multiple single fits working

src/nmreval/fit/minimizer.py

@@ -136,6 +136,8 @@ class FitRoutine(object):
             linked_sender[repl].add(par)
             linked_sender[par].add(repl)
 
+        print(_found_models)
+
         for mm, m_data in _found_models.items():
             if mm.global_parameter:
                 for dd in m_data:
@@ -235,7 +237,7 @@ class FitRoutine(object):
         var = []
         data_pars = []
 
-        # loopyloop over data that belong to one fit (linked or global)
+        # loopy-loop over data that belong to one fit (linked or global)
         for data in data_group:
             actual_pars = []
             for i, (p_k, v_k) in enumerate(data.parameter.items()):
@@ -273,10 +275,10 @@ class FitRoutine(object):
     # COST FUNCTIONS: f(x) - y (least_square, minimize), and f(x) (ODR)
     def __cost_scipy(self, p, data, varpars, used_pars):
         for keys, values in zip(varpars, p):
-            self.parameter[keys].scaled_value = values
+            data.parameter[keys].scaled_value = values
-            self.parameter[keys].namespace[keys] = self.parameter[keys].value
+            data.parameter[keys].namespace[keys] = data.parameter[keys].value
 
-        actual_parameters = [self.parameter[keys].value for keys in used_pars]
+        actual_parameters = [data.parameter[keys].value for keys in used_pars]
         return data.cost(actual_parameters)
 
     def __cost_odr(self, p, data, varpars, used_pars):
@@ -438,15 +440,17 @@ class FitRoutine(object):
 
         # update parameter values
         for keys, p_value, err_value in zip(var_pars, p, err):
-            self.parameter[keys].scaled_value = p_value
+            data.parameter[keys].scaled_value = p_value
-            self.parameter[keys].scaled_error = err_value
+            data.parameter[keys].scaled_error = err_value
+            data.parameter[keys].namespace[keys] = data.parameter[keys].value
+
 
         combinations = list(product(var_pars, var_pars))
         actual_parameters = []
         corr_idx = []
 
         for i, p_i in enumerate(used_pars):
-            actual_parameters.append(self.parameter[p_i])
+            actual_parameters.append(data.parameter[p_i])
             for j, p_j in enumerate(used_pars):
                 try:
                     # find the position of the parameter combinations

src/nmreval/fit/model.py

@@ -80,23 +80,15 @@ class Model(object):
             self.fun_kwargs = {k: v.default for k, v in inspect.signature(model.func).parameters.items()
                                if v.default is not inspect.Parameter.empty}
 
-    def set_global_parameter(self, idx, p, var=None, lb=None, ub=None, default_bounds=False):
+    def set_global_parameter(self, key, value, var=None, lb=None, ub=None, default_bounds=False):
-        if idx is None:
+        idx = [self.params.index(key)]
-            self.parameter = Parameters()
-            self.global_parameter = {}
-            return
-
         if default_bounds:
             if lb is None:
                 lb = [self.lb[i] for i in idx]
             if ub is None:
                 ub = [self.lb[i] for i in idx]
 
-        gp = self.parameter.add_parameter(p, var=var, lb=lb, ub=ub)
+        self.parameter.add(key, value, var=var, lb=lb, ub=ub)
-        for k, v in zip(idx, gp):
-            self.global_parameter[k] = v
-
-        return gp
 
     @staticmethod
     def _prep(param_len, val):

src/nmreval/fit/parameter.py

@@ -60,7 +60,7 @@ class Parameters(dict):
             else:
                 key = f'_p{new_idx}'
             new_keys.append(key)
-            self[key] = Parameter(param[i], var=var[i], lb=lb[i], ub=ub[i])
+            self[key] = Parameter(key, param[i], var=var[i], lb=lb[i], ub=ub[i])
 
         return new_keys
 

src/nmreval/models/spectrum.py

@@ -56,7 +56,6 @@ class Lorentzian:
             off (float): baseline
 
         """
-        # print('Lorentzian', a, mu, sigma, off)
         return (a/np.pi) * 2*sigma / (4*(x-mu)**2 + sigma**2) + off