Merge pull request #23 from lbl-camera/dev2023Sept26

Dev2023 sept26

docs/source/examples/MultiTaskTest.ipynb

@@ -6,7 +6,7 @@
    "metadata": {},
    "source": [
     "# Multi-Task Test\n",
-    "At first we have to install the newest version of fvgGP"
+    "At first we have to install the newest version of fvGP"
    ]
   },
   {

examples/MultiTaskTest.ipynb

@@ -6,7 +6,7 @@
    "metadata": {},
    "source": [
     "# Multi-Task Test\n",
-    "At first we have to install the newest version of fvgGP"
+    "At first we have to install the newest version of fvGP"
    ]
   },
   {
@@ -17,7 +17,7 @@
    "outputs": [],
    "source": [
     "##first install the newest version of fvgp\n",
-    "#!pip install fvgp==4.0.0"
+    "#!pip install fvgp==4.0.5"
    ]
   },
   {
@@ -177,12 +177,11 @@
     "              #gp_kernel_function=mkernel #what happens if comment this out? (adjust bounds below)\n",
     "             )\n",
     "#change this based on kernel choice\n",
-    "hps_bounds = my_gp2.hps_bounds \n",
     "#hps_bounds = np.array([[0.001,10000.],[1.,1000.]])\n",
     "\n",
     "#my_gp2.update_gp_data(x_data3,y_data3)\n",
     "print(\"Global Training in progress\")\n",
-    "my_gp2.train(hps_bounds, max_iter = 200)"
+    "my_gp2.train(max_iter = 2)"
    ]
   },
   {
@@ -272,6 +271,14 @@
     "fig.show()\n",
     "\n"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "1694322d",
+   "metadata": {},
+   "outputs": [],
+   "source": []
   }
  ],
  "metadata": {

examples/SingleTaskTest.ipynb

@@ -24,7 +24,7 @@
    "outputs": [],
    "source": [
     "##first install the newest version of fvgp\n",
-    "#!pip install fvgp==4.0.0"
+    "#!pip install fvgp==4.0.5"
    ]
   },
   {
@@ -215,7 +215,7 @@
    "outputs": [],
    "source": [
     "for i in range(10):\n",
-    "    time.sleep(5)\n",
+    "    time.sleep(2)\n",
     "    my_gp1.update_hyperparameters(opt_obj)\n",
     "    print(my_gp1.hyperparameters)\n",
     "    print(\"\")"

examples/gp2ScaleTest.ipynb

@@ -8,7 +8,7 @@
     "# gp2Scale \n",
     "gp2Scale is a special setting in fvgp that combines non-stationary, compactly-supported kernels, HPC distributed computing, and sparse linear algebra to allow scale-up of exact GPs to millions of data points. Here we run a moderately-sized GP, just because we assume you might run this locally.\n",
     "\n",
-    "I hope it is clear how cool it is what is happening here. If you have a dask client that points to a remote cluster with 500 GPUs, you will distribute the covariance matrix computation across those. The full matrix is sparse will be fast to work with in downstream operations. The algorithm only makes use of naturally-occuring sparsity, so the result is exact in contrast to Vecchia or inducing-point methods."
+    "I hope it is clear how cool it is what is happening here. If you have a dask client that points to a remote cluster with 500 GPUs, you will distribute the covariance matrix computation across those. The full matrix is sparse and will be fast to work with in downstream operations. The algorithm only makes use of naturally-occuring sparsity, so the result is exact in contrast to Vecchia or inducing-point methods."
    ]
   },
   {
@@ -19,7 +19,7 @@
    "outputs": [],
    "source": [
     "##first install the newest version of fvgp\n",
-    "#!pip install fvgp==4.0.0"
+    "#!pip install fvgp==4.0.5"
    ]
   },
   {

examples/gp2Scale_example_HPC_jobscript.sl

@@ -26,7 +26,7 @@ echo ${port}
 echo "starting scheduler"
 dask-scheduler --no-dashboard --no-show --host ${hn} --port ${port} &
 echo "starting workers"
-srun -o dask_worker_info.txt dask-worker ${hn}:${port} &
+srun -o dask_worker_info.txt dask-worker ${hn}:${port} --nthreads 1 &
 echo "starting gp2Scale"
-python -u run_gp2ScaleGPU.py ${hn}:${port} ${number_of_workers}
+python -u gp2Scale_example_HPC_RunScript.py ${hn}:${port} ${number_of_workers}
 

fvgp/fvgp.py

@@ -68,6 +68,10 @@ class provides all the methods described for the GP class.
         fvgp.fvGP.gp_deep_kernel_layer_width. If you specify 
         another kernel, please provide
         init_hyperparameters.
+    hyperparameter_bounds : np.ndarray, optional
+        A 2d numpy array of shape (N x 2), where N is the number of needed hyperparameters.
+        The default is None, in that case hyperparameter_bounds have to be specified
+        in the train calls or default bounds are used. Those only work for the default kernel.
     output_positions : np.ndarray, optional
         A 3-D numpy array of shape (U x output_number x output_dim), so that for each measurement position, the outputs
         are clearly defined by their positions in the output space. The default is np.array([[0],[1],[2],[3],...,[output_number - 1]]) for each
@@ -204,6 +208,7 @@ def __init__(
         x_data,
         y_data,
         init_hyperparameters = None,
+        hyperparameter_bounds = None,
         output_positions = None,
         noise_variances = None,
         compute_device = "cpu",
@@ -262,9 +267,9 @@ def __init__(
             self.hps_bounds[2:] = np.array([-1.,1.])
             init_hps = np.random.uniform(low = self.hps_bounds[:,0], high = self.hps_bounds[:,1],size = len(self.hps_bounds))
             warnings.warn("Hyperparameter bounds have been initialized automatically \
-                    \n for the default kernel in fvgp. They have to be provided to the training.\
-                    \n For instance: GP.train(hyperparameter_bounds = fvgp_obj_name.hps_bounds)\
+                    \n for the default kernel in fvgp. They will automatically used for the training.\
                     \n However, you can also define and provide new bounds.")
+            hyperparameter_bounds = self.hps_bounds
 
 
         ####init GP
@@ -273,6 +278,7 @@ def __init__(
                 x_data,
                 y_data,
                 init_hyperparameters = init_hps,
+                hyperparameter_bounds = hyperparameter_bounds,
                 noise_variances = noise_variances,
                 compute_device = compute_device,
                 gp_kernel_function = gp_kernel_function,