Merge pull request #831 from HEXRD/remove-wppf-parameters-class

Remove WPPF Parameters class

Author: psavery

hexrd/wppf/LeBailCalibration.py

@@ -12,8 +12,6 @@
 calc_Iobs_pvpink
 from hexrd.wppf.spectrum import Spectrum
 from hexrd.wppf import wppfsupport, LeBail
-from hexrd.wppf.parameters import Parameters
-from lmfit import Parameters as Parameters_lmfit
 from hexrd.wppf.phase import Phases_LeBail, Material_LeBail
 from hexrd.imageutil import snip1d, snip1d_quad
 from hexrd.material import Material
@@ -773,16 +771,16 @@ def params(self, param_info):
 
         if(param_info is not None):
             pl = wppfsupport._generate_default_parameters_LeBail(
-                self.phases, self.peakshape, ptype="lmfit")
+                self.phases, self.peakshape)
             self.lebail_param_list = [p for p in pl]
-            if(isinstance(param_info, Parameters_lmfit)):
+            if isinstance(param_info, lmfit.Parameters):
                 """
                 directly passing the parameter class
                 """
                 self._params = param_info
                 params = param_info
             else:
-                params = Parameters_lmfit()
+                params = lmfit.Parameters()
 
                 if(isinstance(param_info, dict)):
                     """
@@ -814,7 +812,7 @@ def params(self, param_info):
         else:
 
             params = wppfsupport._generate_default_parameters_LeBail(
-                self.phases, self.peakshape, ptype="lmfit")
+                self.phases, self.peakshape)
             self.lebail_param_list = [p for p in params]
             wppfsupport._add_detector_geometry(params, self.instrument)
             if "chebyshev" in self.bkgmethod.keys():
@@ -1157,7 +1155,7 @@ def params(self, params):
         set the local value of the parameters to the
         global values from the calibrator class
         """
-        if isinstance(params, Parameters_lmfit):
+        if isinstance(params, lmfit.Parameters):
             if hasattr(self, '_params'):
                 for p in params:
                     if (p in self.lebail_param_list) or (self.name in p):
@@ -1170,7 +1168,7 @@ def params(self, params):
                 xn, wn = roots_legendre(16)
                 self.xn = xn[8:]
                 self.wn = wn[8:]
-                self._params = Parameters_lmfit()
+                self._params = lmfit.Parameters()
                 for p in params:
                     if (p in self.lebail_param_list) or (self.name in p):
                         self._params.add(name=p,

hexrd/wppf/RietveldHEDM.py

@@ -407,7 +407,7 @@ def params(self, param_info):
 
         if(param_info is not None):
             pl = wppfsupport._generate_default_parameters_Rietveld(
-                self.phases, self.peakshape, ptype="lmfit")
+                self.phases, self.peakshape)
             self.rietveld_param_list = [p for p in pl]
             if(isinstance(param_info, Parameters_lmfit)):
                 """

hexrd/wppf/WPPF.py

@@ -33,7 +33,6 @@
 )
 from hexrd.wppf import wppfsupport
 from hexrd.wppf.spectrum import Spectrum
-from hexrd.wppf.parameters import Parameters
 from hexrd.wppf.phase import (
     Phases_LeBail,
     Phases_Rietveld,
@@ -104,6 +103,7 @@ def __init__(
         intensity_init=None,
         peakshape="pvfcj",
         amorphous_model=None,
+        reset_background_params=True,
     ):
 
         self.peakshape = peakshape
@@ -134,6 +134,12 @@ def __init__(
         self.Rwplist = np.empty([0])
         self.gofFlist = np.empty([0])
 
+        if reset_background_params:
+            # Reset background parameters to their initial values
+            self.reset_background_params()
+        else:
+            self._update_bkg(self.params)
+
     def __str__(self):
         resstr = "<LeBail Fit class>\nParameters of \
         the model are as follows:\n"
@@ -654,16 +660,16 @@ class too
 
         return errvec
 
-    def initialize_lmfit_parameters(self):
-
-        params = lmfit.Parameters()
-
-        for p in self.params:
-            par = self.params[p]
-            if par.vary:
-                params.add(p, value=par.value, min=par.lb, max=par.ub)
+    def reset_background_params(self):
+        # Reset background parameters to their initial values
+        if "chebyshev" not in self.bkgmethod:
+            return
 
-        return params
+        params = self.params
+        for i in range(len(self.bkg_coef)):
+            name = f'bkg_{i}'
+            if name in params:
+                params[name].value = self.bkg_coef[i]
 
     def update_parameters(self):
 
