LHC TOP WG meeting

23 Nov 2020, 14:00 → 24 Nov 2020, 18:40 Europe/Zurich

Zoom mtg only (CERN)

Maria Aldaya Martin (DESY), Michelangelo Mangano (CERN), Reinhard Schwienhorst (Michigan State University (US))

This periodic open meeting of the LHC TOP Working Group takes place twice a year at CERN. The meeting provides a forum for discussion between experimentalists and theorists on the latest developments in top quark physics. Everybody is welcome to attend; please register using the link on the left. 

Due to the Covid-19 pandemic restrictions, the meeting will take place remotely via zoom:

https://cern.zoom.us/j/96361746982?pwd=VllVMGw1OTdKSTNtWFJlYit0dkJkQT09 

Meeting ID: 963 6174 6982

Passcode: 112020

Monday:

ATLAS highlight, b-fragmentation:
- Unfolding procedure and neutrinos in jet: in PDF fits, see agreement at particle level, but then disagreement at parton level
- Here, only unfold to stable charged particles (i.e. don’t try to account for or include neutrinos), that keeps dependence on neutrinos at a minimum
- Unfolding to b-hadron would require this additional unfolding step, which has some landmines in it and wasn’t attempted yet
- Fraction of B** mesons that are produced? Since only unfolding to weakly decaying B hadron, this fraction is not measured, and the measurement is not sensitive to it
- Relative rates of production of B mesons and baryons, this measurement has some sensitivity to it, but these were not directly measured.

CMS highlight, CP violation CEDM:
- EFT operators that mimic this CEDM? Have not studied this, but EFT experts in chat say there are CTg and CTW, and for both the imaginary part should lead to this type of CP violation.
- from the chat: most of the current CMS EFT studies are on real coefficients. Would be very interesting to extend to imaginary parts and set the stage for such comparisons

tW and ttbar:
- At NLO, red error band is 7-point scale variation on ttdec
- Scale variation uncertainties - they generally find that ttdec and hvq under-estimate these
- bb4l is only generator where scale uncertainties are propagated through the decay properly
- p8 shows the shape effects of the Y variations are particularly large at low Mtt, studies here show that tW/ttbar interference modeling also shows large variations here, but those are still within the scale uncertainties.

Simultaneous extraction of top mass and alpha_s:
- why not use 2d measurements? Theory predictions for these were not available until very recently
- Systematic uncertainties dominated by theory modeling - these have large impact but are not included here, in particular off-diagonal terms - off-diagonal terms make it very difficult to perform the fit
- trying to extract absolute unfolded distributions from normalized ones is not possible considering large systematic uncertainties - only done for one CMS case and not used in final results.
- Experimental measurements assume a fixed top mass in unfolding
- From chat: The reweighting scheme on p13 with using the p-value of the chi2, doesn't make sense, this gives measurements with zero sigma difference to the prediction a weight of 1 and measurements 1 sigma away a wieght of 0.32
- Discussion continues during the break.
- Concerns on the study raised by the experiments: presenting these top mass and alpha_s results with such precision but not taking into account the dominant systematic uncertainties or systematic correlations among measurements, can be very misleading

Common MC:
- Why not using EvtGen? Not implemented in CMS
- Impact of EvtGen might be small, would have to check. If impact is large, that would make it difficult to use the fully simulated version of this sample, since b-tagging would need to be re-calibrated
- EvtGen is close to Pythia 8, but larger differences to Herwig, for example
- ATLAS document benchmarking EvtGen: https://cds.cern.ch/record/1709132

ttbar+quarks comparison ATLAS and CMS:
- ATLAS, CMS results in different channels are all consistent for ttbb, experiment is above theory prediction, even if plotted differently
- Only CMS all-had looks different, but also has large uncertainties
- Modeling in tt+charm - dedicated charm tagger, where b is a mistag - see p34
- Labeling of additional b quarks  in the reconstruction - b-tag score is used right now, can do better? Maybe change this at particle level, improvement over ghost tagging? - yes, there could be improvements, but would have to study this carefully, would need to be physics motivated, separate b’s from tops from those produced with top

