui self.__initialParameters() def configureDialog(self, oldconfiguration, dialog_type="function"): """Display a dialog, allowing the user to define fit configuration parameters. By default, a common dialog is used for all fit theories. But if the defined a custom dialog using :meth:`associateConfigDialog`, it is used instead. :param dict oldconfiguration: Dictionary containing previous configuration :param str dialog_type: "function" or "background" :return: User defined parameters in a dictionary """ newconfiguration = {} newconfiguration.update(oldconfiguration) if dialog_type == "function": theory = self.fitmanager.selectedtheory configdialog = self.configdialogs[theory] elif dialog_type == "background": theory = self.fitmanager.selectedbg configdialog = self.bgconfigdialogs[theory] # this should only happen if a user specifically associates None # with a theory, to have no configuration option if configdialog is None: return {} # update state of configdialog before showing it if hasattr(configdialog, "setDefault"): configdialog.setDefault(newconfiguration) configdialog.show() if hasattr(configdialog, "exec"): configdialog.exec() else: # Qt5 compatibility configdialog.exec_() if configdialog.result(): newconfiguration.update(configdialog.output) return newconfiguration def estimate(self): """Run parameter estimation function then emit :attr:`sigFitWidgetSignal` with a dictionary containing a status message and a list of fit parameters estimations in the format defined in :attr:`silx.math.fit.fitmanager.FitManager.fit_results` The emitted dictionary has an *"event"* key that can have following values: - *'EstimateStarted'* - *'EstimateFailed'* - *'EstimateFinished'* """ try: theory_name = self.fitmanager.selectedtheory estimation_function = self.fitmanager.theories[theory_name].estimate if estimation_function is not None: ddict = {"event": "EstimateStarted", "data": None} self._emitSignal(ddict) self.fitmanager.estimate(callback=self.fitStatus) else: msg = qt.QMessageBox(self) msg.setIcon(qt.QMessageBox.Information) text = "Function does not define a way to estimate\n" text += "the initial parameters. Please, fill them\n" text += "yourself in the table and press Start Fit\n" msg.setText(text) msg.setWindowTitle("FitWidget Message") msg.exec() return except Exception as e: # noqa (we want to catch and report all errors) _logger.warning("Estimate error: %s", traceback.format_exc()) msg = qt.QMessageBox(self) msg.setIcon(qt.QMessageBox.Critical) msg.setWindowTitle("Estimate Error") msg.setText("Error on estimate: %s" % e) msg.exec() ddict = {"event": "EstimateFailed", "data": None} self._emitSignal(ddict) return self.guiParameters.fillFromFit(self.fitmanager.fit_results, view="Fit") self.guiParameters.removeAllViews(keep="Fit") ddict = {"event": "EstimateFinished", "data": self.fitmanager.fit_results} self._emitSignal(ddict) def startFit(self): """Run fit, then emit :attr:`sigFitWidgetSignal` with a dictionary containing a status message and a list of fit parameters results in the format defined in :attr:`silx.math.fit.fitmanager.FitManager.fit_results` The emitted dictionary has an *"event"* key that can have following values: - *'FitStarted'* - *'FitFailed'* - *'FitFinished'* """ self.fitmanager.fit_results = self.guiParameters.getFitResults() try: ddict = {"event": "FitStarted", "data": None} self._emitSignal(ddict) self.fitmanager.runfit(callback=self.fitStatus) except Exception as e: # noqa (we want to catch and report all errors) _logger.warning("Estimate error: %s", traceback.format_exc()) msg = qt.QMessageBox(self) msg.setIcon(qt.QMessageBox.Critical) msg.setWindowTitle("Fit Error") msg.setText("Error on Fit: %s" % e) msg.exec() ddict = {"event": "FitFailed", "data": None} self._emitSignal(ddict) return self.guiParameters.fillFromFit(self.fitmanager.fit_results, view="Fit") self.guiParameters.removeAllViews(keep="Fit") ddict = {"event": "FitFinished", "data": self.fitmanager.fit_results} self._emitSignal(ddict) return def bkgEvent(self, bgtheory): """Select background theory, then reinitialize parameters""" bgtheory = str(bgtheory) if bgtheory in self.fitmanager.bgtheories: