EFT mini-workshop at IPPP

Other Institutes

Other Institutes

Ogden building
Chris Hays (University of Oxford (GB)), Michael Spannowsky

Session 1: brief review of combined EFT fits and constraints


9 TopFitter (White)

- Warsaw basis: 16 operators affecting top; data constrain 12 operators, neglect others

- correlations included where available

- should we set all but one to zero to understand individual measurements?

  -> has to be presented carefully, not an actual EFT parameter fit

- which operators are primarily constrained by quadratic terms?  cG, maybe 4-fermion?

  -> paper says linear operators included?

- scale of validity quoted?  No but removing Mtt overflow bin did not make much difference

- could you determine the goodness of fit for the SM only?

- What about alternate approach from Butterworth: assume data is SM and constrain c's assuming

  they are zero?  Is that a circular approach?

- any issue with fitting particle-level vs parton-level?  No.

 -> should experimentalists be the ones evaluating particle->parton uncertainties?

- should we make measurements of distributions before background subtraction?  Yes.


9:30 Higgs + dibosons + jets (Plehn)

- Include Dphi(jj) but do not fit CP-odd operators for philosophical reasons

- no experimental combination of dibosons (only ATLAS+CMS ZZ)

- quark 4-fermion operators constrained by 2-3 jets, interference term is zero

- three-gluon operator constrained by >=5 jets

- constraints are at a high scale, can we apply them to Higgs physics?  Consensus is yes.

- What about (DG)^2?  Not in Warsaw, becomes a 4-fermion operator (constrained by 2-3 jets).

- m4l gives little information within the context of VBF -- adding ggF should increase its contribution


10 EW precision & low energy (Trott)

- EDM strongly constrains CP-odd operators in Higgs & top physics

  -> motivates fits either requiring MFV or CP-even

- Benefit to using mW,mZ,GF input scheme because there are two scales (alphaEM would be third)

- EFT uncertainties can be important for precision EW data (higher order terms ~10% of leading terms)



Session 2: status of constraints within LHC experiments


11 Higgs STXS vs data (Hays/Zemaityte)

- note that including quadratic terms becomes more valid when measurements are precise

- should only truncate based on IR physics or MFV symmetry assumption

  -> will truncate using only assumed symmetries

- should remove coefficients from operators and include top loop in ggF

- should include ~1% uncertainty on S parameter from higher order effects


11:30 Higgs DiffXS (Pilkington)

- which is more sensitive: two 1d distributions with correlation or 2d distribution?  Could check.

  -> significant gain expected from 2d distribution in VBF


12 Electroweak (Lohwasser/Price)

- Publish observed numbers of data events?  Useful if validated Delphes model provided.

- Should quote limits from fully optimized reconstructed data and compare to those from unfolded data


12:30 CMS (Milenovic)

- will CMS provide correlations in diffXS?  Aiming to do so, probably starting from 2017 data

- will CMS present results using YR4 PO notation?  Yes, along with historical presentation.



Session 3: future tools/studies in dim-6 (LO & NLO) and dim-8 EFT


14 Warsaw LO UFO (Brivio)

- two implementations of Warsaw basis will soon be available

- ~30 operators if considering W/Z/H pole observables

- first fit probably also needs qq->ttbar?


14:30 Warsaw NLO QCD (Maltoni)

- many NLO SMEFT calculations available

- need to implement four-fermion terms before releasing NLO Madgraph implementation

  -> Provide ggF as a starting point to study?


15 Dim 8 (Sanz)

- no new q^2 dependence coming from dim-8 -- could be due to basis?

- Need to check vertices with fermions: q^2 dependence might come from spinors

- useful for systematic checks

- can VBF be added?  more complicated but possible in principle



Session 4: fit issues


16 Uncertainties (Pecjak)

- Can Madgraph include the running of the EFT operators so that we can use it for EFT scale variations?


16:30: Validity (Hays/Spannowsky)

- For a given validity scale what range of c should we allow?  Up to ~10 okay.

- What should we do about coefficients that are directly probed above the scale used in the fit?  Set to zero?


There are minutes attached to this event. Show them.