Searches using displaced leptons or jets J. Antonelli with input - - PowerPoint PPT Presentation

• J. Antonelli

with input from M. Flowerdew, P. Pais, R. Rosten, H. Russell

Searches using displaced leptons or jets

1

• J. Antonelli UMass Workshop, Nov 13

I’ve spoken with authors from ~all of the relevant searches, here are some general musings resulting from those conversations

2

• These searches are new! ~6 of the 7 searches in this talk have only been done once
• Signature-driven searches have a lot more design choices to make
• You don’t know what your signal is and you don’t know what your backgrounds are!
• Arguably a waste of resources to target a single model
• Potential overlap with existing searches is tougher to evaluate
• Lots of trial and error - this can lead to some “artifacts” that limit sensitivity
• Getting these analyses approved by the collaboration is a struggle 

• Most reviewers don’t have a good frame of reference
• These searches are great at finding software bugs in the “standard” code

subset of the choices we all had to make:

• use half of the event or all of it?
• use standard algorithms (more robust) or homemade ones (more efficient)?
• level of background to target?
• simple event selection (more inclusive) or complex (less background)?
• what lifetimes to target? - huge implications on search strategy!

• J. Antonelli UMass Workshop, Nov 13

Pros and cons of displaced lepton and jets compared to other LL signatures

3

Cons:

• 1. The SM has lots of leptons and jets
• higher backgrounds than other

displaced signatures

• 2. There exist accepted reconstruction

and ID algorithms

• they probably won’t work for your

signal

• 3. Jets are hard to trigger on
• dijet triggers around 500 GeV

in 2015 menu

Pros:

• 1. The SM has lots of leptons and jets
• they are well understood objects
• 2. There exist accepted reconstruction

and ID algorithms

• easy path to reducing backgrounds
• 3. Leptons are good for triggering
• dimuon triggers below 20 GeV

in 2015 menu

• 4. Can access lower signal lifetimes
• compared to HSCP, Disappearing

track, i.e. Simone’s “direct searches”

• J. Antonelli UMass Workshop, Nov 13

List of Run 1 displaced lepton searches

4

[cm]

tt → ~ d0 [ tt → ~ d0 [

˜ t˜ t → be bµ

beamspot

CMS Simulation [GeV]

t ~

M 400 500 600 700 800 [cm] τ Stop c

• 1

10 1 10

2

10

σ 2 ±

σ 1 ± Expected limit Observed limit (8 TeV)

• 1

19.7 fb

CMS

Excluded region

CMS eμ arxiv:1409.4789 CMS ee/μμ arxiv:1411.6977 EXO-14-12 ATLAS ee/eμ/μμ/l+DV arxiv:1504.05162

• J. Antonelli UMass Workshop, Nov 13

Displaced eμ

5

• Goal to keep search as general as possible and

simple to understand

• Didn’t aim for background-free search
• QCD MC didn’t identify it as major background
• Complicated flavor composition of background
• Factorized cuts to avoid null background counts
• Found cuts hidden in particle flow, triggers
• Simple event selection → simple recasting

planned for run 2: remove deltaR & OS cuts, improved muon trigger/reco

εμ (Run 1) εμ (Run 2)

[GeV]

muon p

∈

CMS Simulation Preliminary [GeV]

electron p

∈

CMS Simulation Preliminary [cm] muon d

∈

CMS Simulation Preliminary [cm] electron d

∈

CMS Simulation Preliminary

lepton reco. efficiencies as a function of d0 lepton selection efficiencies as a function of pT

link to study

• J. Antonelli UMass Workshop, Nov 13

Displaced ee/μμ

6

• Measured displaced tracking efficiency by using cosmics
• Background estimation leveraged symmetry in background
• Presented limits with acceptance factored out

planned for run 2: remove vertex constraint, join forces with Run 1 displaced eμ team, expand to more final states

• J. Antonelli UMass Workshop, Nov 13

Displaced ee/eμ/μμ/l+DV

7

• Additional level of tracking needed
• Couldn’t run on entire triggered dataset
• No isolation requirement on leptons!
• Electron systematics were a sticking point
• No good source of displaced leptons
• Used control sample of KS0 decays
• Need to slowly unblind
• Efficiency depends on a lot of parameters
• No efficiency model so far (studies ongoing)

planned for run 2: improvements in seeding of displaced track reconstruction 4 mm displacement on vertex can maybe be relaxed

