Assorted work on combining the EP and sparse methods

GPy/likelihoods/EP.py

@@ -19,6 +19,7 @@ class EP(likelihood):
         self.data = data
         self.N = self.data.size
         self.is_heteroscedastic = True
+        self.Nparams = 0
 
         #Initial values - Likelihood approximation parameters:
         #p(y|f) = t(f|tau_tilde,v_tilde)
@@ -28,6 +29,7 @@ class EP(likelihood):
         #initial values for the GP variables
         self.Y = np.zeros((self.N,1))
         self.covariance_matrix = np.eye(self.N)
+        self.precision = np.ones(self.N)
         self.Z = 0
         self.YYT = None
 

GPy/likelihoods/Gaussian.py

@@ -4,6 +4,7 @@ from likelihood import likelihood
 class Gaussian(likelihood):
     def __init__(self,data,variance=1.,normalize=False):
         self.is_heteroscedastic = False
+        self.Nparams = 1
         self.data = data
         self.N,D = data.shape
         self.Z = 0. # a correction factor which accounts for the approximation made
@@ -18,7 +19,9 @@ class Gaussian(likelihood):
             self._std = np.ones((1,D))
             self.Y = self.data
 
+        #TODO: make this work efficiently (only compute YYT if D>>N)
         self.YYT = np.dot(self.Y,self.Y.T)
+        self.trYYT = np.trace(self.YYT)
         self._set_params(np.asarray(variance))
 
 
@@ -50,4 +53,4 @@ class Gaussian(likelihood):
         pass
 
     def _gradients(self,partial):
-        return np.sum(np.diag(partial))
+        return np.sum(partial)