Fixed directed info undefined and hhg

Author: olivercliff

pyspi/config.yaml

@@ -103,12 +103,12 @@
   MultiscaleGraphCorrelation:
 
   # Hilbert-Schmidt independence criterion
-  HSIC:
+  HilbertSchmidtIndependenceCriterion:
     - biased: False
     - biased: True
 
   # Heller-Heller-Gorfine (HHG) test
-  HHG:
+  HellerHellerGorfine:
 
   # Multi-scale graph correlation for time series
   CrossMultiscaleGraphCorrelation:

pyspi/fast_config.yaml

@@ -83,7 +83,7 @@
     - biased: True
 
   # Hilbert-Schmidt independence criterion
-  HSIC:
+  HilbertSchmidtIndependenceCriterion:
     - biased: False
     - biased: True
 

pyspi/statistics/distance.py

@@ -44,7 +44,7 @@ def multivariate(self, data):
 """
 
 
-class HSIC(Undirected, Unsigned):
+class HilbertSchmidtIndependenceCriterion(Undirected, Unsigned):
     """Hilbert-Schmidt Independence Criterion (HSIC)"""
 
     name = "Hilbert-Schmidt Independence Criterion"
@@ -63,7 +63,7 @@ def bivariate(self, data, i=None, j=None):
         return stat
 
 
-class HHG(Directed, Unsigned):
+class HellerHellerGorfine(Directed, Unsigned):
     """Heller-Heller-Gorfine independence criterion"""
 
     name = "Heller-Heller-Gorfine Independence Criterion"

pyspi/statistics/infotheory.py

@@ -499,15 +499,16 @@ def __init__(self, n=5, **kwargs):
     def _compute_entropy_rates(self, targ):
 
         targ = np.squeeze(targ)
+        m_utils = jp.JPackage("infodynamics.utils").MatrixUtils
 
         entropy_rate_sum = 0
         for i in range(1, self._n + 1):
             # Compute entropy for an i-dimensional embedding
             self._entropy_calc.initialise(i)
 
-            Yi = self.m_utils.makeDelayEmbeddingVector(jp.JArray(jp.JDouble, 1)(targ), i)
+            Yi = m_utils.makeDelayEmbeddingVector(jp.JArray(jp.JDouble, 1)(targ), i)
             self._entropy_calc.setObservations(Yi)
-            entropy_rate_sum = entropy_rate_sum + super(DirectedInfo, self).computeAverageLocalOfObservations() / i
+            entropy_rate_sum = entropy_rate_sum + self._entropy_calc.computeAverageLocalOfObservations() / i
 
         return entropy_rate_sum
 
@@ -516,7 +517,7 @@ def bivariate(self, data, i=None, j=None):
         """Compute directed information from i to j"""
 
         entropy_rates = self._compute_entropy_rates(data.to_numpy(squeeze=True)[j])
-        causal_entropy = self._causal_entropy.bivariate(data, i=i, j=j)
+        causal_entropy = super().bivariate(data, i=i, j=j)
 
         return entropy_rates - causal_entropy