Understanding Tau Identification in ATLAS: A Brief Overview

  • Context: Graduate 
  • Thread starter Thread starter ChrisVer
  • Start date Start date
  • Tags Tags
    Atlas Identification
Click For Summary
SUMMARY

This discussion focuses on tau identification techniques in the ATLAS experiment, particularly the interpretation of figures from the paper "Understanding Tau Identification" (arXiv:1201.5466v1). Key concepts include inverse background efficiency, also known as rejection rate, and the use of boosted decision trees (BDT) for event classification. The discussion clarifies that higher signal efficiency leads to lower inverse background efficiency, and explains the process of reconstructing hadronic tau using barycenters and four-vectors of constituent clusters. The methodology for calculating the tau's energy through jet trajectory analysis is also addressed.

PREREQUISITES
  • Understanding of inverse background efficiency and rejection rates in particle physics.
  • Familiarity with boosted decision trees (BDT) for event classification.
  • Knowledge of four-vector calculations in high-energy physics.
  • Basic concepts of tau decay modes and their identification in collider experiments.
NEXT STEPS
  • Study the application of boosted decision trees in particle identification.
  • Learn about four-vector analysis in high-energy physics, focusing on tau reconstruction.
  • Explore the significance of background efficiency in collider experiments.
  • Review the methodologies for jet trajectory analysis in tau identification.
USEFUL FOR

Particle physicists, researchers in high-energy physics, and anyone involved in tau identification and analysis within collider experiments like ATLAS.

ChrisVer
Science Advisor
Messages
3,372
Reaction score
465
I have some questions concerning this paper (mainly about the tau identification):
http://arxiv.org/pdf/1201.5466v1.pdf

especially the figures 3-6...

I don't understand what is the inverse background efficiency, neither the likelihood score. As a result, I am unable to understand what info the figures give. Any help?
 
Last edited:
Physics news on Phys.org
Inverse background efficiency is also called the rejection rate. A value of 100 means you identify 1 out of 100 background events as signal.
The higher the signal efficiency you want to be, the more events you have to keep, which means you will identify more background events as signal, which gives a lower inverse background efficiency.
The different curves are different tau decay modes, pT regions, background samples and so on.

A boosted decision tree will give you a numerical value for each event, where larger values mean "more signal-like". Figure 3 shows the distribution of the BDT outputs for signal and background events. You can then choose different cut values, each cut value will give you one point in the diagrams 4-6.
 
  • Like
Likes   Reactions: ChrisVer
Thanks,
I still have some further questions in case you can help me.
In the paper I mentioned in the OP (in Sec 2), as well as here
http://arxiv.org/pdf/1409.0343v1.pdf In Sec 2, is mentioned how to reconstruct the hadronic tau.
It says:

A barycenter is formed consisting of the sum of the four-vectors of the constituent topological clusters, assuming zero mass for each of the constituents. Then, the τ_had detector axis is calculated by using clusters within ∆R =...< 0.2 around the barycenter. The four-vectors of those clusters are recalculated using the tau vertex coordinate system and the vectors are summed up.

Does that mean that the barycenter is like a disk within the ∆R<0.2 ?
And by that you are able to identify the vertex coordinate system by running back the jet's trajectory and see where the trajectories intersect?
So then you know they should come from taus and you are able to sum the jet's four vectors and calculate the tau's energy?
 
Why are you using a few page proceedings as your source when it references a more complete note?
 

Similar threads

  • · Replies 11 ·
Replies
11
Views
3K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 13 ·
Replies
13
Views
3K
  • · Replies 3 ·
Replies
3
Views
5K
  • · Replies 0 ·
Replies
0
Views
2K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 49 ·
2
Replies
49
Views
12K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K