Input from HiggsXSWG:
- difference between ATLAS and CMS PH+PY ttbb scale uncertainties? There are some different settings (different scales), but it’s not clear why the variations should be different - likely due to scale choice, this impacts variation band. CMS scale choice is also suggested by theory
- Difference between ATLAS and CMS is small for the same generators as long as are looking within relevant phase space
- Want to agree on not just nominal model for ttbb, but also prescriptions for systematic uncertainties. Focus here is not to get a common sample
- What is needed to constrain these by ttbb measurements? Need differential measurements with 10% uncertainty or better, likely not achievable in Run 2

ttZ comparison ATLAS and CMS:
- ATLAS differential results will be used in future EFT fits, but will have also dedicated EFT samples in analysis in next ATLAS round
- Including EFT operators in backgrounds, in detector-level analysis (like CMS), where it is currently not included, not even for tZ - but at the moment this is negligible
- Similar for using unfolded distributions in EFT fits, would want to also estimate impact of EFT operators on background
- improvements in precision requires mainly more data, especially if want to reduce modeling uncertainty. May even need more than Run2+Run3 data

Tuesday:

EFT workshop summary:
- is everyone in top now using SMEFT at NLO? Are we waiting for any other developments? For CP-even operators that involve top, all of it is there, and we can switch. For analyses that involve CP-odd operators, not yet available at NLO, stay at LO.
- Flavor question: For top analyses, separate out top couplings for each coefficient, while Higgs doesn’t do this at the moment.
- SMEFT@NLO comes with a list of practical questions on the details, like details about Madgraph, how to interface to Madspin, what the default settings for top should be, etc. Not clear that all of this is sufficiently implemented and can fully be used.
- STSX which are used in Higgs have so far not been needed in top since lots of relevant differential distributions are available. But also issue of operators affecting backgrounds. Also, STSX doesn’t include acceptance corrections, so this is becoming more similar to differential XS measurements as used in top.
From chat, response by Gauthier:
 1) NLO for FCNC in a UFO by Cen et al? It is in our plans to update the NLO-capable https://feynrules.irmp.ucl.ac.be/wiki/TopFCNC to match the LHC TOP WG conventions
 2) we should keep in mind that we need to be able to translate the flavour assumptions in dim6top to SMEFTsim. I believe the authors indeed provided restriction cards to match LHC TOP WG conventions. In the context of the EFT validation effort [1906.12310], equivalent cards for SMEFTsim and dim6top were also produced (see the note and https://bazaar.launchpad.net/~rwgtdim6/mg5amcnlo/plugin_eft_contrib/files/head:/example)

learning from PDFs:
- When different distributions give different fits for parameters of interest, the key is to understand prediction and experimental data in detail, need to discuss with both theorists and experimentalists
--> (similar comments as to Matthew's talk): from the ATLAS and CMS sides, what additional information should be made public so that all the necessary information can be included in their fit?
- For example, as NNLO calculations became available, discussion was to stick to one consistent order for all fits (NLO) or to use NNLO for some of the distributions (jet and ttbar) - This improves prediction, but is now inconsistent in the order of alpha_s - will be a similar issue with EFT fits


Lessons from Higgs for top:
- STSX is still integral, these are key ingredients in combination

Discussion:
- Some top quark channels don’t have any large backgrounds (ttbar dilepton), thus these are actually very nice/important
- Others (4-top, ttH) have a lot of top backgrounds
- Unfolding checks cover a lot of the possible EFT operator effects and check for biases
- However, there are always special cases (see Adinda’s example for Z4l in Higgs), in top this could be tW in ttbar dilepton, for example.
- Including both signal and important SM backgrounds in the signal would help with this (eg 4-top and ttW (and ttZ) together)
- Unfolding signal and backgrounds together to particle-level seems ok, but this is not straightforward to parton level (i.e. top quarks)
- To preserve fitting analyses for the future, could use JSON format, which is becoming more standard

Elastic top production:
- Can measure top mass? Clean environment. But won’t have enough statistics to get a precise measurement

Exclusive top production:
- Cannot measure COM system due to neutrino, but can at least get lower limit. Could matching this to forward protons give an additional constraint? Only for fully exclusive production

Comparison of ATLAS and CMS differential distributions:
- Statement that once all theory uncertainties are included, all MC-data disagreements are covered was also found to be true in ATLAS studies
- Where trends are different between ATLAS and CMS, would be good to find out exactly what the reason is. 
- Could also make more comparisons at fiducial level - yes, this is planned in context of common MC
- Parton level is the same for both experiments, only particle level is different. So can already now define common phase spaces at parton level. 
- to be able to compare at parton level we need discussion/agreement again on common binning - currently only possible for the delta_phi distribution at 13 TeV and the tt 1D differential cross sections at 8 TeV in l+jets 
- At particle level, will have to deal with the differences in the details of the object definitions. Need to discuss this and come to agreements on updated particle-level definitions
- Should go into more detail on these comparisons (ATLAS compared to CMS and data compared to prediction) at next closed meeting

Sherpa:
- p18: the speed-ups here affect very few samples, but even within samples, some distributions are affected while others aren’t
- MC samples used by ATLAS and CMS have been older (2.2.2), now using more recent ones (2.2.7, 2.2.8)

MiNNLO:
- At the moment use stable tops in this calculation
- While working on this, would be wiling to release previous versions of the MiNNLO software, which would be useful for studies to experiments.
- Adding NLO top decay to this? At the moment not possible because it would be some work to have full top-quark polarization information.
- p5: 2-loop corrections for ttbar are public (Czakon et al), don’t need color indices here.

