betaFeedback

WorksPhysics and Astronomy

The impact of transverse Slavnov-Taylor identities on dynamical chiral symmetry breaking

Abstract

We extend earlier studies of transverse Ward-Fradkin-Green-Takahashi identities in QED, their usefulness to constrain the transverse fermion-boson vertex and their importance for multiplicative renormalizability, to the equivalent gauge identities in QCD. To this end, we consider transverse Slavnov-Taylor identities that constrain the transverse quark-gluon vertex and derive its eight associated scalar form factors. The complete vertex can be expressed in terms of the quark's mass and wave-renormalization functions, the ghost-dressing function, the quark-ghost scattering amplitude and a set of eight form factors. The latter parametrize the hitherto unknown nonlocal tensor structure in the transverse Slavnov-Taylor identity which arises from the Fourier transform of a four-point function involving a Wilson line in coordinate space. We determine the functional form of these eight form factors with the constraints provided by the Bashir-Bermudez vertex and study the effects of this novel vertex on the quark in the Dyson-Schwinger equation using lattice QCD input for the gluon and ghost propagators. We observe significant dynamical chiral symmetry breaking and a mass gap that leads to a constituent mass of the order of 500 MeV for the light quarks. The flavor dependence of the mass and wave-renormalization functions as well as their analytic behavior on the complex momentum plane is studied and as an application we calculate the quark condensate and the pion's weak decay constant in the chiral limit. Both are in very good agreement with their reference values.

Prepared for submission to JHEP

Introduction

Whilst the Brout-Englert-Higgs (BEH) mechanism has been established as the essential explicit source of elementary particles’ masses, the same cannot be said of Nature’s composite building blocks, namely the atoms and their nuclei. Even the lightest Nambu-Goldstone mode of Quantum Chromodynamics (QCD), the pion, is more than an order of magnitude heavier than the sum of the current masses of its constituents provided by the BEH mechanism. QCD is not a conformal theory and this mass relation is only sensible at a certain energy scale. Commonly, the light quark’s current masses are quoted at 2 GeV in the MS scheme [1], at which the sum of two uu and one dd current-quark current masses amounts to merely 1% of the proton mass.

The overwhelming contribution to the light hadron’s masses does not stem from the aforementioned mechanism and a recent lattice-QCD simulation [2] concludes that the kinetic quark energy and the gluon field are together responsible for 68% of the proton mass, whereas the trace anomaly contributes 23%. Thus, only 9% of the proton’s mass is due to the scalar condensate, which exhibits the major current-quark mass dependence. This recent analysis of the proton’s “mass budget” has come to strengthen long-standing observations made by groups who apply functional continuum methods to QCD. In particular, starting from light current quarks with mu,d≈3m_{u,d} \approx3–44 MeV at a reasonable perturbative scale, solving the Dyson-Schwinger equation (DSE) for the quark propagator in QCD [3, 4] and making use of the three-body Faddeev equation yields the proton’s mass and that of the Roper, the nucleon’s parity partners and the Δ\Delta baryons in a consistent symmetry-preserving truncation [5, 6, 7, 8, 9, 10, 11, 12, 13, 14]. These results not only preserve the correct mass ordering but are also remarkably accurate within the uncertainties of the meson cloud effect. In these approaches one can also compute the nucleon’s σ\sigma term, which is a measure of its current-quark mass dependence [15], and it turns out that σN≈50\sigma_N \approx50–60 MeV60\ \mathrm{MeV}. The evidence is conclusive that the vast majority of the nucleon’s mass is due to dynamical chiral symmetry breaking (DCSB), which is in contrast to explicit chiral symmetry breaking whose origin lies solely in the BEH mechanism and is described by the quarks’ mass terms in the QCD Lagrangian.

The Faddeev bound-state calculations are limited to the leading approximation of the gap equation and three-body interaction kernel, with the exception of a certain degree of beyond-ladder truncations effects modeled in the diquark amplitudes and propagators. In this simplification, referred to as rainbow-ladder truncation, the quark-gluon vertex is simply described by its bare form defined in the Lagrangian coupling. This is a drastic simplification, as the one-loop dressed vertex already leads to non-vanishing coefficients for all 12 independent tensor structures [16, 17]. In fact, DCSB is manifest not only in the quark propagator but also in the quark-gluon vertex, as 6 of the 12 structures are only generated dynamically and their feed-back into the gap equation enhances the generation of quark masses. If one solves the gap equation with merely the bare vertex, a realistic strong coupling and a gluon propagator obtained with the appropriate DSE or lattice-QCD simulations [18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41], the resulting DCSB is too small and practically irrelevant to realistic constituent-quark and hadron masses. Therefore, in using a model for the gluon dressing in the rainbow-ladder truncation, a scale parameter is adjusted to the experimental pion and kaon masses.

This procedure has proven to be very successful for ground and first excited states of mesons and quarkonia in the pseudoscalar and vector channels. However, higher excited and exotic states, as well as the scalar and axialvector qˉq\bar{q}q states, are not well described in this leading truncation [5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58]. A particularly problematic case is the highly asymmetric momentum distribution in flavored mesons, such as the DD and BB mesons, as the quark-gluon vertex dressing has a different impact on a light quark than on charm and bottom quarks. More precisely, while a bare vertex can reliably be used for the heavy-quark interaction in the Bethe-Salpeter equation (BSE), this is not the case for light quarks [59, 60, 61, 62, 63, 64, 65]. Effects of a dressed quark-gluon vertex in heavy-light mesons have also been addressed in Refs. [66, 67, 68].

