Ready for flavour physics

27 July 2010

Event Display: "Event display of an ATLAS event with a b-tagged jet on the right."


As we are approaching the first pb-1of integrated luminosity, it is time to pave the road to new discoveries. Many ingredients are needed to build a solid foundation that we can use for our higher luminosity physics analyses. An important one is the ability of our detector to tag the flavour of the hadrons we detect. Let's take a look at how this is done and how the taggers are prepared for action…

Flavour tagging is a general term but it mostly means the identification of jets that stem from decays of b-hadrons, which are of special interest to many types of physics measurements. On short term, it will be crucial to have this identification power for finding top quarks and the first-time measurement of the top production cross-section at 7 TeV centre of mass energy. On a longer scope, it will be important for certain channels of the Higgs boson search and a valuable tool in many searches for supersymmetry.

There are a lot of different flavour taggers conceived to be used for physics analyses, too many in fact to describe them in a short article like this. However, only a handful of the taggers are being commissioned for the early physics phase and they are an interesting subset since they demonstrate the different techniques used and we can already show how surprisingly well they work.

To build a flavour tagger, we need to carefully exploit characteristics of bottom quarks to spot them in a jungle of the light quark background. We know that many b-hadrons have mean lifetimes that allow them to travel a measurable distance from the primary interaction before they decay. This is already the main idea used by most of the taggers and it illustrates how important our fine granularity and high precision tracking devices are.

Three of the early taggers use the following principle described. The first one is the JetProb tagger, which uses the displacement from the primary interaction of tracks in a jet. These are then compared to templates. This yields the track probability of being a prompt track from a W or Z decay, and therefore not from a b-hadron decay. The second tagger is TrackCounting. It also looks at tracks associated to a jet and requires them to be significantly displaced. And last but not least, we commissioned a secondary vertex based tagger, the SV0 tagger. It tries to directly reconstruct the decay vertex of the b-hadron and judge on its flavour by using the displacement of this vertex from the primary interaction.

These taggers rely strongly on a good understanding of our tracking devices since position and resolution of objects like tracks and vertices are input quantities. It is due to the hard work of many people in the performance groups that we actually do have this understanding. Therefore we were able to already show at the ICHEP conference that these taggers indeed work. More results based on the latest data are in the pipeline for upcoming conferences.

Taking plots out of conference notes and putting them here puts them out of context a bit so I'd rather show you only two selected images: First a comparison of one of the tagger weights between data and simulation. The agreement is decent and it's clear that a cut on a high tagger weight enriches the sample with b-jets. Additionally there is an event display with a displaced reconstructed decay vertex that illustrates the taggers principle. In the top right jet, you can see that the tracks in the jet can be used to form a displaced secondary vertex. The jet also contains a muon, which leads to the fourth tagger being commissioned.

This tagger looks at a different quirk of bottom quarks: A fraction of them decays to leptons. The high initial quark mass (higher than the light quark mass that is) pushes the lepton away from the direction of flight of the initial hadron. By comparing the jet direction of flight with its associated leptons direction, we can “soft lepton tag” this jet. Not only is this an interesting way of using the decay kinematics, but it also is a complementary method to the ones described before. Therefore it allows us to measure the taggers performance!

Performance measurements are among the main current projects in the flavour-tagging group. We use complementary taggers like the soft lepton tagger and physics processes like di-jet events with tagged back-to-back jets to measure the efficiencies and make assumptions about the impact parameter distributions and estimate the fake-rates. Eventually, events containing top quarks will be used for performance measurements that will give better results, but this will only happen in several months.

We have already found out that the early taggers work more or less the way we expect them to work. Now the amount of integrated luminosity is getting to the realm where we can make quantitative statements about the performance of the taggers. This is the “pavement” we need: what would a physics analysis be without good knowledge of the efficiencies and fake-rates of the used techniques?

Distribution: "Secondary vertex tagger weight. A cut on high weights enriches the fraction of b-jets."



 

Florian Hirsch,

Universitaet Dortmund