Triple Gluon Operator in SMEFT

• Triple Gluon Operator is a CP-even, dimension-six SMEFT entity constructed purely from gluon fields, redefining gluon self-interactions.
• It modifies the momentum structure of three- and four-gluon vertices, affecting observables in Higgs, top, and jet production through interference patterns.
• Beyond collider signatures, the operator is pivotal in two-loop QCD renormalization, influencing running, mixing, and normalization conventions in SMEFT.

The triple gluon operator is the dimension-six pure-gluon operator of the Standard Model Effective Field Theory (SMEFT) built from three gluon field strengths and the $SU(3)$ structure constants. In common notation it is written as

$$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu},$$

or equivalently

$$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$

up to index placement and dummy-index relabeling. Multiple papers describe it as the unique CP-even, dimension-six operator involving only gluon fields, although the normalization convention varies: some definitions include an explicit factor of $g_s$, while others absorb that factor into the field strength and Wilson coefficient conventions (Ghosh et al., 2014).

1. Definition and operator conventions

In the Higgs-plus-jet analysis of Grojean et al., the operator is introduced through

$$L_6=\frac{C_g}{\Lambda^2}O_g+\frac{C_{3g}}{\Lambda^2}O_{3g}+\left(\frac{C_t}{\Lambda^2}O_t+\rm h.c.\right)+\left(\frac{C_b}{\Lambda^2}O_b+\rm h.c.\right),$$

with

$$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu}.$$

That paper states explicitly that no extra explicit factor of $g_s$ is inserted into the operator definition and that the sign convention is

$+\frac{C_{3g}}{\Lambda^2}O_{3g}$

in the Lagrangian (Ghosh et al., 2014).

Later SMEFT renormalization papers use closely related notation. In the $\overline{\text{MS}}$ analysis,

$$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$

with

$$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu},$$0

and the operator appears as

$$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu},$$1

The same work emphasizes that $$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu},$$2 is normalized without an explicit factor of $$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu},$$3, and notes that this convention affects loop counting and translations to SMEFT@NLO conventions (Duhr et al., 3 Mar 2025).

By contrast, collider reinterpretation papers on dijets and multijets often define

$$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu},$$4

again inserted as

$$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu},$$5

These are convention changes, not different physical operators. This suggests that care is required when comparing Wilson coefficients across papers, since the presence or absence of an explicit $$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu},$$6 changes the quoted numerical normalization (Goldouzian et al., 2020).

2. Structural role in the SMEFT

Several papers characterize the triple gluon operator as the only CP-even, genuinely gluonic operator at dimension six. In the on-shell renormalization study it is described as “the unique CP-even operator of dimension six that only involves gluon fields,” and the analysis is restricted to the CP-conserving sector and to terms linear in dimension-six coefficients (Duhr et al., 6 Aug 2025).

Its physical effect is not to rescale a single Higgs or top coupling. Instead, as emphasized in the Higgs-plus-jet study, it “generates three and four gluon vertices with a modified momentum dependence and additional vertices with up to six gluons.” This is the key distinction from operators such as $$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu},$$7, $$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu},$$8, and $$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu},$$9: the triple gluon operator alters the self-interaction structure of QCD itself, and collider sensitivity arises indirectly through modified gluonic subamplitudes rather than through a direct Higgs insertion (Ghosh et al., 2014).

This role recurs across applications. In top-pair production it modifies gluon self-interactions, in particular the three-gluon vertex, and the paper on boosted tops notes that the quartic gluon coupling is also affected (Bardhan et al., 2020). In Higgs production from anomalous gluon dynamics, the operator

$$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$0

does not generate a tree-level $$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$1 interaction by itself; instead its leading effect in $$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$2 first arises at two loops and is inseparable from operator mixing into $$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$3 and $$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$4 (Haisch, 8 Mar 2025).

A recurring misconception is that a pure-gluon operator can be ignored in Higgs and top studies because it contains no Higgs or quark fields. The literature summarized here argues the opposite: the operator propagates into Higgs and top observables through modified gluonic interactions, and in some analyses it correlates with more familiar Higgs- and top-sector directions (Ghosh et al., 2014).