self.fitmanager.setbackground(bgtheory) else: functionsfile = qt.QFileDialog.getOpenFileName( self, "Select python module with your function(s)", "", "Python Files (*.py);;All Files (*)", ) if len(functionsfile): try: self.fitmanager.loadbgtheories(functionsfile) except ImportError: qt.QMessageBox.critical(self, "ERROR", "Function not imported") return else: # empty the ComboBox while self.guiConfig.BkgComBox.count() > 1: self.guiConfig.BkgComBox.removeItem(1) # and fill it again for key in self.fitmanager.bgtheories: self.guiConfig.BkgComBox.addItem(str(key)) i = 1 + list(self.fitmanager.bgtheories.keys()).index( self.fitmanager.selectedbg ) self.guiConfig.BkgComBox.setCurrentIndex(i) self.__initialParameters() def funEvent(self, theoryname): """Select a fit theory to be used for fitting. If this theory exists in :attr:`fitmanager`, use it. Then, reinitialize table. :param theoryname: Name of the fit theory to use for fitting. If this theory exists in :attr:`fitmanager`, use it. Else, open a file dialog to open a custom fit function definition file with :meth:`fitmanager.loadtheories`. """ theoryname = str(theoryname) if theoryname in self.fitmanager.theories: self.fitmanager.settheory(theoryname) else: # open a load file dialog functionsfile = qt.QFileDialog.getOpenFileName( self, "Select python module with your function(s)", "", "Python Files (*.py);;All Files (*)", ) if len(functionsfile): try: self.fitmanager.loadtheories(functionsfile) except ImportError: qt.QMessageBox.critical(self, "ERROR", "Function not imported") return else: # empty the ComboBox while self.guiConfig.FunComBox.count() > 1: self.guiConfig.FunComBox.removeItem(1) # and fill it again for key in self.fitmanager.theories: self.guiConfig.FunComBox.addItem(str(key)) i = 1 + list(self.fitmanager.theories.keys()).index( self.fitmanager.selectedtheory ) self.guiConfig.FunComBox.setCurrentIndex(i) self.__initialParameters() def weightEvent(self, flag): """This is called when WeightCheckBox is clicked, to configure the *WeightFlag* field in :attr:`fitmanager.fitconfig` and set weights in the least-square problem.""" self.configure(WeightFlag=flag) if flag: self.fitmanager.enableweight() else: # set weights back to 1 self.fitmanager.disableweight() def __initialParameters(self): """Fill the fit parameters names with names of the parameters of the selected background theory and the selected fit theory. Initialize :attr:`fitmanager.fit_results` with these names, and initialize the table with them. This creates a view called "Fit" in :attr:`guiParameters`""" self.fitmanager.parameter_names = [] self.fitmanager.fit_results = [] for pname in self.fitmanager.bgtheories[self.fitmanager.selectedbg].parameters: self.fitmanager.parameter_names.append(pname) self.fitmanager.fit_results.append( { "name": pname, "estimation": 0, "group": 0, "code": "FREE", "cons1": 0, "cons2": 0, "fitresult": 0.0, "sigma": 0.0, "xmin": None, "xmax": None, } ) if self.fitmanager.selectedtheory is not None: theory = self.fitmanager.selectedtheory for pname in self.fitmanager.theories[theory].parameters: self.fitmanager.parameter_names.append(pname + "1") self.fitmanager.fit_results.append( { "name": pname + "1", "estimation": 0, "group": 1, "code": "FREE", "cons1": 0, "cons2": 0, "fitresult": 0.0, "sigma": 0.0, "xmin": None, "xmax": None, } ) self.guiParameters.fillFromFit(self.fitmanager.fit_results, view="Fit") def fitStatus(self, data): """Set *status* and *chisq* in status bar""" if "chisq" in data: if data["chisq"] is None: self.guistatus.ChisqLine.setText(" ") else: chisq = data["chisq"] self.guistatus.ChisqLine.setText("%6.2f" % chisq) if "status" in data: status = data["status"] self.guistatus.StatusLine.setText(str(status)) def dismiss(self): """Close FitWidget""" self.close()
if __name__ == "__main__": import numpy x = numpy.arange(1500).astype(numpy.float64) constant_bg = 3.14 p = [ 1000, 100.0, 30.0, 500, 300.0, 25.0, 1700, 500.0, 35.0, 750, 700.0, 30.0, 1234, 900.0, 29.5, 302, 1100.0, 30.5, 75, 1300.0, 21.0, ] y = functions.sum_gauss(x, *p) + constant_bg a = qt.QApplication(sys.argv) w = FitWidget() w.setData(x=x, y=y) w.show() a.exec()