FIX: pass scheme to call of MultiCompartmentSphericalHarmonicsModel (#85

)

* FIX: now correctly uses acquisition_scheme parameter in _construct_convolution_kernel call.

dmipy/core/modeling_framework.py

@@ -863,7 +863,7 @@ def _check_acquisition_scheme_has_b0s(self, acquisition_scheme):
             raise ValueError('acquisition scheme must have b0-measurements '
                              'for signal-attenuation-based model fitting.')
 
-    def _construct_convolution_kernel(self, **kwargs):
+    def _construct_convolution_kernel(self, acquisition_scheme=None, **kwargs):
         """
         Helper function that constructs the convolution kernel for the given
         multi-compartment model and the initial condition x0_vector.
@@ -883,8 +883,8 @@ def _construct_convolution_kernel(self, **kwargs):
 
         Parameters
         ----------
-        x0_vector: array of size (N_parameters),
-            Contains the fixed parameters of the convolution kernel.
+        acquisition_scheme: DmipyAcquisitionScheme instance,
+            An acquisition scheme that has been instantiated using Dmipy.
 
         Returns
         -------
@@ -894,6 +894,9 @@ def _construct_convolution_kernel(self, **kwargs):
         """
         parameters_dict = self.add_linked_parameters_to_parameters(kwargs)
 
+        if acquisition_scheme is None:
+            acquisition_scheme = self.scheme
+
         if self.volume_fractions_fixed:
             if len(self.models) > 1:
                 partial_volumes = [
@@ -915,7 +918,7 @@ def _construct_convolution_kernel(self, **kwargs):
                     )
                 kernel += S0 * partial_volume * (
                     model.convolution_kernel_matrix(
-                        self.scheme, self.sh_order, **parameters))
+                        acquisition_scheme, self.sh_order, **parameters))
         else:
             kernel = []
             for model, S0 in zip(self.models, self.S0_responses):
@@ -929,10 +932,10 @@ def _construct_convolution_kernel(self, **kwargs):
                     )
                 if 'orientation' in model.parameter_types.values():
                     kernel.append(S0 * model.convolution_kernel_matrix(
-                        self.scheme, self.sh_order, **parameters))
+                        acquisition_scheme, self.sh_order, **parameters))
                 else:
                     kernel.append(S0 * model.convolution_kernel_matrix(
-                        self.scheme, 0, **parameters))
+                        acquisition_scheme, 0, **parameters))
 
             kernel = np.hstack(kernel)
         return kernel
@@ -2161,13 +2164,11 @@ def __call__(self, acquisition_scheme, **kwargs):
         ----------
         acquisition_scheme : DmipyAcquisitionScheme instance,
             An acquisition scheme that has been instantiated using dMipy.
-        quantity : string
-            can be 'signal', 'FOD' or 'stochastic cost function' depending on
-            the need of the model.
         kwargs: keyword arguments to the model parameter values,
             Is internally given as **parameter_dictionary.
         """
-        A = self._construct_convolution_kernel(**kwargs)
+        A = self._construct_convolution_kernel(
+            acquisition_scheme=acquisition_scheme, **kwargs)
         self.S0_responses = kwargs.get('S0_responses', self.S0_responses)
         self.fit_S0_response = kwargs.get(
             'fit_S0_response', self.fit_S0_response)