3. Phenomenology in Higgs, top, and jet observables

In Higgs-plus-jet production, the operator contributes to gluon fusion processes such as

$$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$5

by modifying the momentum dependence of the three-gluon vertex appearing in the loop-induced topology. The differential cross section is treated linearly in the Wilson coefficients,

$$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$6

so only the interference with the Standard Model is retained. The central phenomenological result is that the $$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$7 distribution due to $$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$8 is “a linear combination of the distributions due to $$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$9 and $g_s$0,” implying that transverse-momentum spectra alone do not disentangle these effects. By contrast, the jet rapidity $g_s$1 and rapidity difference $g_s$2 do separate them: the paper states that the ratio curve for $g_s$3 in $g_s$4 is completely flat while that for $g_s$5 is peaked around $g_s$6 (Ghosh et al., 2014).

In top-pair production, the phenomenology is controlled by helicity structure and by the fact that the interference in

$g_s$7

is proportional to $g_s$8. The boosted-top study therefore identifies $g_s$9 as the natural quark channel and finds strongest sensitivity in the high-$$L_6=\frac{C_g}{\Lambda^2}O_g+\frac{C_{3g}}{\Lambda^2}O_{3g}+\left(\frac{C_t}{\Lambda^2}O_t+\rm h.c.\right)+\left(\frac{C_b}{\Lambda^2}O_b+\rm h.c.\right),$$0 and high-$$L_6=\frac{C_g}{\Lambda^2}O_g+\frac{C_{3g}}{\Lambda^2}O_{3g}+\left(\frac{C_t}{\Lambda^2}O_t+\rm h.c.\right)+\left(\frac{C_b}{\Lambda^2}O_b+\rm h.c.\right),$$1 tails. In its single-operator analysis, the strongest observed 95% C.L. bound comes from the $$L_6=\frac{C_g}{\Lambda^2}O_g+\frac{C_{3g}}{\Lambda^2}O_{3g}+\left(\frac{C_t}{\Lambda^2}O_t+\rm h.c.\right)+\left(\frac{C_b}{\Lambda^2}O_b+\rm h.c.\right),$$2 tail of $$L_6=\frac{C_g}{\Lambda^2}O_g+\frac{C_{3g}}{\Lambda^2}O_{3g}+\left(\frac{C_t}{\Lambda^2}O_t+\rm h.c.\right)+\left(\frac{C_b}{\Lambda^2}O_b+\rm h.c.\right),$$3 up to 1 jet,

$$L_6=\frac{C_g}{\Lambda^2}O_g+\frac{C_{3g}}{\Lambda^2}O_{3g}+\left(\frac{C_t}{\Lambda^2}O_t+\rm h.c.\right)+\left(\frac{C_b}{\Lambda^2}O_b+\rm h.c.\right),$$4

with the paper highlighting $$L_6=\frac{C_g}{\Lambda^2}O_g+\frac{C_{3g}}{\Lambda^2}O_{3g}+\left(\frac{C_t}{\Lambda^2}O_t+\rm h.c.\right)+\left(\frac{C_b}{\Lambda^2}O_b+\rm h.c.\right),$$5 as the headline result (Bardhan et al., 2020).

For inclusive jets, the most important subtlety is that the Standard Model and $$L_6=\frac{C_g}{\Lambda^2}O_g+\frac{C_{3g}}{\Lambda^2}O_{3g}+\left(\frac{C_t}{\Lambda^2}O_t+\rm h.c.\right)+\left(\frac{C_b}{\Lambda^2}O_b+\rm h.c.\right),$$6 amplitudes in $$L_6=\frac{C_g}{\Lambda^2}O_g+\frac{C_{3g}}{\Lambda^2}O_{3g}+\left(\frac{C_t}{\Lambda^2}O_t+\rm h.c.\right)+\left(\frac{C_b}{\Lambda^2}O_b+\rm h.c.\right),$$7 dijet channels are helicity-orthogonal, so there is effectively no interference at $$L_6=\frac{C_g}{\Lambda^2}O_g+\frac{C_{3g}}{\Lambda^2}O_{3g}+\left(\frac{C_t}{\Lambda^2}O_t+\rm h.c.\right)+\left(\frac{C_b}{\Lambda^2}O_b+\rm h.c.\right),$$8. The leading observable effect therefore appears at $$L_6=\frac{C_g}{\Lambda^2}O_g+\frac{C_{3g}}{\Lambda^2}O_{3g}+\left(\frac{C_t}{\Lambda^2}O_t+\rm h.c.\right)+\left(\frac{C_b}{\Lambda^2}O_b+\rm h.c.\right),$$9 through the square of the dimension-six amplitude. This underlies both the reappraisal of anomalous gluon self-interactions and the dedicated reinterpretation of CMS dijet angular distributions (Hirschi et al., 2018).

