(r36) SusySubjet < Main

Tags: view all tags
---+ Identifying massive hadronic decays in SUSY decay chains.

This page is intended to help discussion between Are, Erckan, John and Jon and isn't intended to be used by anyone else.

---++ Project plan

We'll start by assuming that in a SUSY pair production, one decay chain has allowed us to remove all non-SUSY backgrounds, and what we are trying to do is interrogate/unpick the second decay chain to learn more about the SUSY spectrum. In particular we are looking at high ET jets and seeing they come from hadronic decays of W or Z.

John: I suggest that we use the benchmarks postulated in [[http://www-spires.dur.ac.uk/cgi-bin/spiface/hep/www?rawcmd=FIND+EPRINT+HEP-PH+AND+EPRINT+0508198][hep-ph/0508198]], particularly scenarios \alpha, \beta, \gamma and \delta. The first three have neutralino dark matter but non-universal Higgs mases (NUHM), with parameters chosen to get varying branching ratios for \chi_2^0 \to Z,H \chi_1^0, as seen in Fig. 2. I do not know offhand about the \chi^\pm \to W \chi decays. 

The point \delta has gravitino dark matter, but the neutralino is the next-lightest sparticle and its decays are invisible. Hence, all four points have 'standard' missing energy signatures.

Since there are issues in treating NUHM models using ISAJET, we might want to start with point \delta. It is in any case the most interesting, as it has relatively large masses and hence is quite challenging using conventional techniques (see Fig. 9).

After our discussion at CERN on 19 July, we decided the immediate way ahead was as follows:

   * Are to produce the PYTHIA config files from these and upload to this page. 
   * Are to run the [http://hepforge.cedar.ac.uk/hzsteer HZSteer] analysis and show us the key distributions (mass, y, ET, eta etc... as shown previously).
   * Erckan to run similar jobs with Athena ATLFast and compare to Are's.

After this, we should have a good idea how much mileage there is in this kind of study, and hopefully be in a position to start looking at more challenging SUSY scenarios and treating non-SUSY background events.

---++ Are's initial attempts

The SLHA file for John's \delta point has been generated with ISAJET 7.69, and SUSY decays added with SDecay 1.1. I checked the decays with the ones given by ISAJET and modulo small numerical differences they look the same. The file is attached. Also attached is the NLO cross sections for \delta as calculated with Prospino2.

Plots for \delta have been produced and attached (NEW version as of 10.08.06). As a first attempt we are looking for events were two W have been produced in chargino decays. The decay chain we are looking for is ~q_L -> q + ~chargino_1 -> q + W + ~neutralino_1. For a ~q_L pair, either produced directly or from gluino decays, the total branching ration into a chargino pair is 42.1%, while the charginos decay to Ws in 97.5% of cases, making this a very common event. 

A short description of the plots follow:
   * Top left: Plot of pT of jets. Cut on pT>15 and |\eta| < 2.5. In red is shown jets matched to a parton level W (by the usual distance \Delta R < 0.1).
   * Top right: Plot of jet mass m_W as given from jet algorithm. Same color coding. Cut on jet pT: 150 < pT < 600. 
   * Bottom left: Plot of jet log(pT*sqrt(y)). Same color coding. Cut on jet pT: 150 < pT < 600. Cut on jet mass: 80 < m_W < 90.
   * Bottom right: Plot of invariant mass of w+jet system. We require two W's identified in each event from the cuts given above pluss a cut on separation scale: 1.3 < log(pT*sqrt(y)) < 2.0. The W candidates are combined with the two hardest jets in the event. In red (signal) events with two decays of the type mentioned above.

Note that the three first plots have been normalized, and the cuts accumulated through the plots, which was not done in the first version.
From the distributions of jet pT and mass, cuts look to be potentially very effective. However, in the log(pT*sqrt(y)) distribution the W-tagged jets have almost the same distribution as other jets, with only a slight excess just below log(pT*sqrt(y))=2. As a result the cut on separation scale is not very effective, and in the invariant mass distribution we see that while the signal plot (red) has roughly the expected endpoints, one can show analytically that for the \delta masses we should have 247 < m_{qW} < 1272, it is burried in SUSY background.

As you can see there is no longer a top peak in the invariant mass distribution as in the first version, so this is no longer a problem. There are still too many background jets, and rather surprisingly the separation scale cut does not work very well when you look at it after adding the other cuts. I am not sure that I understand fully why this is. One possible problem is that the higgs is another source of jets at this scale. Another is the large number of jets that have the same mass, of which a fraction should have a separation scale similar to the Ws. Also, very few leptons are produced at \delta, so I have no possibility to "guarantee" little jet activity in one chain, e.g. by looking for a chain that produces two leptons. Perhaps \delta is too difficult?