There are not only phenomenological motivations for improving our knowledge about the tensor structure and analytic behavior of the quark-gluon vertex in the nonperturbative domain. On pure field-theoretical grounds the behavior of the fermion-boson vertex is of great interest, as i) is a crucial object to explain DCSB in QED (with an artificially scaled-up coupling) and in QCD, and b) plays an eminent role in its contribution to the infrared behavior of Green functions as well as to quark and gluon fragmentation functions, and therefore to the elucidation of the confinement mechanism. Starting from the perturbative limit, truncation models for the vertex are commonly constructed that retain the essential features of the underlying theory, which are known to be respected at every order of a perturbative formulation by construction. It is a natural requirement to achieve a satisfactory determination of relevant physical observables. For small values of the coupling constant, perturbation theory is the paradigmatic example of such a truncation scheme, yet inadequate to calculate hadronic bound states. Much progress towards a more detailed understanding of the fermion-boson vertex in QED and QCD has been made in the past years, either by direct calculation of the DSE for the vertex [69, 70, 71, 72, 73], invoking gauge identities and multiplicative renormalizability [74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97], in perturbative approaches [98, 99, 100, 101] or by direct numerical sampling of QCD on the lattice [102, 103, 104, 105, 106, 107].

In this work we extend earlier studies [85] which are based on gauge covariance and combine the constraints of the well known Ward-Fradkin-Green-Takahashi (WFGTI) identity [108, 109, 110, 111] and two additional transverse Takahashi identities (TTI) [112, 113, 114, 115, 116]. The WFGTI has long been known as an expression of gauge symmetry and current conservation and allows for an expansion of the photon-fermion vertex in terms of a well-constrained “longitudinal” part1 [16] and undetermined transverse components. The symmetry which leads to the transverse identities is the Lorentz transformation acting on the usual infinitesimal gauge transformation. Indeed, while the WFGTI relates the divergence of the fermion-photon vertex to the inverse fermion propagator, the TTI expresses the curl of this vertex. Though the TTI were verified to one-loop order [117, 118], the complexity of these additional structures made these identities for the longest time not amicable to a straightforward determination of the transverse part of the photon-fermion vertex. Moreover, the TTI couple the vector and axialvector vertices, though, as demonstrated in Ref. [88], the uncoupling can be achieved by judicious tensor projections and leaves one with two identities that involve only the vector vertex.2 Nonetheless, besides the inverse fermion propagator and the vector vertex, these uncoupled TTI still involve a nontrivial tensor structure that arises from the Fourier transform of a four-point like function in coordinate-space with a necessary Wilson line.

In Ref. [85] we demonstrated that this tensor structure can be parametrized and is constrained by multiplicative renormalizability. In that capacity, the TTI are intimately connected to another consequence of local gauge covariance, namely the Landau-Khalatnikov-Fradkin transformations (LKFT) [109, 119] which describe the response of the Green functions to an arbitrary gauge transformation and express multiplicative renormalizability of the massless fermion propagator in 4 space-time dimensions. This implies that not any functional form of the tensor structures in the transverse vertex is possible and we showed that, for a given form of the transverse vertex that satisfies the TTI, the critical QED coupling above which chiral symmetry is dynamically broken is gauge invariant.

We here build upon these results and explore them in the context of QCD. As it is well known, color-gauge invariance in QCD is preserved by the Slavnov-Taylor identity (STI) for the quark-gluon vertex [120, 121] which also leaves the transverse vertex undetermined. The TTI were generalized to two transverse STIs (TSTI) [122] from which the vector vertex can be extracted that involves, as the usual STI, the inverse quark propagators, the ghost-dressing function and the quark-ghost scattering amplitude, but furthermore a nonlocal four-point like function which is a consequence of gauge invariance. As in QED, this latter term can be parametrized most generally by four tensor structures and corresponding form factors. Similarly, the quark-ghost scattering kernel can be described by four matrix-valued amplitudes which can be computed within a nonperturbative dressed-propagator approach [86, 90, 92, 93, 94]. Note that, for the very first time, we now also have generalized LKFT (GLKFT) for QCD [82, 123, 124], though our understanding of these is in its infancy and their complexity still prevents us from imposing tangible constraints on the quark-gluon vertex.

In analogy with the approach taken in Refs. [85, 88], we divide the quark-gluon vertex into longitudinal and transverse tensor structures using the tensor basis of Ref. [125]. Solving a system of coupled equations we obtain the expressions of the longitudinal form factors λi(k,p)\lambda_i(k,p), i=1,…,4i=1,\ldots,4, from the usual STI as in Refs. [86, 90, 92, 93, 94], and additionally derive the functional form of the transverse form factors τj(k,p)\tau_j(k,p), j=1,…,8j=1,\ldots,8, where kk and pp are the outgoing and incoming quark momenta, respectively. Six of the twelve vertex form factors depend on the quark-ghost interaction kernel and on the ghost-dressing function. In addition, the transverse form factors are characterized by a functional dependence on eight scalar functions, Yi(k,p)Y_i(k,p), which parametrize the aforementioned nonlocal tensor structure in the uncoupled TSTI for the vector vertex. As we are currently in no condition to calculate the corresponding four-point function in momentum space, we rely on the well established Bashir-Bermudez ansatz [81] for the fermion-gauge boson vertex that preserves multiplicative renormalizability and is constrained by gauge covariance and perturbative QCD in a given kinematic limit. Thus, we trade our ignorance of the Yi(k,p)Y_i(k,p) functions for eight parameters aia_i employed in this ansatz and solve the quark DSE with the complete vertex structure for a large representative sample of aia_i-sets. The latter are constrained by multiplicative renormalizability in the range, −2≤ai≤+2-2 \le a_i \le+2, and we employ gluon and ghost propagators from lattice QCD.

