boosted higgs and jet substructure
DESCRIPTION
Boosted Higgs and jet substructure. Mrinal Dasgupta The University of Manchester Atlas-UK Higgs Workshop, Birmingham, 25 September, 2014. Overview. Quick introduction to jet substructure and boosted particle searches/studies. E xamples of recent (and not so recent) Higgs studies. - PowerPoint PPT PresentationTRANSCRIPT
Boosted Higgs and jet substructure
Mrinal Dasgupta
The University of Manchester
Atlas-UK Higgs Workshop, Birmingham,
25 September, 2014
Overview• Quick introduction to jet substructure and
boosted particle searches/studies.• Examples of recent (and not so recent) Higgs
studies.• Introduction to substructure tools, challenges
and open questions.• Understanding substructure better.• The future: creating robust tools for future
applications (LHC run 2 and beyond)
Introduction
Period since 2008 has seen emergence of a vibrant research area dedicated to use of jet substructure for LHC discoveries.
Basic idea goes back to 1993 and work by Seymour.
It all started with the Higgs….
The famous 2008 “BDRS” paper. Discovery potential for Higgs ~ 120 GeV in VH production with H to bb.
Basic ideas
Signal vs background : tagging and grooming
How to tell signal (e.g. jet originating from a Higgs decay) from “boring” QCD jets i.e. tagging signal?• Exploit differences in “energy-sharing”. Cut on z discriminates against
background.• No large-angle soft radiation from colour singlet Higgs.
How to clean signal from contamination i.e. ISR, UE pile-up? Removal of soft large angle junk.
We have 10-20 different methods invented for this with > 100 papers over 5 years!
Since BDRS
• LHC Run 1 has shown that these methods actually work. Run 2 with higher energies will mean they become very important methods.
• And if the future holds 100 TeV colliders everything will be boosted!
Examples: ttH
Higgs production in association with tt was taken off list of promising discovery channels due to poor S/B. Important also for direct access to top Yukawa coupling.
Using fat jet methods and top+Higgs tagging Plehn, Spannowsky Salam turned S/B ~1/9 to S/B ~ 1/2 for same significance.
Examples: Boosted Higgs pair production
Presented by J.Rojo at BOOST 2014
Examples: hhvv couplingF
Requires a combination of resolved and boosted analysis. Also work on SM hh production with h to bb by Konstantinides et. al
Examples: CP properties
Uses jet substructure to reconstruct Higgs and study angular variables that discriminate between SM and BSM structure of HVV vertex. BSM contributions from higher dimensional operators enhanced by selection cuts in boosted analysis. Godbole, Miller, Mohan White 2013 and 2014
So what is left to do?The techniques are valuable in a variety of contexts and viable. However a number of open questions remain:• There are numerous tools (Taggers/groomers). Which one to
pick?• How do we compare tools? How do results obtained depend on
many parameters of the taggers? Optimal values?• How do we know we have made a discovery and have not
unearthed a feature of the taggers?• How to quantify theory uncertainties in substructure studies?
What tools to use? MC? Fixed-order? Combinations thereof?• Can results depend on shower models, tunes, jet algorithms?
What we need are formulae….
QCD jet mass
Comparison of taggers
Taggers look similar…..
Comparison of taggers
But only for a limited mass range. How to understand what we are seeing? Position of kinks etc? Calls for analysis and calculation.
Analytical vs MC : trimming
MD Fregoso Marzani and Salam 2013
Analytical vs MC: pruning
MD Fregoso Marzani and Salam 2013
Analytical vs MC mMDT (BDRS)
Non perturbative effects
Tagger performance
We now have a much better idea of the factors behind tagger performance and are using this to develop better tools
Outlook• Boosted object techniques have already been demonstrated to
be of great value for several Higgs studies. Many examples of previously unfavourable channels being converted to promising avenues.
• The focus now is on understanding the tools both in and beyond perturbation theory.
• The focus is on performance i.e signal significance but also on robustness and reliability.
• The understanding of substructure is still in its infancy.