One additional plot I made is a scatter plot of the separation \Delta R between the two quarks in the W decay and the pT of the W (also attached).  I find this a little interesting because it shows at what sort of pT the quarks come out in the same direction, and can make a single W jet. Above 300-400 Gev seems to be good enough, but notice how few events are out there. This is reflected in the small fraction of Ws that decay hadronically that one can find a matching, single jet, for (less than one third with the \delta R cut given above). Perhaps there is simply too little boost for the Ws? I would like to know how this compares to the WW scattering situation.

I made another attempt by trying to find a Higgs in the decay chain ~q_L -> q + ~neutralino_2 -> q + h + ~neutralino_1, throught its decay to a bbbar pair. The conclusions are the same. While the jet pT and mass cuts work, the separation scale cut doesn't. You end up with the expected signal distribution, but with too much background. The relevant plots, following the same pattern as for the W is also attached. Here there might however be some more hope. No attempt at b-tagging the jet was made, which might give enought rejection on the background to see something.
 
---++ Some tests from Jon

I have run HZSteer with Are's Pythia settings and version of hztemplate to generate an ntuple of about 10k events. I then looked in these events to find the jets and the true W's. I split the jets into two categories based upon whether they had an eta-phi match (0.1 unit) with a true W or not. Those which did I called "W jets" and the others I called "QCD jets". I then plotted the y and the subjet scale (sqrt(y)*pT) for each jet. The plots are here:

   * [[%ATTACHURL%/ytest.eps][ytest.eps]]: Test of the y cut for W and other jets.

They show a marked difference (as expected) in both distributions, with the W jets peaking near to the W mass scale. 

This is encouraging but not comparable directly to Are's plots since I haven't made any cuts on the jets at all and I can't yet pick out jets from squark decays (these will be included in the "QCD jet" sample at present). Nevertheless these are probably easy plots for you to reproduce Are? If we agree on these, that would at least remove one possible source of bugs.

I will look into this as a function of some other cuts (pT etc).

Here's the same plot with a cut at pT>250 GeV applied to both types of jets.

   * [[%ATTACHURL%/ytest250.eps][ytest250.eps]]: ytest250.eps

---++ Are's attempt at reproduction

I've tried to reproduce John's test and failed miserably. My corresponding plots are found here:

   * [[%ATTACHURL%/ytesta.eps][ytesta.eps]]: Test of y distributions

   * [[%ATTACHURL%/ytest250a.eps][ytest250a.eps]]: Test of y distributions

As usuall the red distribution is jets matched to true W's.

-- Main.JonButterworth - 19 Jul 2006

-- Main.ErkcanOzcan 
   * [[%ATTACHURL%/delta_slha.dat][delta_slha.dat]]: SLHA file for the \delta benchmark

   * [[%ATTACHURL%/delta.xsec][delta.xsec]]: Cross sections at NLO for \delta calculated with Prospino2

   * [[%ATTACHURL%/delta.ps][delta.ps]]: Some plots for \delta

   * [[%ATTACHURL%/pTvsDeltaR.ps][pTvsDeltaR.ps]]: pT of W versus \Delta R between quark decay products

   * [[%ATTACHURL%/deltahiggs.ps][deltahiggs.ps]]: Plot for higgs at \delta
Attachments
Topic attachments
I	Attachment	Action	Size	Date	Who	Comment
ps	delta.ps	manage	29.0 K	2006-08-10 - 14:27	AreRaklev	Som start-up plots for \delta
xsec	delta.xsec	manage	10.2 K	2006-07-20 - 19:30	AreRaklev	Cross sections at NLO for \delta calculated with Prospino2
dat	delta_slha.dat	manage	45.8 K	2006-07-20 - 13:02	AreRaklev	SLHA file for the \delta benchmark
ps	deltahiggs.ps	manage	27.7 K	2006-08-10 - 15:35	AreRaklev	Plot for higgs at \delta
ps	pTvsDeltaR.ps	manage	82.2 K	2006-08-10 - 15:35	AreRaklev	pT of W versus \Delta R between quark decay products
eps	ytest250a.eps	manage	17.5 K	2006-09-12 - 14:53	AreRaklev	Test of y variable
eps	ytesta.eps	manage	18.2 K	2006-09-12 - 14:52	AreRaklev	Test of y variable
Topic revision: r36 - 2006-09-12 - AreRaklev
~~Edit~~
~~Attach~~