From 442a3212d6222696e9fe3386b14084bb15abd696 Mon Sep 17 00:00:00 2001
From: "Michael K. Wilkinson" <michael.k.wilkinson@gmail.com>
Date: Wed, 3 Apr 2024 12:26:03 -0400
Subject: [PATCH] explain need for normalizing in DemoReweighting notebook

---
 notebooks/DemoReweighting.ipynb | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/notebooks/DemoReweighting.ipynb b/notebooks/DemoReweighting.ipynb
index f7d68e8..a7a4570 100644
--- a/notebooks/DemoReweighting.ipynb
+++ b/notebooks/DemoReweighting.ipynb
@@ -144,6 +144,14 @@
     "                                         weights1=new_original_weights, weights2=numpy.ones(len(target), dtype=float)))"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Notice:\n",
+    "Setting `density=True` in `hist_settings` above means that the histograms will be drawn normalized below, so that the area under each histogram integrates to 1. This can obscure the fact that the weights produced by `predict_weights` are not normalized; if you want the number of effective events to be the same after reweighting, you must renormalize the weights manually--see the `reweight` documentation.\n"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},