Merge branch 'gh-pages' of github.com:FNALLPC/jets-hats into gh-pages

_episodes/01-jets101.md

@@ -76,6 +76,10 @@ These algorithms follow this recipe:
 > ## How the different jet algorithms look like in our events?
 > <img src="../fig/episode1/JHEP04_2008_063.jpg" alt="" style="width: 600px;"/>
 > Comparison of jet areas for four different jet algorithms, from "The anti-kt Clustering Algorithm" by Cacciari, Salam, and Soyez [JHEP04, 063 (2008), arXiv:0802.1189].
+>
+> In the figure, you can observe the high irregularity in the kt algorithm's area, the conical shape of the anti-kt jets illustrating this algorithm's preference for hard radiation, and the smaller effective radius of the SIScone jets. This is due to the split-merge procedure, which results in smaller jet areas and two jets being resolved where there is just one grey jet.
+>
+> Despite their differences, all these algorithms share a crucial feature — infrared and collinear safety. This ensures that the presence of a hard particle or a softly radiated gluon does not alter the number or contents of the jet. This condition is essential for making meaningful comparisons between theoretical predictions and experimental data.
 {: .callout}
 
 Some excellent references about jet algorithms can be found here:
@@ -255,6 +259,8 @@ Also possible to customize nanoAOD. JME/BTV have their extended format with more
 
 
 > ## Note
+> Even though the stored jets are corrected in MiniAOD and NanoAOD, these corrections should be removed, and the latest version recommended by the JME should be applied. This is required by the JME POG.
+> 
 > There are several advanced tools on the market which allow you to do sophisticated analysis using nanoAOD format, including [RDataFrame](https://root.cern/doc/master/classROOT_1_1RDataFrame.html), [NanoAOD-tools](https://github.com/cms-nanoAOD/nanoAOD-tools), or [Coffea](https://github.com/CoffeaTeam/coffea). We encourage you to look at them and use the one you like the most. However, we are going to use coffea for this tutorial.
 {: .callout}
 
@@ -290,4 +296,3 @@ A short list of jet properties that we can find in nanoAOD are:
 
 
 {% include links.md %}
-

_episodes/02-Pileup_jetID.md

@@ -50,7 +50,9 @@ We need to simulate out-of-time interactions, time structure of detector sensiti
 Many clever ways have been devised to remove the effects of pileup from physics analyses and
 objects. Pileup affects all objects (MET, muons, etc.). We are focusing on jets today.
 
-### $\rho$ pileup correction
+To mitigate PU in jets, we can do it in two steps, first at particle-level and then at the jet-level.
+
+### jet-level PU mitigation: $\rho$ pileup correction
 
 Imagine making a grid out of your detector, then $\rho$ is the median patch value (pT/area). Therefore, the corrected jet momentum is:
 $$p_T^{corr} = p_T^{raw} - (\rho \times area)$$
@@ -109,38 +111,35 @@ There are 12 interactions before and 3 after.
 > <img src="../fig/episode2/efracs_particles_8TeV.png" alt="Jet Composition MC" width="380px" />
 {: .solution}
 
-### Charged Hadron Subtraction (CHS)
-
-Tracking is a major tool in CMS. We can identify most charged particles from non-leading primary vertices, CHS removes these particles. 
-
-
-### PileUp Per Particle Identification (PUPPI)
+### particle-level PU mitigation
 
-Unfortunately, pileup is not really isotropic, it is uneven:
+Unfortunately, pileup is not really isotropic, it is uneven. Pileup particles produce additional tracks and deposits in the calorimeters, which are also reconstructed as PF candidates and can overlap with PF candidates belonging to a jet.
 
 <img src="../fig/episode2/puppi_rho.svg" alt="">
 
-PUPPI is trying to have an inherently local correction based on the following information: A particle from the hard scatter process is likely near (geometrically) other particles from the same interaction and have a generally higher pT. We expect particles from pileup to have no shower structure, have generally lower pT, and be uncorrelated with particles from the leading vertex.
+The paticle-level PU mitigation techniques ensure an inherently local correction. The two techniques used in CMS are:
 
-<img src="../fig/episode2/puppi.svg" alt="">
+- **Charged Hadron Subtraction (CHS)** mitigates this charged pileup contribution by removing charged particles originating from pileup vertices. However, it is limited to the tracker-covered region and does not account for neutral pileup contributions.
+- **PileUp Per Particle Identification (PUPPI)** complements CHS by assigning a probability to each particle, quantifying its similarity to pileup. It operates under the assumption that particles originating from the hard scatter process are likely to be near (geometrically) other particles from the same interaction and generally have higher pT​. In contrast, particles from pileup tend to lack shower structure, have lower pT, and are uncorrelated with particles from the leading vertex. These estimated probabilities are then used to assign weights to the four-momentum of the particles.
 
+<img src="../fig/episode2/puppi.svg" alt="">
 
-### Exercise 2.2
+#### Exercise 2.2
 
 > ## Open a notebook
 >
 > For this part open the notebook called `Pileup.ipynb` and run the Exercise 2.2
 {: .checklist}
 
-> ## Discussion 2.1
+> ## Discussion 2.2
 > Do you see any difference in the jet pt for CHS and PUPPI jets? Where you expecting these results?
 {: .discussion}
 
 ## Pileup reweighting
 
-Start with chosen input distribution – the instantaneous luminosity for a given event is sampled from this distribution to obtain the mean number of interactions in each beam crossing. The number of interactions for each beam crossing that will be part of the event (in- and out-of-time) is taken from a poisson distribution with the predetermined mean. The input distribution is thus smeared by convolving with a poisson distribution in each bin. This is what the observed distribution should look like after the poisson fluctuations of each interaction
+<!-- Comment: Start with chosen input distribution – the instantaneous luminosity for a given event is sampled from this distribution to obtain the mean number of interactions in each beam crossing. The number of interactions for each beam crossing that will be part of the event (in- and out-of-time) is taken from a poisson distribution with the predetermined mean. The input distribution is thus smeared by convolving with a poisson distribution in each bin. This is what the observed distribution should look like after the poisson fluctuations of each interaction -->
 
-<img src="../fig/episode2/pu_reweight.svg" alt="" style="width:50%">
+<!-- Comment: <img src="../fig/episode2/pu_reweight.svg" alt="" style="width:50%"> -->
 
 The __Goal__ of the pileup reweighting procedure is to match the generated pileup distribution to the one found in data:
  * Step 1: Create the weights
@@ -273,3 +272,20 @@ C++ from a miniAOD file:
 
 {% include links.md %}
 
+## Noisy event filters
+Detector noise, cosmic rays, and beam-halo particles can lead to large anomalous missing energy in the detector and can be an indication of problematic event reconstruction. To ensure good event reconstruction, JME POG requires the use of Noisy event filters, formerly called MET filters. For 2018, you can find the list of recommended filters on the [twiki](https://twiki.cern.ch/twiki/bin/view/CMS/MissingETOptionalFiltersRun2#Run_2_recommendations).
+ 
+### Exercise 2.5
+
+Here we are going add the Noise filters from the[twiki](https://twiki.cern.ch/twiki/bin/view/CMS/MissingETOptionalFiltersRun2#Run_2_recommendations).
+
+> ## Open a notebook
+>
+> For this part open the notebook called `Pileup.ipynb` and after Excercise 3, apply the filters for 2018 in an `AND` combination.
+{: .checklist}
+
+> ## Question 2.4
+> How many events do you reject with this requirement?
+{: .challenge}
+
+