Clean up of tutorial

Author: klapo

tutorials/tutorial20/costs-tutorial_toy-data.ipynb

@@ -510,40 +510,7 @@
    "id": "d6d35728-7d23-408b-b98e-987fe65b90c5",
    "metadata": {},
    "source": [
-    "How big should the window be? This is a non-trivial question. First, let's find the number of time steps in a single cycle for the high frequency transient feature."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 9,
-   "id": "4fa5ad74-9efe-4613-81f3-fdb9558a3627",
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Number of time steps per cycle of the transient feature: 785\n"
-     ]
-    }
-   ],
-   "source": [
-    "dt = scipy.stats.mode(np.diff(time.flatten()))[0]\n",
-    "\n",
-    "num_steps_cycle = (2 * np.pi / f_transient) / dt\n",
-    "print(\n",
-    "    \"Number of time steps per cycle of the transient feature: {:.0f}\".format(\n",
-    "        num_steps_cycle[0]\n",
-    "    )\n",
-    ")"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "e74244b0-dcb0-42fa-8e03-edfde8bc98c2",
-   "metadata": {},
-   "source": [
-    "Following the advice in Dylewsky et al., (2019) we should have a window size somewhere around 2 * the period of the fastest feature. \n",
+    "How big should the window be? This is a non-trivial question. Following the advice in Dylewsky et al., (2019) we should have a window size somewhere around 2 * the period of the fastest feature. \n",
     "\n",
     "When the window is too large and/or the svd rank is too small to fit the data well, these high frequency features are simply dropped. This behavior can be accidentally nice when wanting to ignore high frequency components but frustrating in the cases when these components need to be retained.\n",
     "\n",