Add subheadings

BOLFI.ipynb

@@ -175,7 +175,14 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "Now we can immediately proceed with the inference. However, when dealing with a Gaussian process, it is advisable to take a logarithm of the discrepancies in order to reduce the effect that high discrepancies have on the GP. In ELFI such transformed node can be created easily:"
+    "### Fitting the surrogate model"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Now we can immediately proceed with the inference. However, when dealing with a Gaussian process, it may be beneficial to take a logarithm of the discrepancies in order to reduce the effect that high discrepancies have on the GP. In ELFI such transformed node can be created easily:"
    ]
   },
   {
@@ -201,7 +208,9 @@
   {
    "cell_type": "code",
    "execution_count": 4,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "bolfi = elfi.BOLFI(log_d, batch_size=5, initial_evidence=20, update_interval=10, \n",
@@ -365,7 +374,9 @@
   {
    "cell_type": "code",
    "execution_count": 9,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "post2 = bolfi.extract_posterior(-1.)"
@@ -398,6 +409,13 @@
     "post.plot(logpdf=True)"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Sampling"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},