ENH/FIX: coerce 'data' and 'independent_var' in Model class

- input 'data' will always become 'float64' or 'complex128'
- coercion is only done for certain data types in case of
    an 'independent_var' to still allow for more complicated
    (user-defined) data types

See: lmfit#723
Closes: lmfit#727

doc/whatsnew.rst

@@ -22,6 +22,7 @@ New features:
 Bug fixes/enhancements:
 - do not overwrite user-specified figure titles in Model.plot() functions and allow setting with 'title' keyword argument (PR #711)
 - preserve Parameters subclass in deepcopy (@jenshnielsen; PR #719)
+- coerce ``data`` and ``indepdent_vars`` to NumPy array with ``dtype=float64`` whenever applicable (Issues #723 and #727)
 
 Various:
 - update asteval dependency to >=0.9.22 to avoid DeprecationWarnings from NumPy v1.20.0 (PR #707)

lmfit/model.py

@@ -982,15 +982,6 @@ def fit(self, data, params=None, weights=None, method='leastsq',
             msg += 'Non initialized parameters: %s' % str(blank)
             raise ValueError(msg)
 
-        # Do not alter anything that implements the array interface (np.array, pd.Series)
-        # but convert other iterables (e.g., Python lists) to NumPy arrays.
-        if not hasattr(data, '__array__'):
-            data = np.asfarray(data)
-        for var in self.independent_vars:
-            var_data = kwargs[var]
-            if isinstance(var_data, (list, tuple)):
-                kwargs[var] = np.asfarray(var_data)
-
         # Handle null/missing values.
         if nan_policy is not None:
             self.nan_policy = nan_policy
@@ -1005,12 +996,30 @@ def fit(self, data, params=None, weights=None, method='leastsq',
 
         # If independent_vars and data are alignable (pandas), align them,
         # and apply the mask from above if there is one.
-
         for var in self.independent_vars:
             if not np.isscalar(kwargs[var]):
-                # print("Model fit align ind dep ", var, mask.sum())
                 kwargs[var] = _align(kwargs[var], mask, data)
 
+        # Make sure `dtype` for data is always `float64` or `complex128`
+        if np.isrealobj(data):
+            data = np.asfarray(data)
+        elif np.iscomplexobj(data):
+            data = np.asarray(data, dtype='complex128')
+
+        # Coerce `dtype` for independent variable(s) to `float64` or
+        # `complex128` when the variable has one of the following types: list,
+        # tuple, numpy.ndarray, or pandas.Series
+        for var in self.independent_vars:
+            var_data = kwargs[var]
+            if isinstance(var_data, (list, tuple)):
+                kwargs[var] = np.asfarray(var_data)
+            elif (isinstance(var_data, (np.ndarray, Series)) and
+                    np.isrealobj(var_data)):
+                kwargs[var] = np.asfarray(var_data)
+            elif (isinstance(var_data, (np.ndarray, Series)) and
+                    np.iscomplexobj(var_data)):
+                kwargs[var] = np.asarray(var_data, dtype='complex128')
+
         if fit_kws is None:
             fit_kws = {}