We find that only a very limited combination leads to a mass function M(p)M(p) that exhibits the DCSB and functional behavior observed in phenomenological models, and the same holds for the wave renormalization function Z(p)Z(p). In many cases, uninteresting solutions are found, i.e. M(0)≲100M(0) \lesssim100 MeV, or the iteration process to solve the integral equations converges poorly or not at all. We here present the first solution of the quark’s DSE that leads to significant DCSB with M(0)≈500M(0) \approx500 MeV, employing αs(μ)≃0.3\alpha_s(\mu) \simeq0.3, a gluon-dressing function, Δ(q2,μ)\Delta(q^2,\mu), and a ghost-dressing function, G(q2,μ2)G(q^2,\mu^2), renormalized at μ=4.3\mu= 4.3 GeV in agreement with lattice QCD. No additional strength via a form factor or other modifications in the DSE kernel are introduced and even in the chiral limit the DCSB is still considerable.

This paper is organized as follows: in Section 2 we review the DSE for a quark in QCD and motivate our renormalization procedure, after which we go into the details of the construction of the fully dressed quark-gluon vertex and its most general tensor structure. We then introduce the longitudinal and transverse STIs that constrain the quark-gluon vertex, discuss their content, in particular the quark-ghost scattering kernel, and show how their manipulations with appropriate tensor projections leads to the analytic form of the 12 associated form factors. Since one nonlocal tensor structure in the transverse STIs is hitherto unknown, at least nonperturbatively, we introduce its parametrization and constrain it with the Bashir-Bermudez ansatz in [81] and multiplicative renormalizability. In § 3 we turn our attention to the gauge sector and list the parametrizations of the gluon and ghost propagators which are fitted to the data from three lattice QCD groups. In § 4 we present our main results which consist of the mass and wave-renormalization functions and the leading quark-ghost form factor in case of omitting the transverse vertex and using the full vertex in the DSE. Strong coupling variations and quark-flavor dependence of the DSE solutions are also studied. We wrap up that section with the solutions of the DSE on the complex plane for light quarks, followed by some applications of our results, namely the calculation of the pion’s weak decay constant and the quark condensate in the chiral limit, in § 4.3. Finally, in § 5 we comment on our results and propose future steps that could lead to a parameter-free determination of all transverse vertex functions.

The gap equation and gauge-symmetry constraints on the quark-gluon vertex

Dyson-Schwinger equation for a quark

The most prominent occurrence of the quark-gluon vertex is in the DSE, which is nothing else than the relativistic equation of motion of the quark in QCD formulated in a nonperturbative manner. In essence, the DSE describes the nonperturbative gluon dressing of the current quark by a self-energy term in its propagator. For a given flavor the DSE of the inverse quark propagator is,3

S−1(p)=Z2(iγ⋅p+mbm)+Σ(p2)=Z2iγ⋅p+Z4m+Z1g2∫Λd4k(2π)4Δμνab(q)γμtaS(k)Γνb(k,p) ,(1)\begin{aligned} S^{-1}(p) = Z_2\left(i\gamma\cdot p + m^{\mathrm{bm}}\right) + \Sigma(p^2) \\ &= Z_2 i\gamma\cdot p + Z_4 m + Z_1 g^2 \int^\Lambda\frac{\mathrm{d}^4 k}{(2\pi)^4} \Delta_{\mu\nu}^{ab}(q) \gamma_\mu t^a S(k) \Gamma_\nu^b(k,p) \, , \tag*{(1)} \end{aligned}

where mbmm^{\mathrm{bm}} is the bare quark mass, mm is the renormalized or current quark mass and Z1(μ,Λ)Z_1(\mu,\Lambda) and Z2(μ,Λ)Z_2(\mu,\Lambda) are the vertex and wave-function renormalization constants, respectively. The first two terms in Eq. (1) are the inverse free quark propagator and the integral expresses the quark’s self-energy Σ(p2)\Sigma(p^2). In this integral, Δμν(q)\Delta_{\mu\nu}(q) is the dressed-gluon propagator in Landau gauge with momentum q=k−pq = k - p,

Δμνab(q)=δab(δμν−qμqνq2)Δ(q2) ,(2)\Delta_{\mu\nu}^{ab}(q) = \delta^{ab}\left(\delta_{\mu\nu} - \frac{q_\mu q_\nu}{q^2}\right)\Delta(q^2) \, , \tag*{(2)}

where Δ(q2)\Delta(q^2) is the nonperturbative dressing function, Δ(q2)→1/q2\Delta(q^2) \to1/q^2 for large q2q^2, Γμa(k,p)=Γμ(k,p)ta\Gamma_\mu^a(k,p) = \Gamma_\mu(k,p)t^a, is the dressed quark-gluon vertex and ta=λa/2t^a = \lambda^a/2 are the SU(3) group generators with λa\lambda^a in the fundamental representation; a,ba,b generally represent color indices.

