reverted stupid merge error

GPy/core/gp.py

@@ -96,7 +96,8 @@ class GP(GPBase):
         model for a new variable Y* = v_tilde/tau_tilde, with a covariance
         matrix K* = K + diag(1./tau_tilde) plus a normalization term.
         """
-        return - 0.5 * self.num_data * self.output_dim * np.log(2.*np.pi) - 0.5 * self.output_dim * self.K_logdet + self._model_fit_term() + self.likelihood.Z
+        return (-0.5 * self.num_data * self.output_dim * np.log(2.*np.pi) -
+            0.5 * self.output_dim * self.K_logdet + self._model_fit_term() + self.likelihood.Z)
 
 
     def _log_likelihood_gradients(self):

GPy/core/svigp.py

@@ -4,7 +4,7 @@
 import numpy as np
 import pylab as pb
 from .. import kern
-from ..util.linalg import linalg, pdinv, mdot, tdot, dpotrs, dtrtrs, jitchol, backsub_both_sides
+from ..util.linalg import pdinv, mdot, tdot, dpotrs, dtrtrs, jitchol, backsub_both_sides
 from ..likelihoods import EP
 from gp_base import GPBase
 from model import Model
@@ -269,6 +269,7 @@ class SVIGP(GPBase):
     def optimize(self, iterations, print_interval=10, callback=lambda:None, callback_interval=5):
 
         param_step = 0.
+
         #Iterate!
         for i in range(iterations):
 
@@ -287,7 +288,6 @@ class SVIGP(GPBase):
             #compute the steps in all parameters
             vb_step = self.vb_steplength*natgrads[0]
             if (self.epochs>=1):#only move the parameters after the first epoch
-                # print "it {} ep {} par {}".format(self.iterations, self.epochs, param_step)
                 param_step = self.momentum*param_step + self.param_steplength*grads
             else:
                 param_step = 0.
@@ -295,8 +295,8 @@ class SVIGP(GPBase):
             self.set_vb_param(self.get_vb_param() + vb_step)
             #Note: don't recompute everything here, wait until the next iteration when we have a new batch
             self._set_params(self._untransform_params(self._get_params_transformed() + param_step), computations=False)
-            #print messages if desired
 
+            #print messages if desired
             if i and (not i%print_interval):
                 print i, np.mean(self._ll_trace[-print_interval:]) #, self.log_likelihood()
                 print np.round(np.mean(self._grad_trace[-print_interval:],0),3)

GPy/examples/dimensionality_reduction.py

@@ -24,7 +24,7 @@ def BGPLVM(seed=default_seed):
     Y = np.random.multivariate_normal(np.zeros(N), K, Q).T
     lik = Gaussian(Y, normalize=True)
 
-    k = GPy.kern.rbf(Q, ARD=True) + GPy.kern.bias(Q) + GPy.kern.white(Q)
+    k = GPy.kern.rbf_inv(Q, ARD=True) + GPy.kern.bias(Q) + GPy.kern.white(Q)
     # k = GPy.kern.rbf(Q) + GPy.kern.bias(Q) + GPy.kern.white(Q, 0.00001)
     # k = GPy.kern.rbf(Q, ARD = False)  + GPy.kern.white(Q, 0.00001)
 

GPy/examples/regression.py

@@ -67,8 +67,8 @@ def toy_ARD(optim_iters=1000, kernel_type='linear', N=300, D=4):
     X4 = np.log(np.sort(np.random.rand(N,1),0))
     X = np.hstack((X1, X2, X3, X4))
 
-    Y1 = np.asmatrix(2*X[:,0]+3).T
+    Y1 = np.asarray(2*X[:,0]+3).T
-    Y2 = np.asmatrix(4*(X[:,2]-1.5*X[:,0])).T
+    Y2 = np.asarray(4*(X[:,2]-1.5*X[:,0])).T
     Y = np.hstack((Y1, Y2))
 
     Y = np.dot(Y, np.random.rand(2,D));