Monday, 23 November

14:00 → 14:10 Introduction

Speakers: Maria Aldaya Martin (DESY), Michelangelo Mangano (CERN), Reinhard Schwienhorst (Michigan State University (US))

LHCtopWG_open_meeting_intro_nov20.pdf

14:15 → 14:35 ATLAS highlight: measuring b-fragmentation in ttbar events with ATLAS

Speaker: Chris Pollard (Deutsches Elektronen-Synchrotron (DE))

LHCTopWG20201123.pdf

14:45 → 15:05 CMS Highlight: Search for CP violating anomalous top quark coupling in pp collisions at sqrt(s) = 13 TeV

Speaker: Seungkyu Ha (Yonsei University (KR))

lhctopWG_ttbarcpv_sk_v9.pdf

15:15 → 15:30 tW production in ttbar phase space

Speaker: Tomas Jezo (University of Zurich)

TJezo.pdf

15:50 → 16:05 Simultaneous extraction of mt and alphaS from LHC tt differential distributions

Speaker: Matthew Lim

mtas.pdf

16:10 → 16:40 Coffee Break

16:40 → 16:55 Common MC update

Speakers: Giulia Negro (Purdue University (US)), Michael James Fenton (University of California Irvine (US))

commonMC_LHCtopWG_23Nov2020.pdf

17:00 → 17:20 tt+quarks production in ATLAS and CMS

Speaker: Seth Moortgat (Vrije Universiteit Brussel (BE))

ttquark_CMSandATLAS_LHCTopWG_SethMoortgat.pdf

17:30 → 17:50 Input from LHC Higgs XS WG on tt+quarks needs

Speaker: Andrea Helen Knue (Albert-Ludwigs-Universitaet Freiburg (DE))

ttbb_Studies_ATLAS_CMS_AndreaKnue.pdf

18:00 → 18:20 tt+Z comparison ATLAS and CMS

Speaker: Fabio Cardillo (Univ. of Valencia and CSIC (ES))

TopLHC-23-11-20.pdf

Tuesday, 24 November

14:00 → 14:20 LHC EFT workshop summary and next steps for EFT interpretations

Speaker: Eleni Vryonidou

TopWGNov.pdf

14:30 → 14:50 PDF fits including top and learning from PDF fits for EFT fit

Speakers: C.-P. Yuan (Michigan State University), C.-P. Yuan (Michigan State University)

lhctopwg_2020_yuan.pdf

15:00 → 15:20 What can top EFT fits learn from Higgs EFT fits

Speaker: Adinda De Wit (Universitaet Zuerich (CH))

HiggsEFT_TOPWG.pdf

15:30 → 16:00 Discussion on how to include EFT dependence in unfolding/background

If you have any input on how to handle the dependence of acceptances and background estimates on anomalous operators, send your points to the EFT contacts listed here.

Speakers: Kirill Skovpen (Ghent University (BE)), Laura Barranco Navarro (Stockholm University (SE)), Peter Berta (Deutsches Elektronen-Synchrotron (DE))

TopEFTxBackgrounds.pdf

16:00 → 16:30 Coffee Break

16:30 → 16:45 Elastic production of top quarks

Speaker: James William Howarth (University of Glasgow (GB))

ATLAS_LHCTopWG_ForwardTops.pdf

16:50 → 17:05 Exclusive top-quark production

Speaker: Marek Tasevsky (Czech Academy of Sciences (CZ))

Tasevsky_LHCTop_241120_v2.pdf

17:15 → 17:35 Higher-level comparison of multiple distributions for MC generators

Speaker: Dr Marino Romano (INFN Bologna (IT))

Romano - Higher-level comparison of multiple distributions for MC generators.pdf

17:45 → 18:05 Modeling of uncertainties in Sherpa

Speaker: Frank Siegert (Technische Universitaet Dresden (DE))

2020-11-24-Sherpa-LHC-Top-WG.pdf

18:15 → 18:35 MiNNLOPS and prospects for top

Speakers: Javier Mazzitelli (Universidad de Buenos Aires (AR)), Javier Mazzitelli

mazzitelli_MiNNLO.pdf