fitting (#43)

adjustments to fit to see if fit is at least running, could help with #39

Co-authored-by: Dominik Demuth <dominik.demuth@physik.tu-darmstadt.de>
Reviewed-on: #43

src/gui_qt/lib/utils.py

@@ -58,7 +58,7 @@ class RdBuCMap:
         elif val < self.min:
             col = QtGui.QColor.fromRgb(*RdBuCMap._rdbu[-1])
         else:
-            col = QtGui.QColor.fromRgb(*(float(self.spline[i](val)) for i in range(3)))
+            col = QtGui.QColor.fromRgb(*(int(self.spline[i](val)) for i in range(3)))
 
         return col
 

src/gui_qt/main/mainwindow.py

@@ -66,7 +66,6 @@ class NMRMainWindow(QtWidgets.QMainWindow, Ui_BaseWindow):
         self._block_window_change = False
 
         self.fname = None
-        self.tim = QtCore.QTimer()
 
         self.settings = QtCore.QSettings('NMREVal', 'settings')
         self._init_gui()
@@ -887,6 +886,11 @@ class NMRMainWindow(QtWidgets.QMainWindow, Ui_BaseWindow):
 
         self.fit_dialog.fit_button.setEnabled(False)
         self.management.start_fit(parameter, links, fit_options)
+        self.status.setText('Fit running...'.format(self.management.fitter.step))
+        tim = QtCore.QTimer()
+        tim.setInterval(500)
+        tim.timeout.connect(lambda: self.status.setText(f'Fit running... ({self.management.fitter.step} evaluations)'))
+        tim.start(500)
 
     @QtCore.pyqtSlot(dict, int, bool)
     def show_fit_preview(self, funcs: dict, num: int, show: bool):
@@ -910,6 +914,7 @@ class NMRMainWindow(QtWidgets.QMainWindow, Ui_BaseWindow):
     @QtCore.pyqtSlot(list)
     def show_fit_results(self, results: list):
         self.fit_dialog.fit_button.setEnabled(True)
+        self.status.setText('')
         if results:
             res_dialog = QFitResult(results, self.management, parent=self)
             res_dialog.add_graphs(self.management.graphs.list())

src/gui_qt/main/management.py

@@ -398,7 +398,7 @@ class UpperManagement(QtCore.QObject):
 
         self.__fit_options = (parameter, links, fit_options)
 
-        fitter = FitRoutine()
+        self.fitter = FitRoutine()
         models = {}
         fit_limits = fit_options['limits']
         fit_mode = fit_options['fit_mode']
@@ -442,18 +442,18 @@ class UpperManagement(QtCore.QObject):
                                 lb=model_p['lb'], ub=model_p['ub'],
                                 fun_kwargs=set_params[1])
 
-                fitter.add_data(d)
+                self.fitter.add_data(d)
 
             model_globs = model_p['glob']
             if model_globs:
                 m.set_global_parameter(**model_p['glob'])
 
         for links_i in links:
-            fitter.set_link_parameter((models[links_i[0]], links_i[1]),
+            self.fitter.set_link_parameter((models[links_i[0]], links_i[1]),
-                                      (models[links_i[2]], links_i[3]))
+                                           (models[links_i[2]], links_i[3]))
 
         with busy_cursor():
-            self.fit_worker = FitWorker(fitter, fit_mode)
+            self.fit_worker = FitWorker(self.fitter, fit_mode)
             self.fit_thread = QtCore.QThread()
             self.fit_worker.moveToThread(self.fit_thread)
 
@@ -469,7 +469,7 @@ class UpperManagement(QtCore.QObject):
 
     @QtCore.pyqtSlot(list, bool)
     def end_fit(self, result: list, success: bool):
-        print('FIT FINISHED')
+        logger.info('FIT FINISHED')
         if success:
             self.fitFinished.emit(result)
         else:

src/nmreval/fit/minimizer.py

@@ -1,3 +1,4 @@
+import time
 import warnings
 from itertools import product
 
@@ -31,6 +32,7 @@ class FitRoutine(object):
         self.result = []
         self.linked = []
         self._abort = False
+        self.step = 0
 
     def add_data(self, x, y=None, we=None, idx=None):
         if isinstance(x, Data):
@@ -165,7 +167,7 @@ class FitRoutine(object):
             self.find_paths(neighbor, graph, coupled_nodes, visited_nodes)
 
     def abort(self):
-        print('ABORT ???')
+        logger.info('Fit aborted by user')
         self._abort = True
 
