A complete top-quark EFT implementation

Under the umbrella of the LHC TOP WG, common standards and prescriptions were established for the EFT interpretation of top-quark measurements at the LHC. They are summarized in the note at Details concerning the present UFO model implementation are provided in Appendix B.1.

EFT degrees of freedom that are natural for the description of top-quark processes were defined as linear combinations of Warsaw basis operator coefficients. They match the interference structure of the EFT with SM amplitudes and give a direct parametrization of the top-quark couplings to physical W and Z gauge bosons (see Appendices C, D, and E of the note for degrees of freedom definitions).

The CKM matrix is approximated as a unit matrix. Masses and Yukawa couplings of all fermions except for the top and bottom quarks are neglected by default. A U(2)q+u+d flavour symmetry is imposed on the first two generations of quarks (see Section 4 of the note for details). All operators of the Warsaw basis involving a top quark and satisfying this flavour assumption are included (four-quark, two-quark, and two-quark-two-lepton operators). Baryon and lepton number violating operators are not included. In total the model includes O(90) flavour-conserving degrees of freedom which have a DIM6=1 coupling order.

Enhancing the U(2)q+u+d flavour symmetry to U(2)q×U(2)u×U(2)d (the baseline scenario considered as prescription) is done by neglecting the 10+10 CPV supplementary degrees of freedom. A further restriction to the top-philic can also be obtained following the constraints given in the note.

Top-quark FCNCs are also included by allowing one quark bilinear in each operator to break the U(2)q×U(2)u×U(2)d baseline flavour symmetry requirement and couple the third generation with either the first or the second. Operators with one light and one heavy quark, one light quark one heavy quark and two leptons, one light quark and three heavy quarks, three light and one heavy quark are included. The O(300) degrees of freedom are assigned a FCNC=1 coupling order. Requiring FCNC=0 at the generation level is required to not consider them.

Two versions of the models are provided dim6top_LO_UFO and dim6top_LO_UFO_each_coupling_order. In the second one, each degree of freedom is assigned an individual coupling order, like DIM6_ctZ, DIM6_cQQ1, or FCNC_cqq11x3331, in addition to the DIM6 or FCNC one. This allows for the selection of individual degrees of freedom interferences at the generation level in MG5_aMC@NLO. The syntax:

> generate p p > t t~ FCNC=0 DIM6=1 DIM6^2==1 DIM6_ctZ^2==1
> generate p p > t t~ FCNC=0 DIM6=1 DIM6^2==2 DIM6_ctZ^2==1 DIM6_ctW^2==1

would for instance respectively retain only the interference between SM and ctZ amplitudes, and the interferences between ctZ and ctW amplitudes, in top pair production. The FCNC=0 specification excludes FCNC interactions, DIM6=1 allows for at most one operator insertion at the amplitude level (to be removed to include multiple insertions too), DIM6^2==n specifies the overall EFT order of the selected squared amplitude, DIM6_ctZ^2==n specifies the order of the squared amplitude in the specific ctZ operator coefficient.

Only tree-level simulation in the unitary gauge is possible. Loop induced couplings of the Higgs boson to pairs of gluons, photons, or are not included.

A positive QED=n coupling order is also assigned to operators involving n Higgs doublet fields in the unbroken phase to compensate for the QED=-1 coupling order of the Higgs vev which appears in the broken phase.

The bottom quark is massive by default. Switching to the five-flavour factorization scheme can be achieve by using a restriction card in which MB is set to 0, or by redefining

> define p = p b b~
> define j = p

before process generation and setting MB to 0 in the param_card (setting ymb to 0 may also be desired, for consistency).

Benchmark results for the linear and quadratic dimension-six EFT dependences of total rates are provided in Tables 10-21 of the note, for processes like pp→tt̅, tt̅bb̅, tt̅tt̅, tt̅e⁺ν, tt̅e⁺e⁻, tt̅γ, tt̅h and pp→tj, te⁻ν, tje⁺e⁻, tjγ, tjh.

Please refer to for further details and when using this model.

Contact persons are Gauthier Durieux and Cen Zhang.

