various merge conflicts from the newGP branch

GPy/examples/BGPLVM_demo.py

@@ -17,7 +17,7 @@ K = k.K(X)
 Y = np.random.multivariate_normal(np.zeros(N),K,D).T
 
 # k = GPy.kern.rbf(Q) + GPy.kern.bias(Q) + GPy.kern.white(Q, 0.00001)
-k = GPy.kern.linear(Q, ARD = False)  + GPy.kern.white(Q, 0.00001)
+k = GPy.kern.rbf(Q, ARD = False)  + GPy.kern.white(Q, 0.00001)
 m = GPy.models.Bayesian_GPLVM(Y, Q, kernel = k,  M=M)
 m.constrain_positive('(rbf|bias|noise|white|S)')
 # m.constrain_fixed('S', 1)

GPy/models/BGPLVM.py

@@ -33,7 +33,7 @@ class Bayesian_GPLVM(sparse_GP, GPLVM):
         kernel = kern.rbf(Q) + kern.white(Q)
 
         S = np.ones_like(X) * 1e-2#
-        sparse_GP.__init__(self, X, Gaussian(Y), X_uncertainty = S, Z=Z,**kwargs)
+        sparse_GP.__init__(self, X, Gaussian(Y), X_uncertainty=S, Z=Z,**kwargs)
 
     def _get_param_names(self):
         X_names = sum([['X_%i_%i'%(n,q) for n in range(self.N)] for q in range(self.Q)],[])
@@ -46,7 +46,7 @@ class Bayesian_GPLVM(sparse_GP, GPLVM):
         The resulting 1-D array has this structure:
 
         ===============================================================
-        |       mu       |        S        |    Z    | beta |  theta  |
+        |       mu       |        S        |    Z    | theta |  beta  |
         ===============================================================
 
         """

GPy/models/GPLVM.py

@@ -8,9 +8,10 @@ import sys, pdb
 from .. import kern
 from ..core import model
 from ..util.linalg import pdinv, PCA
-from GP_regression import GP_regression
+from GP import GP
+from ..likelihoods import Gaussian
 
-class GPLVM(GP_regression):
+class GPLVM(GP):
     """
     Gaussian Process Latent Variable Model
 
@@ -22,10 +23,13 @@ class GPLVM(GP_regression):
     :type init: 'PCA'|'random'
 
     """
-    def __init__(self, Y, Q, init='PCA', X = None, **kwargs):
+    def __init__(self, Y, Q, init='PCA', X = None, kernel=None, **kwargs):
         if X is None:
             X = self.initialise_latent(init, Q, Y)
-        GP_regression.__init__(self, X, Y, **kwargs)
+        if kernel is None:
+            kernel = kern.rbf(Q) + kern.bias(Q)
+        likelihood = Gaussian(Y)
+        GP.__init__(self, X, likelihood, kernel, **kwargs)
 
     def initialise_latent(self, init, Q, Y):
         if init == 'PCA':
@@ -34,23 +38,19 @@ class GPLVM(GP_regression):
             return np.random.randn(Y.shape[0], Q)
 
     def _get_param_names(self):
-        return (sum([['X_%i_%i'%(n,q) for n in range(self.N)] for q in range(self.Q)],[])
-                + self.kern._get_param_names_transformed())
+        return sum([['X_%i_%i'%(n,q) for n in range(self.N)] for q in range(self.Q)],[]) + GP._get_param_names(self)
 
     def _get_params(self):
-        return np.hstack((self.X.flatten(), self.kern._get_params_transformed()))
+        return np.hstack((self.X.flatten(), GP._get_params(self)))
 
     def _set_params(self,x):
         self.X = x[:self.X.size].reshape(self.N,self.Q).copy()
-        GP_regression._set_params(self, x[self.X.size:])
+        GP._set_params(self, x[self.X.size:])
 
     def _log_likelihood_gradients(self):
-        dL_dK = self.dL_dK()
+        dL_dX = 2.*self.kern.dK_dX(self.dL_dK,self.X)
 
-        dL_dtheta = self.kern.dK_dtheta(dL_dK,self.X)
-        dL_dX = 2*self.kern.dK_dX(dL_dK,self.X)
-
-        return np.hstack((dL_dX.flatten(),dL_dtheta))
+        return np.hstack((dL_dX.flatten(),GP._log_likelihood_gradients(self)))
 
     def plot(self):
         assert self.Y.shape[1]==2

GPy/testing/unit_tests.py

@@ -160,8 +160,8 @@ class GradientTests(unittest.TestCase):
         Y = np.hstack([np.ones(N/2),np.repeat(-1,N/2)])[:,None]
         kernel = GPy.kern.rbf(1)
         distribution = GPy.likelihoods.likelihood_functions.probit()
-        likelihood = GPy.likelihoods.EP(Y,distribution)
-        m = GPy.models.GP(X,kernel,likelihood=likelihood)
+        likelihood = GPy.likelihoods.EP(Y, distribution)
+        m = GPy.models.GP(X, likelihood, kernel)
         m.ensure_default_constraints()
         self.assertTrue(m.EPEM)