Fix hyper model caching, adds data check

Caching method now checks whether input data is the same along with the input parameters when evaluating the probability.

Closes #524 (closed)

bilby/hyper/model.py

@@ -3,9 +3,9 @@ from ..core.utils import infer_args_from_function_except_n_args
 
 class Model(object):
     """
-    Population model
+    Returns the probability for a given population model using
+    data and population parameters.
 
-    This should take population parameters and return the probability.
     """
 
     def __init__(self, model_functions=None):
@@ -18,14 +18,33 @@ class Model(object):
         self.models = model_functions
         self._cached_parameters = {model: None for model in self.models}
         self._cached_probability = {model: None for model in self.models}
+        self._cached_data = {model: None for model in self.models}
 
         self.parameters = dict()
 
     def prob(self, data, **kwargs):
+        """
+        Parameters
+        ----------
+        data: dict, DataFrame
+            Data to evaluate probability for.
+
+        Returns
+        -------
+        probability: float
+            Probability evaluated for a given population model.
+            Uses data and function parameters.
+        """
         probability = 1.0
+        import pandas as pd
         for ii, function in enumerate(self.models):
             function_parameters = self._get_function_parameters(function)
-            if self._cached_parameters[function] == function_parameters:
+            A = self._cached_parameters[function] == function_parameters
+            if isinstance(data, pd.DataFrame):
+                B = data.equals(self._cached_data[function])
+            else:
+                B = data == self._cached_data[function]
+            if A and B:
                 new_probability = self._cached_probability[function]
             else:
                 new_probability = function(
@@ -33,13 +52,14 @@ class Model(object):
                 )
                 self._cached_parameters[function] = function_parameters
                 self._cached_probability[function] = new_probability
+                self._cached_data[function] = data
             probability *= new_probability
         return probability
 
     def _get_function_parameters(self, func):
         """If the function is a class method we need to remove more arguments"""
         param_keys = infer_args_from_function_except_n_args(func, n=0)
-        ignore = ['dataset', 'self', 'cls']
+        ignore = ["dataset", "self", "cls"]
         for key in ignore:
             if key in param_keys:
                 del param_keys[param_keys.index(key)]