The most general Poincaré-covariant form of the solutions to Eq. (1) is written in terms of covariant scalar and vector amplitudes:

S(p)=−iγ⋅p σv(p2)+σs(p2)=1iγ⋅p A(p2)+B(p2)=Z(p2)iγ⋅p+M(p2).(3)S(p) = -i\gamma\cdot p\,\sigma_{v}(p^{2}) + \sigma_{s}(p^{2}) = \frac{1}{i\gamma\cdot p\,A(p^{2}) + B(p^{2})} = \frac{Z(p^{2})}{i\gamma\cdot p + M(p^{2})}. \tag*{(3)}

In the integral, Λ\Lambda is an ultraviolet cut-off, μ\mu is the renormalization scale and one typically chooses Λ≫μ\Lambda\gg\mu in DSE studies of QCD. These scales are implicit in our notation, namely A(p2)≡A(p2,μ2,Λ2)A(p^{2}) \equiv A(p^{2},\mu^{2},\Lambda^{2}) and B(p2)≡B(p2,μ2,Λ2)B(p^{2}) \equiv B(p^{2},\mu^{2},\Lambda^{2}), as is a flavor index ff for these quantities as well as for all renormalization constants. The flavor-dependent nonperturbative mass and wave renormalization functions are,

M(p2)=B(p2,μ2,Λ2)/A(p2,μ2,Λ2),(4)M(p^{2}) = B(p^{2},\mu^{2},\Lambda^{2})/A(p^{2},\mu^{2},\Lambda^{2}), \tag*{(4)}

and,

Z(p2,Λ2,μ2)=1/A(p2,Λ2,μ2),(5)Z(p^{2},\Lambda^{2},\mu^{2}) = 1/A(p^{2},\Lambda^{2},\mu^{2}), \tag*{(5)}

respectively. The scale μ\mu is commonly chosen such that the dressed functions match the ones in perturbation theory, i.e. Z(μ2)=1Z(\mu^{2}) = 1 and M(μ2)=m=Z2(μ,Λ)/Z4(μ,Λ) mbm(Λ)M(\mu^{2}) = m = Z_{2}(\mu,\Lambda)/Z_{4}(\mu,\Lambda)\,m^{\mathrm{bm}}(\Lambda), where Z4(μ,Λ)Z_{4}(\mu,\Lambda) is the renormalization constant associated with the Lagrangian’s mass term.

The object of our interest is the fully dressed vertex, Γμ(k,p)\Gamma_{\mu}(k,p), which satisfies its own DSE. As already mentioned, we do not intend to solve the DSE of the vertex in a given truncation scheme but rather make use of three STIs presented in § 2.2 and § 2.3. This allows us to derive the scalar form factors associated with the tensor structure of the vertex. Before we engage on this path, a note on the renormalization method is in order. After taking the color trace and with CF=4/3C_{F} = 4/3 in the fundamental representation, the unrenormalized DSE for a given quark reads:

S0−1(p)=iγ⋅p+mbm+g02CF∫d4k(2π)4 Δμν0(q)γμS0(k)Γν0(k,p).(6)S_{0}^{-1}(p) = i\gamma\cdot p + m^{\mathrm{bm}} + g_{0}^{2}C_{F}\int\frac{d^{4}k}{(2\pi)^{4}}\,\Delta_{\mu\nu}^{0}(q)\gamma_{\mu}S_{0}(k)\Gamma_{\nu}^{0}(k,p). \tag*{(6)}

Relating bare propagators, the coupling and vertex to their renormalized expressions via the following procedure,

S(p,μ2)=Z2−1(μ2)S0(p),(7)S(p,\mu^{2}) = Z_{2}^{-1}(\mu^{2})S_{0}(p), \tag*{(7)}
Δ(q2,μ2)=Z3−1(μ2)Δ0(q2),(8)\Delta(q^{2},\mu^{2}) = Z_{3}^{-1}(\mu^{2})\Delta_{0}(q^{2}), \tag*{(8)}
Γμ(k,p,μ2)=Z1(μ2)Γμ0(k,p),(9)\Gamma_{\mu}(k,p,\mu^{2}) = Z_{1}(\mu^{2})\Gamma_{\mu}^{0}(k,p), \tag*{(9)}
g(μ2)=Zg−1(μ2)g0,(10)g(\mu^{2}) = Z_{g}^{-1}(\mu^{2})g_{0}, \tag*{(10)}

and considering the relation Zg2=Z12/(Z22ZA)Z_{g}^{2} = Z_{1}^{2}/(Z_{2}^{2}Z_{A}) for the strong coupling [126] we obtain:

Z2−1S−1(p)=iγ⋅p+mbm+CFg2Z12Z22ZA∫d4k(2π)4 ZAΔμν(q)γμZ2S(k)Γν(k,p)Z1.(11)Z_{2}^{-1}S^{-1}(p) = i\gamma\cdot p + m^{\mathrm{bm}} + \frac{C_{F}g^{2}Z_{1}^{2}}{Z_{2}^{2}Z_{A}}\int\frac{d^{4}k}{(2\pi)^{4}}\,Z_{A}\Delta_{\mu\nu}(q)\gamma_{\mu}Z_{2}S(k)\frac{\Gamma_{\nu}(k,p)}{Z_{1}}. \tag*{(11)}

