x_out optional even for multi-task and bug fix with entropy related p…

…osterior quantities
lbl-camera · Nov 27, 2024 · c06d12b · c06d12b
1 parent 4910147
commit c06d12b
Show file tree

Hide file tree

Showing 12 changed files with 146 additions and 146 deletions.
diff --git a/docs/source/examples/MultiTaskTest.ipynb b/docs/source/examples/MultiTaskTest.ipynb
@@ -16,8 +16,9 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "##first install the newest version of fvgp\n",
-    "#!pip install fvgp~=4.5.3"
+    "##first, install the newest version of fvgp\n",
+    "#!pip install fvgp~=4.5.6\n",
+    "#!pip install torch"
    ]
   },
   {
@@ -133,9 +134,10 @@
    "outputs": [],
    "source": [
     "#mean and standard deviation\n",
-    "mean = my_gp2.posterior_mean(x_pred=x_pred1d.reshape(50,1), x_out=np.array([0,1]))[\"f(x)\"]\n",
+    "mean = my_gp2.posterior_mean(x_pred=x_pred1d.reshape(50,1))[\"f(x)\"]\n",
     "std = np.sqrt(my_gp2.posterior_covariance(x_pred=x_pred1d.reshape(50,1), x_out=np.array([0,1]))[\"v(x)\"])\n",
     "\n",
+    "\n",
     "plt.plot(x_pred1d.reshape(50,1),mean[:,0], label = \"mean task 1\")\n",
     "plt.plot(x_pred1d.reshape(50,1),mean[:,1], label = \"mean task 2\")\n",
     "plt.scatter(x_data,y_data1) \n",
@@ -411,9 +413,9 @@
     "\n",
     "\n",
     "bounds = np.zeros((n.number_of_hps+2,2))\n",
-    "bounds[0] = np.array([0.001,10.])\n",
-    "bounds[1] = np.array([0.001,10.])\n",
-    "bounds[2:] = np.array([-1,1])\n",
+    "bounds[0] = np.array([0.1,10.])\n",
+    "bounds[1] = np.array([0.1,10.])\n",
+    "bounds[2:] = np.array([0.01,1])\n",
     "my_gp2.train(hyperparameter_bounds=bounds,max_iter = 2)"
    ]
   },
@@ -508,13 +510,29 @@
     "fig.show()\n",
     "\n"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "811a3337-a377-44bd-85b7-9a699c299cd0",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "069a28fb-fecf-42e9-bc27-16fcccec3080",
+   "metadata": {},
+   "outputs": [],
+   "source": []
   }
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3 (ipykernel)",
+   "display_name": "fvgp_dev",
    "language": "python",
-   "name": "python3"
+   "name": "fvgp_dev"
   },
   "language_info": {
    "codemirror_mode": {

diff --git a/docs/source/examples/NonEuclideanInputSpaces.ipynb b/docs/source/examples/NonEuclideanInputSpaces.ipynb
@@ -19,7 +19,7 @@
    "outputs": [],
    "source": [
     "#install the newest version of fvgp\n",
-    "#!pip install fvgp~=4.5.3"
+    "#!pip install fvgp~=4.5.6"
    ]
   },
   {
@@ -214,7 +214,8 @@
    "outputs": [],
    "source": [
     "x_pred = [\"dwed\",\"dwe\"]\n",
-    "my_gp2.posterior_mean(x_pred, x_out = np.array([0,1,2,3]))"
+    "my_gp2.posterior_mean(x_pred, x_out = np.array([0,1,2,3]))\n",
+    "my_gp2.posterior_mean(x_pred)"
    ]
   },
   {

