Chapter 5:
1. If you have higher pT, you want to make wider jet radius in order to catch the radiation and stuff
   1. Why? If you have higher pT, you get collimated jets but the phase space is larger, so you need larger radius jet to capture all particles
2. If you are picking better choices for jet algorithm, you want the dijet resonance mass to be centered around actual resonance and the width to be more or less consistent
   1. There are many factors that affect this
3. 5.2 pile up subtraction:
   1. You can subtract average fluctuations but not the fluctuations
   2. The wider the jet area, the better is pT resolution you get
4. Substructure:
   1. Use anti-kt to find jets and use C/A to recluster the jet
   2. It is very uncommon to use same algorithm to recluster jet
5. Cone algorithms:
   1. Tevatron and stuff are super computationally expensive, scale as N^3
   2. In 2008, anti-kt and fastjet came out and no one uses cone algorithms anymore
6. 3 prong decays of top:
   1. Ttbar all had channel used 3 prong decays to measure cross-sections
7. Top mass:
   1. Tevatron has a small uncertainty ~0.5 GeV
   2. Same as ATLAS, CMS
   3. Now people are debating what we mean by top mass
8. Why R=0.4?
   1. Historical
   2. CMS used R=0.5 throughout Run 1 which made comparison between ATLAS and CMS impossible
   3. Now they moved on to R=0.4
9. This paper is well written and recommended for thorough reading
10. What have been developments since this paper?
   1. Nsubjettiness
   2. Grooming is another area
   3. Softdrop is now everyone’s fav. tagger
   4. And multi prong tagging
   5. Subjet b tagging and b tagging within jets has considerbaly improved
   6. Use machine learning on everything improves
   7. Precision QCD using jet substructure
11. If you are doing PDF calculations, then you should know DGLAP equation and QCD evolution, discussed in second link
   1. Bjorken scaling
12. Link 2, figure 17:
   1. Not IR safe
   2. So this is scaled as sqrt(alpha_s)
13. q/g discrimination:
   1. More forward stuff is quark like and central stuff is more gluon like
   2. This is due to event selection:
      1. If protons have similar energy and partons have similar energy then they are more central or else they both will move in an eta direction and partons wont have same x
   3. No. of particles inside a jet is a good discriminant for q/g tagging but it is not IRC safe.
      1. There was a paper proving emperically that the multiplicity is best metric for this discrimination
      2. Paper is arxiv: 1704.06266
      3. Casimir scaling gets saturated with stats while poisson scaling variables get asymptotically better with stats
14. Where have we successfuly used q/g tagging in ATLAS?
   1. Explicitly, few places
      1. High mass resonance search: Looks at m_qq 
      2. CMS uses it much more even in subjet tagging. In ATLAS, we developed calibratedd q/g tagging very recently
   2. Implicitly, everywhere. Jet calibration uses q/g tagging
   3. Hypothetically, if there is a signal 
15. PETRA: gluon discovery