@@ -704,10 +710,7 @@ def Refine(self):
         >> @DETAILS: this routine performs the least squares refinement for all
                      variables which are allowed to be varied.
         """
-
-        params = self.initialize_lmfit_parameters()
-
-        if len(params) > 0:
+        if self.num_vary > 0:
             fdict = {
                 "ftol": 1e-6,
                 "xtol": 1e-6,
@@ -717,7 +720,7 @@ def Refine(self):
                 "method": "trf",
                 "jac": "2-point",
             }
-            fitter = lmfit.Minimizer(self.calcRwp, params)
+            fitter = lmfit.Minimizer(self.calcRwp, self.params)
 
             res = fitter.least_squares(**fdict)
             return res
@@ -743,8 +746,7 @@ def updatespectrum(self):
         simulated and experimental spectra
         """
         # ???: is this supposed to return something, or is it incomplete?
-        params = self.initialize_lmfit_parameters()
-        errvec = self.calcRwp(params)
+        errvec = self.calcRwp(self.params)
 
     def _update_shkl(self, params):
         """
@@ -1257,6 +1259,10 @@ def weights(self):
     def params(self):
         return self._params
 
+    @property
+    def num_vary(self) -> int:
+        return sum(x.vary for x in self.params.values())
+
     @property
     def init_bkg(self):
         degree = self.bkgmethod["chebyshev"]
@@ -1291,15 +1297,15 @@ def params(self, param_info):
         self.wn = wn[8:]
 
         if param_info is not None:
-            if isinstance(param_info, Parameters):
+            if isinstance(param_info, lmfit.Parameters):
                 """
                 directly passing the parameter class
                 """
                 self._params = param_info
                 params = param_info
 
             else:
-                params = Parameters()
+                params = lmfit.Parameters()
 
                 if isinstance(param_info, dict):
                     """
@@ -1309,8 +1315,8 @@ def params(self, param_info):
                         params.add(
                             k,
                             value=float(v[0]),
-                            lb=float(v[1]),
-                            ub=float(v[2]),
+                            min=float(v[1]),
+                            max=float(v[2]),
                             vary=bool(v[3]),
                         )
 
@@ -1792,6 +1798,7 @@ def __init__(
         particle_size=1.0,
         phi=0.0,
         amorphous_model=None,
+        reset_background_params=True,
     ):
 
         self.bkgmethod = bkgmethod
@@ -1823,6 +1830,12 @@ def __init__(
         self.Rwplist = np.empty([0])
         self.gofFlist = np.empty([0])
 
+        if reset_background_params:
+            # Reset background parameters to their initial values
+            self.reset_background_params()
+        else:
+            self._update_bkg(self.params)
+
     def __str__(self):
         resstr = "<Rietveld Fit class>\nParameters of \
         the model are as follows:\n"
@@ -2187,16 +2200,15 @@ def calcRwp(self, params):
 
         return errvec
 
-    def initialize_lmfit_parameters(self):
-
-        params = lmfit.Parameters()
-
-        for p in self.params:
-            par = self.params[p]
-            if par.vary:
-                params.add(p, value=par.value, min=par.lb, max=par.ub)
+    def reset_background_params(self):
+        if "chebyshev" not in self.bkgmethod:
+            return
 
-        return params
+        params = self.params
+        for i in range(len(self.bkg_coef)):
+            name = f'bkg_{i}'
+            if name in params:
+                params[name].value = self.bkg_coef[i]
 
     def update_parameters(self):
 
@@ -2212,10 +2224,7 @@ def Refine(self):
         >> @DETAILS: this routine performs the least squares refinement for all
                      variables that are allowed to be varied.
         """
-
-        params = self.initialize_lmfit_parameters()
-
-        if len(params) > 0:
+        if self.num_vary > 0:
             fdict = {
                 "ftol": 1e-6,
                 "xtol": 1e-6,
@@ -2226,7 +2235,7 @@ def Refine(self):
                 "jac": "2-point",
             }
 
-            fitter = lmfit.Minimizer(self.calcRwp, params)
+            fitter = lmfit.Minimizer(self.calcRwp, self.params)
 
             self.res = fitter.least_squares(**fdict)
 
@@ -2521,14 +2530,14 @@ def params(self, param_info):
         self.xn = xn[8:]
         self.wn = wn[8:]
         if param_info is not None:
-            if isinstance(param_info, Parameters):
+            if isinstance(param_info, lmfit.Parameters):
                 """
                 directly passing the parameter class
                 """
                 self._params = param_info
                 params = param_info
             else:
-                params = Parameters()
+                params = lmfit.Parameters()
 
                 if isinstance(param_info, dict):
                     """
@@ -2538,8 +2547,8 @@ def params(self, param_info):
                         params.add(
                             k,
                             value=float(v[0]),
-                            lb=float(v[1]),
-                            ub=float(v[2]),
+                            min=float(v[1]),
+                            max=float(v[2]),
                             vary=bool(v[3]),
                         )
 
@@ -2586,6 +2595,10 @@ def params(self, param_info):
 
         self._set_params_vals_to_class(params, init=True, skip_phases=True)
 
+    @property
+    def num_vary(self) -> int:
+        return sum(x.vary for x in self.params.values())
+
     @property
     def spectrum_expt(self):
         vector_list = [s.y for s in self._spectrum_expt]