GPy/models/bayesian_gplvm.py

@@ -115,8 +115,8 @@ class BayesianGPLVM(SparseGP, GPLVM):
         self.dbound_dZtheta = SparseGP._log_likelihood_gradients(self)
         return np.hstack((self.dbound_dmuS.flatten(), self.dbound_dZtheta))
 
-    def plot_latent(self, *args, **kwargs):
+    def plot_latent(self, plot_inducing=True, *args, **kwargs):
-        return plot_latent.plot_latent(self, *args, **kwargs)
+        return plot_latent.plot_latent(self, plot_inducing=plot_inducing, *args, **kwargs)
 
     def do_test_latents(self, Y):
         """

GPy/models/gplvm.py

@@ -36,10 +36,10 @@ class GPLVM(GP):
         self.ensure_default_constraints()
 
     def initialise_latent(self, init, input_dim, Y):
+        Xr = np.random.randn(Y.shape[0], input_dim)
         if init == 'PCA':
-            return PCA(Y, input_dim)[0]
+            Xr[:, :Y.shape[1]] = PCA(Y, input_dim)[0]
-        else:
+        return Xr
-            return np.random.randn(Y.shape[0], input_dim)
 
     def getstate(self):
         return GP.getstate(self)

GPy/testing/unit_tests.py

@@ -23,7 +23,7 @@ class GradientTests(unittest.TestCase):
         self.X2D = np.random.uniform(-3., 3., (40, 2))
         self.Y2D = np.sin(self.X2D[:, 0:1]) * np.sin(self.X2D[:, 1:2]) + np.random.randn(40, 1) * 0.05
 
-    def check_model_with_white(self, kern, model_type='GPRegression', dimension=1):
+    def check_model_with_white(self, kern, model_type='GPRegression', dimension=1, uncertain_inputs=False):
         # Get the correct gradients
         if dimension == 1:
             X = self.X1D
@@ -36,6 +36,9 @@ class GradientTests(unittest.TestCase):
 
         noise = GPy.kern.white(dimension)
         kern = kern + noise
+        if uncertain_inputs:
+            m = model_fit(X, Y, kernel=kern, X_variance=np.random.rand(X.shape[0], X.shape[1]))
+        else:
             m = model_fit(X, Y, kernel=kern)
         m.randomize()
         # contrain all parameters to be positive
@@ -141,6 +144,26 @@ class GradientTests(unittest.TestCase):
         rbf = GPy.kern.rbf(2)
         self.check_model_with_white(rbf, model_type='SparseGPRegression', dimension=2)
 
+    def test_SparseGPRegression_rbf_linear_white_kern_1D(self):
+        ''' Testing the sparse GP regression with rbf and white kernel on 2d data '''
+        rbflin = GPy.kern.rbf(1) + GPy.kern.linear(1)
+        self.check_model_with_white(rbflin, model_type='SparseGPRegression', dimension=1)
+
+    def test_SparseGPRegression_rbf_linear_white_kern_2D(self):
+        ''' Testing the sparse GP regression with rbf and white kernel on 2d data '''
+        rbflin = GPy.kern.rbf(2) + GPy.kern.linear(2)
+        self.check_model_with_white(rbflin, model_type='SparseGPRegression', dimension=2)
+
+    def test_SparseGPRegression_rbf_linear_white_kern_2D_uncertain_inputs(self):
+        ''' Testing the sparse GP regression with rbf, linear and white kernel on 2d data with uncertain inputs'''
+        rbflin = GPy.kern.rbf(2) + GPy.kern.linear(2)
+        self.check_model_with_white(rbflin, model_type='SparseGPRegression', dimension=2, uncertain_inputs=1)
+
+    def test_SparseGPRegression_rbf_linear_white_kern_1D_uncertain_inputs(self):
+        ''' Testing the sparse GP regression with rbf, linear and white kernel on 1d data with uncertain inputs'''
+        rbflin = GPy.kern.rbf(1) + GPy.kern.linear(1)
+        self.check_model_with_white(rbflin, model_type='SparseGPRegression', dimension=1, uncertain_inputs=1)
+
     def test_GPLVM_rbf_bias_white_kern_2D(self):
         """ Testing GPLVM with rbf + bias and white kernel """
         N, input_dim, D = 50, 1, 2