Merge pull request #60 from ziatdinovmax/mc-dev-notebooks

Notebook rework for smoke tests

.github/workflows/notebook_smoke.yml

@@ -1,5 +1,8 @@
 name: notebooks
-
+
+env:
+  CI_SMOKE: True
+
 on:
   pull_request:
     branches:

examples/compare_GPs.ipynb

@@ -20,7 +20,7 @@
     "\n",
     "This is a simple notebook to compare timings and results of two different commonly used GPs. One trained using NUTS, and the other trained using SVI.\n",
     "\n",
-    "*Prepared by Matthew R. Carbone & Maxim Ziatdinov (2023)*"
+    "*Prepared by Matthew R. Carbone & Maxim Ziatdinov (2023). Last updated in October 2023.*"
    ]
   },
   {
@@ -66,7 +66,7 @@
     "id": "86iUwKxLO7qE"
    },
    "source": [
-    "Install GPax package:"
+    "Install the latest GPax package from PyPI (this is best practice, as it installs the latest, deployed and tested version). Please do not install from a GitHub url!"
    ]
   },
   {
@@ -125,6 +125,37 @@
     "gpax.utils.enable_x64()  # enable double precision"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Enable some pretty plotting."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import matplotlib as mpl"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "mpl.rcParams['mathtext.fontset'] = 'stix'\n",
+    "mpl.rcParams['font.family'] = 'STIXGeneral'\n",
+    "mpl.rcParams['text.usetex'] = False\n",
+    "plt.rc('xtick', labelsize=12)\n",
+    "plt.rc('ytick', labelsize=12)\n",
+    "plt.rc('axes', labelsize=12)\n",
+    "mpl.rcParams['figure.dpi'] = 200"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {
@@ -168,11 +199,12 @@
     "y = f(X) + np.random.normal(0., NOISE_LEVEL, NUM_INIT_POINTS)\n",
     "\n",
     "# Plot generated data\n",
-    "plt.figure(dpi=100)\n",
-    "plt.xlabel(\"$x$\")\n",
-    "plt.ylabel(\"$y$\")\n",
-    "plt.scatter(X, y, marker='x', c='k', zorder=1, label='Noisy observations')\n",
-    "plt.ylim(-1.8, 2.2);"
+    "fig, ax = plt.subplots(1, 1, figsize=(6, 2))\n",
+    "ax.set_xlabel(\"$x$\")\n",
+    "ax.set_ylabel(\"$y$\")\n",
+    "ax.scatter(X, y, marker='x', c='k', zorder=1, label='Noisy observations')\n",
+    "ax.set_ylim(-1.8, 2.2)\n",
+    "plt.show()"
    ]
   },
   {
@@ -271,7 +303,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "Note that SVI (the `viGP`) is significantly faster. SVI is usually better to use on larger datasets and is more easily scalable. In this case, they produce similar results."
+    "Note that SVI (the `viGP`) is significantly faster (use the `% timeit` magic before the command to see). SVI is usually better to use on larger datasets and is more easily scalable. In this case, they produce similar results."
    ]
   },
   {
@@ -314,7 +346,7 @@
    },
    "outputs": [],
    "source": [
-    "_, ax = plt.subplots(1, 1, figsize=(6, 2), dpi=200)\n",
+    "_, ax = plt.subplots(1, 1, figsize=(6, 2))\n",
     "\n",
     "ax.set_xlabel(\"$x$\")\n",
     "ax.set_ylabel(\"$y$\")\n",
@@ -339,6 +371,13 @@
     "\n",
     "plt.show()"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
   }
  ],
  "metadata": {