After simplification one arrives at the renormalized DSE (1), where the integral is multiplied by the μ\mu-dependent vertex renormalization constant Z1Z_{1}. Now, recall that the STI for the quark-gluon vertex (16) imposes the all-order constraint on the renormalization factors,

Z1=ZHZ2Zc(12)Z_{1} = \frac{Z_{H}Z_{2}}{Z_{c}} \tag*{(12)}

in which ZHZ_{H}, defined by H(k,p,μ2)=ZH(μ2)H0(k,p)H(k,p,\mu^{2}) = Z_{H}(\mu^{2})H_{0}(k,p), is the renormalization of the quark-ghost kernel discussed in § 2.4 and ZcZ_{c} is the renormalization constant of the ghost propagator (17):

G(q2,μ2)=Zc−1(μ2)G0(q2)(13)G(q^{2},\mu^{2}) = Z_{c}^{-1}(\mu^{2})G_{0}(q^{2}) \tag*{(13)}

It is known that the ghost-gluon vertex is not ultraviolet divergent in Landau gauge and that the quark-ghost kernel is finite at one-loop order [121]. This also is the case in the dressed-perturbative approach [90] we employ and we choose ZH=1Z_{H}=1. The STI condition (12) thus simplifies to [127],

Z1=Z2Zc(14)Z_{1} = \frac{Z_{2}}{Z_{c}} \tag*{(14)}

which is still not the relation Z1=Z2Z_{1}=Z_{2} imposed by the WFGTI in QED. Since the self-energy Σ(p2)\Sigma(p^{2}) is also finite at one loop in Landau gauge, the authors of Ref. [90] set Z2=1Z_{2}=1 which simplifies the STI condition further to Z1=1/ZcZ_{1}=1/Z_{c}, and in order to insure multiplicative renormalizability the replacement Zc−1→G(q2)Z_{c}^{-1} \to G(q^{2}) was advocated. Similarly, also insisting on the renormalization-point independence of M(p2)M(p^{2}), the Curtis-Pennington vertex [74] is multiplied by a factor of G(q2)G(q^{2}) in a non-Abelian ansatz for the quark-gluon vertex [69].

An intuitive motivation for this substitution is provided by the observation that the product g2Δ(q2)G2(q2)g^{2}\Delta(q^{2})G^{2}(q^{2}) is a renormalization group invariant quantity in the Taylor scheme4 and can be used to define an effective strong coupling in the DSE kernel. As argued in Ref. [90], the prescription Zc−1→G(q2)Z_{c}^{-1} \to G(q^{2}) compensates for the missing transverse part of the quark-gluon vertex in their approach. This is because omitting the transverse vertex components leads to mishandling of overlapping divergences and therefore compromises the multiplicative renormalizability of the DSE. Indeed, we demonstrated in Ref. [85] that the transverse vertex is crucial to satisfy multiplicative renormalizability and in here we include all components of the vertex.

If we solved simultaneously the coupled quark, gluon and ghost DSE, we would know the value of ZcZ_{c}. This is not our task here, as we pursue the hybrid approach of solving the quark DSE (1) using as input renormalized gluon and ghost propagators from lattice QCD. Note that the STI and TSTI to be discussed in § 2.2 and § 2.3 involve the ghost propagator and the quark-ghost scattering amplitude. Consequently, the vertex Γμa(k,p)\Gamma_{\mu}^{a}(k,p) we derive therefrom also depends on their scalar dressing functions. We therefore absorb ZcZ_{c} on the right-hand side of Eq. (14) in the ghost-dressing function G(q2)G(q^{2}) (see Eqs. (44) to (55)) or, similarly, we may absorb ZcZ_{c} in the strong coupling αs=g2/4π\alpha_{s}=g^{2}/4\pi which is an overall factor in front of the integral. As we allow for a certain variation of αs(μ2)\alpha_{s}(\mu^{2}) at a given renormalization scale imposed by lattice QCD, the impact of this simplification is minimal. Thus, the quark DSE we are concerned with in this work is,

S−1(p)=Z2 iγ⋅p+Z4 m+Z216παs3∫Λd4k(2π)4 Δμν(q) γμS(k) Γν(k,p)(15)S^{-1}(p) = Z_2\,i\gamma\cdot p + Z_4\,m + Z_2 \frac{16\pi\alpha_s}{3} \int^\Lambda\frac{d^4 k}{(2\pi)^4}\,\Delta_{\mu\nu}(q)\,\gamma_\mu S(k)\,\Gamma_\nu(k,p) \tag*{(15)}

with all the 12 tensor structures of the fully dressed quark-gluon vertex we derive in the next sections.