diff --git a/docs/source/examples/SingleTaskTest.ipynb b/docs/source/examples/SingleTaskTest.ipynb
@@ -13,7 +13,7 @@
    "id": "a1b1c026",
    "metadata": {},
    "source": [
-    "This is the new test for fvgp version 4.3.0 and later."
+    "This is the new test for fvgp version 4.5.5 and later."
    ]
   },
   {
@@ -23,7 +23,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "#!pip install fvgp~=4.5.3"
+    "#!pip install fvgp~=4.5.5"
    ]
   },
   {
@@ -170,7 +170,7 @@
     "                       [0.001,0.1],  #noise\n",
     "                       [0.01,1.]  #mean\n",
     "                      ])\n",
-    "my_gp1.update_gp_data(x_data, y_data, noise_variances_new=np.ones(y_data.shape) * 0.01)\n",
+    "my_gp1.update_gp_data(x_data, y_data, noise_variances_new=np.ones(y_data.shape) * 0.01) #this is just for testing, not needed\n",
     "print(\"Standard Training\")\n",
     "my_gp1.train(hyperparameter_bounds=hps_bounds)\n",
     "print(\"Global Training\")\n",
@@ -316,15 +316,13 @@
     "my_gp1.gp_entropy(x_test)\n",
     "my_gp1.gp_entropy_grad(x_test, 0)\n",
     "my_gp1.gp_kl_div(x_test, np.ones((len(x_test))), np.identity((len(x_test))))\n",
-    "my_gp1.gp_kl_div_grad(x_test, np.ones((len(x_test))), np.identity((len(x_test))), 0)\n",
     "my_gp1.gp_relative_information_entropy(x_test)\n",
     "my_gp1.gp_relative_information_entropy_set(x_test)\n",
     "my_gp1.posterior_covariance(x_test)\n",
     "my_gp1.posterior_covariance_grad(x_test)\n",
     "my_gp1.posterior_mean(x_test)\n",
     "my_gp1.posterior_mean_grad(x_test)\n",
     "my_gp1.posterior_probability(x_test, np.ones((len(x_test))), np.identity((len(x_test))))\n",
-    "my_gp1.posterior_probability_grad(x_test, np.ones((len(x_test))), np.identity((len(x_test))),0)\n",
     "\n"
    ]
   },

diff --git a/docs/source/examples/gp2ScaleTest.ipynb b/docs/source/examples/gp2ScaleTest.ipynb
@@ -19,7 +19,8 @@
    "outputs": [],
    "source": [
     "##first install the newest version of fvgp\n",
-    "#!pip install fvgp~=4.5.3"
+    "#!pip install fvgp~=4.5.6\n",
+    "#!pip install imate"
    ]
   },
   {

diff --git a/examples/MultiTaskTest.ipynb b/examples/MultiTaskTest.ipynb
@@ -17,7 +17,8 @@
    "outputs": [],
    "source": [
     "##first, install the newest version of fvgp\n",
-    "#!pip install fvgp~=4.5.5"
+    "#!pip install fvgp~=4.5.6\n",
+    "#!pip install torch"
    ]
   },
   {
@@ -412,9 +413,9 @@
     "\n",
     "\n",
     "bounds = np.zeros((n.number_of_hps+2,2))\n",
-    "bounds[0] = np.array([0.001,10.])\n",
-    "bounds[1] = np.array([0.001,10.])\n",
-    "bounds[2:] = np.array([-1,1])\n",
+    "bounds[0] = np.array([0.1,10.])\n",
+    "bounds[1] = np.array([0.1,10.])\n",
+    "bounds[2:] = np.array([0.01,1])\n",
     "my_gp2.train(hyperparameter_bounds=bounds,max_iter = 2)"
    ]
   },
@@ -517,13 +518,21 @@
    "metadata": {},
    "outputs": [],
    "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "069a28fb-fecf-42e9-bc27-16fcccec3080",
+   "metadata": {},
+   "outputs": [],
+   "source": []
   }
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3 (ipykernel)",
+   "display_name": "fvgp_dev",
    "language": "python",
-   "name": "python3"
+   "name": "fvgp_dev"
   },
   "language_info": {
    "codemirror_mode": {

diff --git a/examples/NonEuclideanInputSpaces.ipynb b/examples/NonEuclideanInputSpaces.ipynb
@@ -19,7 +19,7 @@
    "outputs": [],
    "source": [
     "#install the newest version of fvgp\n",
-    "#!pip install fvgp~=4.5.3"
+    "#!pip install fvgp~=4.5.6"
    ]
   },
   {
@@ -214,7 +214,8 @@
    "outputs": [],
    "source": [
     "x_pred = [\"dwed\",\"dwe\"]\n",
-    "my_gp2.posterior_mean(x_pred, x_out = np.array([0,1,2,3]))"
+    "my_gp2.posterior_mean(x_pred, x_out = np.array([0,1,2,3]))\n",
+    "my_gp2.posterior_mean(x_pred)"
    ]
   },
   {