• J. Antonelli UMass Workshop, Nov 13

List of Run 1 displaced jet searches

8

CMS displaced dijet in tracker arxiv:1411.6530 ATLAS displaced jets in ID/MS arxiv:1504.03634 ATLAS displaced jets in HCAL arxiv:1501.04020

CMS

(8 TeV)

• 1

18.5 fb

[cm] τ X c

1 10

2

10

) [pb] q q → (X

2

XX) B → (H σ

• 3

10

• 2

10

• 1

10

95% CL limits: = 150 GeV

m = 350 GeV

m ) σ 1 ±

• Exp. limits (

= 1000 GeV

m

• J. Antonelli UMass Workshop, Nov 13

Displaced dijets in tracker

9

candidates / 0.6 cm

K

10

10

10

Data Simulation transverse decay length [cm]

s

K

10 20 30 40 50 60

Data / Sim

0.9 1 1.1

CMS

(8 TeV)

• 1

17 pb

• HT trigger reduced sensitivity to SM Higgs portal
• Validated vertex reconstruction with KS sample
• Used complex multivariate discriminants 


instead of material veto planned for Run 2: jet-tagging algorithms

• maximize lifetime coverage
• remove need for dijet vertexing
• a lot of interest!
• N.B. non “refresh” analyses will take longer

+ =

arxiv:1503.05923

• J. Antonelli UMass Workshop, Nov 13

Displaced jets in ID/MS

10

• Used same displaced re-tracking dataset as displaced

DV search

• Needed to write MS tracklet reconstruction algorithm
• Different trigger strategies depending on signal model
• Used zero-bias data event overlaid on background

instead of simulated pileup

• Recasting is very model dependent, found two legs


were not independent possibilities for run 2: combine with HCAL jets and and DV search, use long-lived hadron gun to estimate efficiency, displaced muons will be in the standard reco!

• J. Antonelli UMass Workshop, Nov 13

Displaced jets in HCAL

11

• Considered both trackless jet (ECAL or HCAL) and low-EMF jet (HCAL)
• Found search feasibility was much better with the extra background rejection

from an H/E cut

• Multiple jets will still create one HCAL cluster!
• Timing cut to reduce cosmic background may be tighter than needed
• Some efficiency loss for high mass signals

planning for run 2: pileup removal will be key because of track isolation in trigger, combine with ID/MS analysis to have additional categories (ID/HCAL, MS/HCAL, etc.)

• J. Antonelli UMass Workshop, Nov 13

Run 2 outlook / conclusion

12

We’ve been through a lot of trial and error! Some Run 1 searches ended up being somewhat over-engineered (hindsight is 20/20) In Run 2, results should come significantly faster Some limitations are unavoidable from detector technology, but much improvement is being made on the software side Now that the many displaced objects have been validated, make way for combinations! (ATLAS has a head-start in this respect)

• J. Antonelli UMass Workshop, Nov 13

fin

13

• J. Antonelli UMass Workshop, Nov 13

ATLAS Displaced dilepton search parameters

14

Triggers: Mu50 Eta1.07 Photon120 Diphoton40

• J. Antonelli UMass Workshop, Nov 13

ATLAS Displaced lepton in DV search parameters

15

• J. Antonelli UMass Workshop, Nov 13

CMS Displaced eμ search parameters

16

Trigger: HLT Mu22 Photon22 CaloIdL

• J. Antonelli UMass Workshop, Nov 13

CMS Displaced ee/μμ search parameters

17

ee trigger: HLT Photon36 R9Id85 OR CaloId10 Iso50 Photon22 R9Id85 OR CaloId10 Iso50 μμ trigger: L2DoubleMu23 NoVertex 2Cha Angle2p5

• J. Antonelli UMass Workshop, Nov 13

ATLAS HCAL jet search parameters

18

L1 Trigger: narrow jet (tau seed) pT > 40 GeV

L2: log10(EH/EEM) > 1.2, no pointing tracks

L3: anti-kT jet, pT > 35 GeV

• J. Antonelli UMass Workshop, Nov 13

ATLAS displaced jet in ID/MS search parameters

19

Jet ET + ET

miss trigger:

• ffline jet pT > 120 GeV,

ET

miss > 200 GeV

Independent muon ROI trigger

• J. Antonelli UMass Workshop, Nov 13

CMS displaced dijet search parameters

20