The quark-gluon vertex: general tensor structure

As emphasized in § 2.1, the dressed quark-gluon vertex is a fundamental ingredient in the QCD gap equation and its contributions in the infrared domain play a pivotal role for the emergence of a constituent-quark mass. The vertex will result in strong DCSB if at least one form factor associated with its tensor structure provides sufficient enhancement in the infrared domain; this feature is commonly associated with the rainbow-ladder truncation of the gap- and bound-state equations. However, as argued for example in Refs. [4, 48, 51, 62], this simplest ansatz has limitations when applied to radially excited mesons, the axialvector channel and heavy-flavored mesons. It is therefore of crucial importance to understand the individual as well as collective contributions of the full vertex to the gap equation.

To begin with, as in QED, the fermion-gauge-boson vertex in QCD satisfies a set of constraints imposed by gauge symmetry, multiplicative renormalizability, LKFT as well as the CC, PP and TT symmetry properties of the bare vertex; see, for instance, the discussions in Refs. [78, 84]. Therefore, in a general though not unique decomposition, the fermion-gauge-boson vertex can be written in terms of twelve linearly independent structures [16]. In QCD, the STI associated with this vertex is given by,

iq⋅Γa(k,p)=G(q2)[S−1(k)Ha(k,p)−Hˉa(p,k)S−1(p)](16)i q\cdot\Gamma^a(k,p) = G(q^2)\left[S^{-1}(k)H^a(k,p)-\bar{H}^a(p,k)S^{-1}(p)\right] \tag*{(16)}

where q=k−pq=k-p and the momentum flow is depicted in Figure 1 and S−1(p)=iA(p2) γ⋅p+B(p2)S^{-1}(p)=iA(p^2)\,\gamma\cdot p+B(p^2) is the inverse of the propagator in Eq. (3). The color index aa denotes again the implicit SU(3) group generators. The ghost-dressing function G(q2)G(q^2) is defined by the ghost propagator,

Diagrammatic representation of the dressed quark-gluon vertex

Figure 1. Diagrammatic representation of the dressed quark-gluon vertex Γμ(k,p)\Gamma_\mu(k,p) with its momentum flow defined by the incoming quark momentum pp, the outgoing momentum kk and q=k−pq = k-p.

Dab(q2)=−δabG(q2)q2(17)D^{ab}(q^2)=-\delta^{ab}\frac{G(q^2)}{q^2} \tag*{(17)}

renormalized as G(μ2)=1G(\mu^2)=1, Ha(k,p)=H(k,p)taH^a(k,p)=H(k,p)t^a is the quark-ghost scattering amplitude and Hˉa(p,k)\bar{H}^a(p,k) is obtained from the conjugate, H†(−p,−k)H^\dagger(-p,-k), and the momentum exchange k↔−pk\leftrightarrow-p. This identity allows us to determine the four components of the vertex which are not transverse to the gluon momentum qq. However, this doesn’t exhaust the number of independent tensor structures that can be formed from the momenta kk and pp and the Dirac matrices. The full vertex, Γμa(k,p)=Γμ(k,p)ta\Gamma_\mu^a(k,p)=\Gamma_\mu(k,p)t^a contains additional transverse vertex components and can be decomposed as the sum [16],

Γμ(k,p)=ΓμL(k,p)+ΓμT(k,p).(18)\Gamma_\mu(k,p)=\Gamma_\mu^{L}(k,p)+\Gamma_\mu^{T}(k,p). \tag*{(18)}

The transverse vertex ΓμT(k,p)\Gamma_\mu^{T}(k,p) vertex in Eq. (18) is thus naturally defined by,

iq⋅ΓμT(k,p)=0.(19)i q\cdot\Gamma_\mu^{T}(k,p)=0. \tag*{(19)}

The STI (16) constrains the so-called longitudinal vertex, ΓμL(k,p)\Gamma_\mu^{L}(k,p), to four independent structures,

ΓμL(k,p)=λ1(k,p)γμ+λ2(k,p)12tμγ⋅t−iλ3(k,p)tμ−λ4(k,p)σμνtν,(20)\Gamma_\mu^{L}(k,p)=\lambda_1(k,p)\gamma_\mu+\lambda_2(k,p)\frac{1}{2}t_\mu\gamma\cdot t-i\lambda_3(k,p)t_\mu-\lambda_4(k,p)\sigma_{\mu\nu}t_\nu, \tag*{(20)}

where t=p+kt=p+k is the sum of the incoming and outgoing quark momenta.

The transverse vertex, ultraviolet-finite at one-loop order [17, 83], can generally be decomposed into eight independent tensor structures multiplied by associated scalar form factors, τi(k,p)\tau_i(k,p) [16]:

ΓμT(k,p)=∑i=18τi(k,p)Tμi(k,p).(21)\Gamma_\mu^{T}(k,p)=\sum_{i=1}^{8}\tau_i(k,p)T_\mu^i(k,p). \tag*{(21)}

For the kinematical configuration in Figure 1, the tensors are defined as:

Tμ1(k,p)=i[pμ(k⋅q)−kμ(p⋅q)],(22)T_\mu^1(k,p)=i\left[p_\mu(k\cdot q)-k_\mu(p\cdot q)\right], \tag*{(22)}
Tμ2(k,p)=[pμ(k⋅q)−kμ(p⋅q)]γ⋅t,(23)T_\mu^2(k,p)=\left[p_\mu(k\cdot q)-k_\mu(p\cdot q)\right]\gamma\cdot t, \tag*{(23)}
Tμ3(k,p)=q2γμ−qμγ⋅q,(24)T_\mu^3(k,p)=q^2\gamma_\mu-q_\mu\gamma\cdot q, \tag*{(24)}
Tμ4(k,p)=iq2[γμγ⋅t−tμ]+2qμpνkρσνρ,(25)T_\mu^4(k,p)=iq^2\left[\gamma_\mu\gamma\cdot t-t_\mu\right]+2q_\mu p_\nu k_\rho\sigma_{\nu\rho}, \tag*{(25)}
Tμ5(k,p)=σμνqν,(26)T_\mu^5(k,p)=\sigma_{\mu\nu}q_\nu, \tag*{(26)}
Tμ6(k,p)=−γμ(k2−p2)+tμγ⋅q,(27)T_\mu^6(k,p)=-\gamma_\mu\left(k^2-p^2\right)+t_\mu\gamma\cdot q, \tag*{(27)}
Tμ7(k,p)=i2(k2−p2)[γμγ⋅t−tμ]+tμpνkρσνρ,(28)T_\mu^7(k,p)=\frac{i}{2}\left(k^2-p^2\right)\left[\gamma_\mu\gamma\cdot t-t_\mu\right]+t_\mu p_\nu k_\rho\sigma_{\nu\rho}, \tag*{(28)}
Tμ8(k,p)=−iγμpνkρσνρ−pμγ⋅k+kμγ⋅p.(29)T_\mu^8(k,p)=-i\gamma_\mu p_\nu k_\rho\sigma_{\nu\rho}-p_\mu\gamma\cdot k+k_\mu\gamma\cdot p. \tag*{(29)}

Here, we adopt a slightly modified tensor basis [125] with respect to the one derived by Ball and Chiu [16]. This ensures all transverse form factors are independent of any kinematic singularities in one-loop perturbation theory in an arbitrary covariant gauge.

As discussed in more detail in Refs. [78, 84], the fully dressed vertex must exhibit the same properties under charge-conjugation transformation as the bare vertex, which constrains all the τi(k,p)\tau_i(k,p) in Eq. (21) to be symmetric under the momentum interchange k↔pk\leftrightarrow p, with the exception of the odd functions τ4(k,p)\tau_4(k,p) and τ6(k,p)\tau_6(k,p):

τi(k,p)=τi(p,k),i=1,2,3,5,7,8(30)\tau_i(k,p)=\tau_i(p,k),\qquad i=1,2,3,5,7,8 \tag*{(30)}
τi(k,p)=−τi(p,k),i=4,6.(31)\tau_i(k,p)=-\tau_i(p,k),\qquad i=4,6. \tag*{(31)}

Similarly, λ1(k,p)\lambda_1(k,p), λ2(k,p)\lambda_2(k,p), λ3(k,p)\lambda_3(k,p) and λ4(k,p)\lambda_4(k,p) are symmetric under k↔pk \leftrightarrow p, so as to preserve the charge-conjugation properties of the bare vertex.

The STI (16) completely determines the form factors in Eqs. (20) and their expressions are given in In Section 2.3. The transverse scalar functions in Eq. (21), on the other hand, remain undetermined. Nonetheless, as in QED, two additional transverse identities allow to analogously constrain the form of these transverse vertex functions.

Transverse Slavnov-Taylor identities

A complete fermion-boson vertex based on gauge techniques, such as the one derived from the TTI [85, 88], is a first step towards a realistic quark-gluon interaction in QCD valid in any coupling regime. As in QED, gauge invariance as well as Lorentz and charge transformation invariance constrain the tensor structure of the fully dressed three-point function.

As we already saw, the corresponding gauge symmetry preserving relation between Green functions in QCD theory is provided by the STI of Eq. (16), which relates the quark-gluon vertex with the ghost-dressing function, the inverse quark propagator and the quark-ghost scattering kernel, H(k,p)H(k,p), and its conjugate, Hˉ(k,p)\bar{H}(k,p). The non-transverse vertex saturates this STI while the transverse vertex is automatically conserved. As in QED, this allows to decompose the fermion-boson vertex into a so-called longitudinal and a transverse part with a tensor basis for which we choose the one defined in Eq. (20) and Eqs (22)–(29). On the other hand, the STI is a consequence of SU(3) gauge invariance and merely makes a statement about the non-transverse components of the vertex. By itself, it fails to ensure gauge covariance in its totality. Under certain simplifying conditions, the WFGTI can be recovered form the STI which leads to an abelianized form of the longitudinal quark-gluon vertex [87, 90].

How do the DSE solutions of propagators and vertices depend on an arbitrary gauge transformation? In QED the answer to this question can be obtained from the well-defined set of LKFT laws which leave the DSE and WGTI form-invariant. Given a suitable ansatz for the three-point Green function, Γ(k,p)\Gamma(k,p), that is invariant under a LKFT to another gauge, one can ensure gauge covariance and the WFGTI is satisfied in either gauge. The general rules that govern the LKFT, though, are far from trivial and an extension to the quark propagator in any gauge has only recently been established [82].