The dijet-angular analysis uses

$$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu}.$$0

and shows that the $$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu}.$$1 contribution peaks at small $$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu}.$$2, reflecting more central scattering without the Standard Model $$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu}.$$3-channel pole structure. Reinterpreting CMS $$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu}.$$4 data with $$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu}.$$5, the paper obtains

$$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu}.$$6

observed at 95% C.L., with expected limit

$$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu}.$$7

and concludes that this bound lies far below the sensitivity obtainable from top-quark and Higgs measurements (Goldouzian et al., 2020).

4. Matching, mixing, and two-loop renormalization

The renormalization literature elevates the triple gluon operator from a phenomenological nuisance parameter to a genuine two-loop ingredient of SMEFT QCD. In the $$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu}.$$8 study, the calculation is performed off shell in the background-field method, and $$O_{3g}=f^{abc}G^{a\mu}_{\ \ \nu}G^{b\nu}_{\ \ \rho}G^{c\rho}_{\ \ \mu}.$$9 contributes for the first time at two loops to the top-field renormalization, top-mass renormalization, gluon-field renormalization, and strong-coupling renormalization. The paper gives

$g_s$0

$g_s$1

$g_s$2

$g_s$3

and

$g_s$4

It also extracts an $g_s$5 contribution to the QCD $g_s$6-function and emphasizes that consistent two-loop renormalization requires EOM-vanishing and BRST-exact operators in addition to the physical basis (Duhr et al., 3 Mar 2025).

The on-shell companion paper reaches complementary conclusions. There the operator contributes only at two loops to the top mass renormalization $g_s$7, top field renormalization $g_s$8, and gluon field renormalization $g_s$9, while for massless quarks on shell

$+\frac{C_{3g}}{\Lambda^2}O_{3g}$0

The explicit gluon-field result is

$+\frac{C_{3g}}{\Lambda^2}O_{3g}$1

The paper further states that the $+\frac{C_{3g}}{\Lambda^2}O_{3g}$2 contributions to the top field and mass renormalization constants are gauge independent and describes these results as “essential ingredients for any two-loop computation in the QCD sector of the SMEFT” (Duhr et al., 6 Aug 2025).

The two-loop Higgs-production analysis provides a collider realization of this renormalization structure. Starting from

$+\frac{C_{3g}}{\Lambda^2}O_{3g}$3

it shows that a consistent $+\frac{C_{3g}}{\Lambda^2}O_{3g}$4-induced prediction for $+\frac{C_{3g}}{\Lambda^2}O_{3g}$5 requires three ingredients at the same parametric order: direct two-loop matching from $+\frac{C_{3g}}{\Lambda^2}O_{3g}$6, one-loop matching from $+\frac{C_{3g}}{\Lambda^2}O_{3g}$7 after one-loop running, and tree-level matching from $+\frac{C_{3g}}{\Lambda^2}O_{3g}$8 after two-step running. A genuinely new result of that work is the two-loop mixing

$+\frac{C_{3g}}{\Lambda^2}O_{3g}$9

The paper stresses that scale independence at a given order requires combining matching and running contributions computed at different loop orders (Haisch, 8 Mar 2025).

5. Alternative meanings and nearby concepts

The phrase “triple gluon” has several distinct meanings in the literature, and the local CP-even SMEFT operator should be distinguished from them.