     def run(self, mode='lsq'):
@@ -311,14 +313,17 @@ class FitRoutine(object):
         return r
 
     def _least_squares_single(self, data, p0, lb, ub, var):
+        self.step = 0
+
         def cost(p):
+            self.step += 1
             if self._abort:
                 raise FitAbortException(f'Fit aborted by user')
 
             return self.__cost_scipy(p, data, var, data.para_keys)
 
         with np.errstate(all='ignore'):
-            res = optimize.least_squares(cost, p0, bounds=(lb, ub), max_nfev=1000 * len(p0))
+            res = optimize.least_squares(cost, p0, bounds=(lb, ub), max_nfev=500 * len(p0))
 
         err, corr, partial_corr = self._calc_error(res.jac, np.sum(res.fun**2), *res.jac.shape)
         self.make_results(data, res.x, var, data.para_keys, res.jac.shape,
@@ -326,12 +331,13 @@ class FitRoutine(object):
 
     def _least_squares_global(self, data, p0, lb, ub, var, data_pars):
         def cost(p):
+            self.step += 1
             if self._abort:
                 raise FitAbortException(f'Fit aborted by user')
             return self.__cost_scipy_glob(p, data, var, data_pars)
 
         with np.errstate(all='ignore'):
-            res = optimize.least_squares(cost, p0, bounds=(lb, ub), max_nfev=1000 * len(p0))
+            res = optimize.least_squares(cost, p0, bounds=(lb, ub), max_nfev=500 * len(p0))
 
         err, corr, partial_corr = self._calc_error(res.jac, np.sum(res.fun**2), *res.jac.shape)
         for v, var_pars_k in zip(data, data_pars):
@@ -340,25 +346,27 @@ class FitRoutine(object):
 
     def _nm_single(self, data, p0, lb, ub, var):
         def cost(p):
+            self.step += 1
             if self._abort:
                 raise FitAbortException(f'Fit aborted by user')
             return (self.__cost_scipy(p, data, var, data.para_keys)**2).sum()
 
         with np.errstate(all='ignore'):
             res = optimize.minimize(cost, p0, bounds=[(b1, b2) for (b1, b2) in zip(lb, ub)],
-                                    method='Nelder-Mead', options={'maxiter': 1000 * len(p0)})
+                                    method='Nelder-Mead', options={'maxiter': 500 * len(p0)})
 
         self.make_results(data, res.x, var, data.para_keys, (len(data), len(p0)))
 
     def _nm_global(self, data, p0, lb, ub, var, data_pars):
         def cost(p):
+            self.step += 1
             if self._abort:
                 raise FitAbortException(f'Fit aborted by user')
             return (self.__cost_scipy_glob(p, data, var, data_pars)**2).sum()
 
         with np.errstate(all='ignore'):
             res = optimize.minimize(cost, p0, bounds=[(b1, b2) for (b1, b2) in zip(lb, ub)],
-                                    method='Nelder-Mead', options={'maxiter': 1000 * len(p0)})
+                                    method='Nelder-Mead', options={'maxiter': 500 * len(p0)})
 
         for v, var_pars_k in zip(data, data_pars):
             self.make_results(v, res.x, var, var_pars_k, (sum(len(d) for d in data), len(p0)))
@@ -367,6 +375,7 @@ class FitRoutine(object):
         odr_data = odr.Data(data.x, data.y)
 
         def func(p, _):
+            self.step += 1
             if self._abort:
                 raise FitAbortException(f'Fit aborted by user')
             return self.__cost_odr(p, data, var_pars, data.para_keys)
@@ -390,6 +399,7 @@ class FitRoutine(object):
 
     def _odr_global(self, data, p0, var, data_pars):
         def func(p, _):
+            self.step += 1
             if self._abort:
                 raise FitAbortException(f'Fit aborted by user')
             return self.__cost_odr_glob(p, data, var, data_pars)

src/nmreval/io/asciireader.py

@@ -155,8 +155,8 @@ class AsciiReader:
                     # more than one axis, append column number
                     kwargs['name'] = filename + '_' + str(y[j-1])
 
-                if j+num_y+1 < raw_data.shape[2]:
+                if j+num_y < raw_data.shape[2]:
-                    kwargs['y_err'] = raw_data[i, j+num_y+1]
+                    kwargs['y_err'] = raw_data[i, :, j+num_y]
 
                 imported_sets.append(cls(x=raw_data[i, :, 0], y=raw_data[i, :, j:j+single_len].T, **kwargs))