diff --git a/examples/SingleTaskTest.ipynb b/examples/SingleTaskTest.ipynb
@@ -170,7 +170,7 @@
     "                       [0.001,0.1],  #noise\n",
     "                       [0.01,1.]  #mean\n",
     "                      ])\n",
-    "my_gp1.update_gp_data(x_data, y_data, noise_variances_new=np.ones(y_data.shape) * 0.01)\n",
+    "my_gp1.update_gp_data(x_data, y_data, noise_variances_new=np.ones(y_data.shape) * 0.01) #this is just for testing, not needed\n",
     "print(\"Standard Training\")\n",
     "my_gp1.train(hyperparameter_bounds=hps_bounds)\n",
     "print(\"Global Training\")\n",
@@ -316,15 +316,13 @@
     "my_gp1.gp_entropy(x_test)\n",
     "my_gp1.gp_entropy_grad(x_test, 0)\n",
     "my_gp1.gp_kl_div(x_test, np.ones((len(x_test))), np.identity((len(x_test))))\n",
-    "my_gp1.gp_kl_div_grad(x_test, np.ones((len(x_test))), np.identity((len(x_test))), 0)\n",
     "my_gp1.gp_relative_information_entropy(x_test)\n",
     "my_gp1.gp_relative_information_entropy_set(x_test)\n",
     "my_gp1.posterior_covariance(x_test)\n",
     "my_gp1.posterior_covariance_grad(x_test)\n",
     "my_gp1.posterior_mean(x_test)\n",
     "my_gp1.posterior_mean_grad(x_test)\n",
     "my_gp1.posterior_probability(x_test, np.ones((len(x_test))), np.identity((len(x_test))))\n",
-    "my_gp1.posterior_probability_grad(x_test, np.ones((len(x_test))), np.identity((len(x_test))),0)\n",
     "\n"
    ]
   },

diff --git a/examples/gp2ScaleTest.ipynb b/examples/gp2ScaleTest.ipynb
@@ -19,7 +19,8 @@
    "outputs": [],
    "source": [
     "##first install the newest version of fvgp\n",
-    "#!pip install fvgp~=4.5.3"
+    "#!pip install fvgp~=4.5.6\n",
+    "#!pip install imate"
    ]
   },
   {

diff --git a/fvgp/fvgp.py b/fvgp/fvgp.py
@@ -266,8 +266,6 @@ class provides all the methods described for the GP (:py:class:`fvgp.GP`) class.
 
     :py:meth:`fvgp.GP.posterior_probability`
 
-    :py:meth:`fvgp.GP.posterior_probability_grad`
-
     Validation Methods:
 
     :py:meth:`fvgp.GP.crps`

diff --git a/fvgp/gp.py b/fvgp/gp.py
@@ -1232,7 +1232,7 @@ def posterior_probability(self, x_pred, comp_mean, comp_cov, x_out=None):
         comp_mean: np.ndarray
             A vector of mean values, same length as x_pred.
         comp_cov: np.nparray
-            Covariance matrix, in R^{len(x_pred) times len(x_pred)}
+            Covariance matrix, in R^{len(x_pred) x len(x_pred)}
         x_out : np.ndarray, optional
             Output coordinates in case of multi-task GP use; a numpy array of size (N),
             where N is the number evaluation points in the output direction.
@@ -1245,35 +1245,6 @@ def posterior_probability(self, x_pred, comp_mean, comp_cov, x_out=None):
         """
         return self.posterior.posterior_probability(x_pred, comp_mean, comp_cov, x_out=x_out)
 
-    def posterior_probability_grad(self, x_pred, comp_mean, comp_cov, direction, x_out=None):
-        """
-        Function to compute the gradient of the probability of a probabilistic quantity of interest,
-        given the GP posterior at a given point.
-
-        Parameters
-        ----------
-        x_pred : np.ndarray  or list
-            A numpy array of shape (V x D), interpreted as  an array of input point positions, or a list for
-            GPs on non-Euclidean input spaces.
-        comp_mean: np.ndarray
-            A vector of mean values, same length as x_pred.
-        comp_cov: np.nparray
-            Covariance matrix, in R^{len(x_pred) times len(x_pred)}
-        direction : int
-            The direction to compute the gradient in.
-        x_out : np.ndarray, optional
-            Output coordinates in case of multi-task GP use; a numpy array of size (N),
-            where N is the number evaluation points in the output direction.
-            Usually this is np.ndarray([0,1,2,...]).
-
-        Return
-        ------
-        Solution : dict
-            The gradient of the probability of a probabilistic quantity of interest,
-            given the GP posterior at a given point.
-        """
-        return self.posterior.posterior_probability_grad(x_pred, comp_mean, comp_cov, direction, x_out=x_out)
-
     ####################################################################################
     ####################################################################################
     #######################VALIDATION###################################################