remove rng from EnrichmentSource and pass rng to get_value_for_featur…

…e() method

bhepop2/enrichment/base.py

@@ -84,6 +84,19 @@ def _evaluate_feature_on_population(self):
         # implement feature evaluation using a dedicated algorithm
         raise NotImplementedError
 
+    def _get_value_for_feature(self, feature_id):
+        """
+        Get a feature value for the given feature id.
+
+        This method is a helper that class self.source.get_value_for_feature
+        with feature id and self.rng.
+
+        :param feature_id:
+
+        :return: feature value
+        """
+        return self.source.get_value_for_feature(feature_id, self.rng)
+
     # validation and read
 
     @abstractmethod

bhepop2/enrichment/bhepop2.py

@@ -167,7 +167,7 @@ def _draw_feature_value(self, probs):
         feature_index = self.rng.choice(feature_indexes, p=filtered_probs)
 
         # get feature value from distribution
-        value = self.source.get_value_for_feature(feature_index)
+        value = self._get_value_for_feature(feature_index)
 
         return value
 

bhepop2/enrichment/uniform.py

@@ -40,7 +40,7 @@ def _evaluate_feature_on_population(self):
 
     def _draw_feature_value(self):
         feature_index = self.rng.integers(len(self.source.feature_values))
-        return self.source.get_value_for_feature(feature_index)
+        return self._get_value_for_feature(feature_index)
 
     def _validate_and_process_inputs(self):
         assert isinstance(self.source, QuantitativeGlobalDistribution)

bhepop2/sources/base.py

@@ -31,10 +31,6 @@ def __init__(self, data, name: str = None):
 
         self.name: str = DEFAULT_SOURCE_NAME if name is None else name
 
-        # random number generator.
-        # Set by the SyntheticPopulationEnrichment class is used with one.
-        self.rng = None
-
         self.data = data
 
         self._feature_values = None
@@ -77,14 +73,15 @@ def _validate_data(self):
         pass
 
     @abstractmethod
-    def get_value_for_feature(self, feature_index):
+    def get_value_for_feature(self, feature_index, rng):
         """
         Return a feature value for the given feature index.
 
         Generate a singular value from the feature state
         corresponding to the given index.
 
         :param feature_index: index of the feature in self.feature_values
+        :param rng: Numpy random Generator
 
         :return: feature value
         """

bhepop2/sources/global_distribution.py

@@ -44,22 +44,23 @@ def _validate_data(self):
         assert set(self.data.columns) >= {f"D{i}" for i in range(1, 10)}
         assert len(self.data) == 1
 
-    def get_value_for_feature(self, feature_index):
+    def get_value_for_feature(self, feature_index, rng):
         """
         Return a value drawn from the interval corresponding to the feature index.
 
         The first interval is defined as [self._abs_minimum, self.feature_values[0]].
         and so on. The value is drawn using a uniform rule.
 
         :param feature_index:
+        :param rng:
         :return:
         """
 
         interval_values = [self._abs_minimum] + self.feature_values
-
+        print(interval_values)
         lower, upper = interval_values[feature_index], interval_values[feature_index + 1]
 
-        draw = self.rng.uniform()
+        draw = rng.uniform()
 
         drawn_feature_value = lower + (upper - lower) * draw
 

bhepop2/sources/marginal_distributions.py

@@ -196,7 +196,7 @@ def compute_feature_prob(self, attribute=ALL_LABEL, modality=ALL_LABEL):
 
         return res
 
-    def get_value_for_feature(self, feature_index):
+    def get_value_for_feature(self, feature_index, rng):
         # directly return the stored feature value
         return self.feature_values[feature_index]
 
@@ -323,22 +323,23 @@ def compute_feature_prob(self, attribute=ALL_LABEL, modality=ALL_LABEL):
 
         return prob_df
 
-    def get_value_for_feature(self, feature_index):
+    def get_value_for_feature(self, feature_index, rng):
         """
         Return a value drawn from the interval corresponding to the feature index.
 
         The first interval is defined as [self._abs_minimum, self.feature_values[0]].
         and so on. The value is drawn using a uniform rule.
 
         :param feature_index:
+        :param rng:
         :return:
         """
 
         interval_values = [self._abs_minimum] + self.feature_values
 
         lower, upper = interval_values[feature_index], interval_values[feature_index + 1]
 
-        draw = self.rng.uniform()
+        draw = rng.uniform()
 
         drawn_feature_value = lower + (upper - lower) * draw