First, it is distinct from the standard QCD three-gluon vertex. The worldline study of the one-loop three-gluon vertex analyzes the 1PI Green function

$\overline{\text{MS}}$0

and matches its low-energy expansion onto local gauge-invariant operators such as

$\overline{\text{MS}}$1

This is closely related conceptually, but the subject there is the dressed Yang–Mills vertex rather than the SMEFT Wilson coefficient itself (Ahmadiniaz et al., 2012).

Second, it is distinct from the CP-odd three-gluon operator, usually called the Weinberg operator. The nonperturbative-renormalization paper defines

$\overline{\text{MS}}$2

describes it as the gluon chromo-electric dipole moment, and constructs a regularization-independent momentum-subtraction scheme for its lattice renormalization. That operator is $\overline{\text{MS}}$3-odd and $\overline{\text{MS}}$4-odd, unlike the CP-even SMEFT triple gluon operator considered above (Cirigliano et al., 2020).

Third, it is distinct from twist-3 multi-gluon correlation functions in hadron structure. In the SSA study for

$\overline{\text{MS}}$5

the relevant “triple-gluon correlation functions” are nonlocal light-cone matrix elements of three field strengths, resolved into symmetric and antisymmetric color structures,

$\overline{\text{MS}}$6

with the observable depending separately on $\overline{\text{MS}}$7, $\overline{\text{MS}}$8, $\overline{\text{MS}}$9, and $$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$0 (Koike et al., 2010).

Fourth, it is distinct from three-soft-gluon emission operators in soft factorization. The soft-radiation paper derives the nonlocal color-space current

$$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$1

where $$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$2 is the irreducible maximally non-abelian part of the triple-soft current. This is a soft-emission operator, not a local field operator in the Lagrangian (Colferai, 2022).

These distinctions matter because the phrase “triple gluon operator” is sometimes used loosely. In strict SMEFT usage, it denotes the local CP-even dimension-six pure-gluon operator $$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$3; other “triple-gluon” objects belong to different operator classifications and different factorization regimes.

6. Status and interpretation

Across collider studies, the operator has moved from a neglected pure-gauge direction to an explicitly constrained component of practical SMEFT analyses. In 2014 it was argued that a full Higgs-plus-jet EFT interpretation must include $$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$4, because its allowed distortions in kinematic distributions can be similar in size to those from the effective Higgs-gluon operator when each coefficient is taken near expected bounds (Ghosh et al., 2014). Subsequent jet-based analyses then argued that dijet and multijet observables constrain the operator strongly enough that, for practical global fits of top and Higgs data, its impact can often be neglected after an independent bound is imposed (Goldouzian et al., 2020).

That practical conclusion is sharpened by the multijet reappraisal, which supports the robustness of multijet limits even though they are largely driven by $$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$5 terms rather than linear interference. The same paper nevertheless emphasizes that specially designed three-jet angular observables remain theoretically attractive because they are dominated by $$\mathcal O_G=f^{abc}G_{\mu}^{a,\nu}G_{\nu}^{b,\rho}G_{\rho}^{c,\mu},$$6 interference (Hirschi et al., 2018).

At the same time, the renormalization literature shows that the operator cannot be reduced to a simple collider template. It affects wave-function renormalization, mass renormalization, strong-coupling running, and operator mixing at two loops, and therefore enters any RG-improved SMEFT calculation in the QCD sector beyond leading order (Duhr et al., 3 Mar 2025).

A balanced synthesis is therefore the following. The triple gluon operator is a uniquely defined CP-even pure-gluon dimension-six SMEFT deformation, but its normalization is convention-dependent. Phenomenologically, it modifies gluon self-interactions rather than a single Standard Model coupling, so its effects are process- and observable-dependent. In Higgs and top studies it can induce nontrivial correlations; in jet observables it is often constrained most strongly; and in higher-order SMEFT it is an essential ingredient of two-loop renormalization and mixing. This suggests that its modern significance lies not in any single signature, but in the fact that it links collider phenomenology, SMEFT basis conventions, and higher-order QCD renormalization into a single operator framework (Haisch, 8 Mar 2025).