Update get_model_params to use sgp for large datasets

sgptools/utils/gpflow.py

@@ -21,6 +21,8 @@
 import numpy as np
 import matplotlib.pyplot as plt
 
+from .misc import get_inducing_pts
+
 
 def plot_loss(losses, save_file=None):
     """Helper function to plot the training loss
@@ -52,7 +54,8 @@ def get_model_params(X_train, y_train,
                      kernel=None,
                      return_gp=False,
                      **kwargs):
-    """Train a GP on the given training set
+    """Train a GP on the given training set. 
+    Trains a sparse GP if the training set is larger than 1000 samples.
 
     Args:
         X_train (ndarray): (n, d); Training set inputs
@@ -79,12 +82,23 @@ def get_model_params(X_train, y_train,
         kernel = gpflow.kernels.SquaredExponential(lengthscales=lengthscales, 
                                                    variance=variance)
 
-    gpr = gpflow.models.GPR(data=(X_train, y_train), 
-                            kernel=kernel,
-                            noise_variance=noise_variance)
+    if len(X_train) <= 1500:
+        gpr = gpflow.models.GPR(data=(X_train, y_train), 
+                                kernel=kernel,
+                                noise_variance=noise_variance)
+        trainable_variables=gpr.trainable_variables
+    else:
+        inducing_pts = get_inducing_pts(X_train, 500)
+        gpr = gpflow.models.SGPR(data=(X_train, y_train), 
+                                 kernel=kernel,
+                                 inducing_variable=inducing_pts,
+                                 noise_variance=noise_variance)
+        trainable_variables=gpr.trainable_variables[1:]
 
     if max_steps > 0:
-        loss = optimize_model(gpr, max_steps=max_steps, lr=lr, **kwargs)
+        loss = optimize_model(gpr, max_steps=max_steps, lr=lr, 
+                              trainable_variables=trainable_variables,
+                              **kwargs)
     else:
         loss = 0