Feynrule files: dim6top.m,,

UFO model files: dim6top_LO_UFO.tar.gz, dim6top_LO_UFO_each_coupling_order.tar.gz


  • 2020-05-19:
    • changed the electroweak FCNC dipole basis to (ctA...,ctZ...) instead of (ctW...,ctZ...)}
    • added five-point interactions (thanks to CMS TOP EFT group, Florencia Canelli, Reza Goldouzian, Abideh Jafari, Seth Moortgat)
    • added a configuration option to switch from the (aEWM1,Gf,MZ) electroweak inputs to the (Gf,MZ,MW) ones. As inputs are different, different numerical results will be obtained in the two cases. See for the hack, to be used with great care.
    • added a configuration option to normalize all chromo-dipole operator coefficients (ctG and FCNC cousins) with a factor of the strong coupling G. This is currently required for jet merging not to crash. Note however that the jet merging code was not written with non-SM interactions in mind. So results it produces with chromo-dipole operators should be validated. Note also that G runs and that its value at the scale MZ depends on the parton distribution function employed. See for the hack, to be used with great care.
  • 2018-08-23: added back DIM6 coupling orders to the dim6top_LO_UFO_each_coupling_order version (thanks to Jay Howarth)
  • 2018-06-06: corrected the Lorentz structures of the operators relative to cbtud1 and cbtud8 coefficients from scalar to vector (thanks to Céline Degrande)
  • 2018-05-28: moved the definition of 'ctA' below that of 'sw' and 'cw' in, to avoid "MadGraph5Error : Unable to evaluate mdl_ctA = (mdl_ctW-mdl_cw*mdl_ctZ)/mdl_sw: raise error: name 'mdl_cw' is not defined" (thanks to Marcel Vos)
  • 2018-05-21: corrected the Lorentz structure of Oledq operator from tensor to scalar (thanks to Ken Mimasu)

Note on NLO QCD

For NLO QCD simulation, the model authors recommend using the SMEFT@NLO model which is mostly compatible with the baseline prescription of the LHC TOP WG. Few differences between the dim6top and SMEFT@NLO (August 2019 version) models are to be noted:

  • SMEFT@NLO includes Higgs and electroweak operators but does not include top-quark FCNCs.
  • SMEFT@NLO is moreover restricted to CP-conserving operator coefficients.
  • SMEFT@NLO imposes an enhanced U(2)q×U(2)u×U(3)d flavour symmetry in the quark sector, such that the right-handed bottom quark is treated on the same footing as other light down-type quarks. This is required for consistency with the five-flavour scheme at NLO QCD, where MB=0.
  • SMEFT@NLO normalizes the top dipole operator (with coefficient ctG) with a factor of the strong coupling constant G for compatibility with MG 2.x series at NLO in QCD. The 3.x series does not require it. This implies that ctG_dim6top = G * ctG_SMEFTatNLO. Note that G runs.

Altogether, due to the enhanced quark flavour symmetry and to CP-conservation, SMEFT@NLO only includes a subset of 45 top-quark operator coefficients:
cQQ1, cQQ8, cQd1, cQd8, cQe1, cQe2, cQe3, cQl31, cQl32, cQl33, cQlM1, cQlM2, cQlM3, cQq11, cQq13, cQq81, cQq83, cQt1, cQt8, cQu1, cQu8, cblS3, cdp, cpQ3, cpQM, cpt, ctG, ctW, ctZ, ctd1, ctd8, cte1, cte2, cte3, ctl1, ctl2, ctl3, ctlS3, ctlT3, ctp, ctq1, ctq8, ctt1, ctu1, ctu8
and equates cQb1=cQd1, cQb8=cQd8, ctb1=ctd1, ctb8=ctd8. The chirality flipping and charged right-handed currents involving a bottom quark are in particular forbidden.
(Altogether, among the non-FCNC couplings, SMEFT@NLO does not include: ctpI, cpb, cptb, cptbI, ctWI, ctZI, cbW, cbWI, ctGI, ctlS1, ctlSI1, ctlS2, ctlSI2, ctlSI3, ctlT1, ctlTI1, ctlT2, ctlTI2, ctlTI3, cblS1, cblSI1, cblS2, cblSI2, cblSI3, ctQqu1, ctQqu1I, ctQqu8, ctQqu8I, cbQqd1, cbQqd1I, cbQqd8, cbQqd8I, cQtqd1, cQtqd1I, cQtqd8, cQtqd8I, cQbqu1, cQbqu1I, cQbqu8, cQbqu8I, ctQqu1T, ctQqu1TI, ctQqu8T, ctQqu8TI, cbQqd1T, cbQqd1TI, cbQqd8T, cbQqd8TI, cQtqd1T, cQtqd1TI, cQtqd8T, cQtqd8TI, cQbqu1T, cQbqu1TI, cQbqu8T, cQbqu8TI, cbtud1, cbtud1I, cbtud8, cbtud8I, cQtQb1, cQtQb8, cQtQb1I, cQtQb8I.)

A high-precision comparison with dim6top has been performed at leading order following the prescription of 1906.12310, with the plugin_eft_contrib of MG. More details are found in the arxiv note and in the example directory of the plugin.

The SMEFT@NLO version dated from August 2019 does not include an implementation of four-fermion operators at NLO in QCD. Its validation is ongoing.

Last modified 4 months ago Last modified on 05/19/20 19:29:16

Attachments (5)

Download all attachments as: .zip