In analogy to QED, the transverse components of the vertex can be constrained by TSTI that relate to the curl of the quark-gluon vertex. Their Dirac structure is identical to that of the TTI and in addition they involve the ghost-dressing function and the quark-ghost scattering kernel. The TSTI can be derived in a functional approach [122] and read:

qμΓνa(k,p)−qνΓμa(k,p)=G(q2)[S−1(p)σμνHa(k,p)+Hˉa(p,k)σμνS−1(k)]+2imΓμνa(k,p)+tαϵαμνβΓβ5a(k,p)+Aμνa(k,p)(32)q_\mu\Gamma^a_\nu(k,p) - q_\nu\Gamma^a_\mu(k,p) = G(q^2) \left[ S^{-1}(p)\sigma_{\mu\nu}H^a(k,p) + \bar{H}^a(p,k)\sigma_{\mu\nu}S^{-1}(k) \right] + 2im\Gamma^a_{\mu\nu}(k,p) + t_\alpha\epsilon_{\alpha\mu\nu\beta}\Gamma^{5a}_\beta(k,p) + A^a_{\mu\nu}(k,p) \tag*{(32)}
qμΓν5a(k,p)−qνΓμ5a(k,p)=G(q2)[S−1(p)σμν5Ha(k,p)−Hˉa(p,k)σμν5S−1(k)]+tαϵαμνβΓβa(k,p)+Vμνa(k,p)(33)q_\mu\Gamma^{5a}_\nu(k,p) - q_\nu\Gamma^{5a}_\mu(k,p) = G(q^2) \left[ S^{-1}(p)\sigma^5_{\mu\nu}H^a(k,p) - \bar{H}^a(p,k)\sigma^5_{\mu\nu}S^{-1}(k) \right] + t_\alpha\epsilon_{\alpha\mu\nu\beta}\Gamma^a_\beta(k,p) + V^a_{\mu\nu}(k,p) \tag*{(33)}

where mm is the renormalized current-quark mass in the DSE, σμν5=γ5σμν\sigma^5_{\mu\nu}=\gamma_5\sigma_{\mu\nu}, and Γμνa(k,p)\Gamma^a_{\mu\nu}(k,p) is an inhomogeneous tensor vertex. The two tensor structures, AμνaA^a_{\mu\nu} and VμνaV^a_{\mu\nu}, are Fourier transforms of four-point functions in coordinate space which involve a Wilson line to preserve gauge invariance [122]. Moreover, Γμα(k,p)\Gamma^\alpha_\mu(k,p) couples to the axialvector vertex via the fourth term on the right-hand side of Eq. (32) and likewise, Γμ5α(k,p)\Gamma^{5\alpha}_\mu(k,p) couples to the vector vertex. We stress that these identities bear no explicit dependence on the covariant gauge parameter. They are valid for any covariant gauge and are not limited to the perturbative case.

We follow Refs. [85, 88] and proceed in analogous manner to decouple Eqs. (32) and (33). This is done so by introducing the two tensors,

Tμν1=12ϵαμνβtαqβ,(34)T^1_{\mu\nu} = \frac{1}{2}\epsilon_{\alpha\mu\nu\beta}t^\alpha q^\beta, \tag*{(34)}
Tμν2=12ϵαμνβγαqβ.(35)T^2_{\mu\nu} = \frac{1}{2}\epsilon_{\alpha\mu\nu\beta}\gamma^\alpha q^\beta. \tag*{(35)}

Contracting the axialvector identity (33) with the tensors (34) and (35) yields zero on the left-hand side of the equation, whereas operating the contractions on the right-hand sides with the identities,

Tμν1tαϵαμνβΓβ(k,p)=t2q⋅Γ(k,p)−q⋅t t⋅Γ(k,p),(36)T^1_{\mu\nu}t_\alpha\epsilon_{\alpha\mu\nu\beta}\Gamma_\beta(k,p) = t^2q\cdot\Gamma(k,p)-q\cdot t\,t\cdot\Gamma(k,p), \tag*{(36)}
Tμν2tαϵαμνβΓβ(k,p)=γ⋅tq⋅Γ(k,p)−q⋅tγ⋅Γ(k,p),(37)T^2_{\mu\nu}t_\alpha\epsilon_{\alpha\mu\nu\beta}\Gamma_\beta(k,p) = \gamma\cdot tq\cdot\Gamma(k,p)-q\cdot t\gamma\cdot\Gamma(k,p), \tag*{(37)}

one can recast the two contractions of the axialvector identity into the new form,

q⋅t t⋅Γ(k,p)=G(q2)Tμν1[S−1(p)σμν5H(k,p)−Hˉ(p,k)σμν5S−1(k)]+t2q⋅Γ(k,p)+Tμν1Vμν,(38)\begin{aligned} q\cdot t\,t\cdot\Gamma(k,p)={}&G(q^2)T^1_{\mu\nu}\left[S^{-1}(p)\sigma^5_{\mu\nu}H(k,p)-\bar{H}(p,k)\sigma^5_{\mu\nu}S^{-1}(k)\right]\\ &+t^2q\cdot\Gamma(k,p)+T^1_{\mu\nu}V_{\mu\nu}, \tag*{(38)} \end{aligned}
q⋅t γ⋅Γ(k,p)=G(q2)Tμν2[S−1(p)σμν5H(k,p)−Hˉ(p,k)σμν5S−1(k)]+γ⋅tq⋅Γ(k,p)+Tμν2Vμν,(39)\begin{aligned} q\cdot t\,\gamma\cdot\Gamma(k,p)={}&G(q^2)T^2_{\mu\nu}\left[S^{-1}(p)\sigma^5_{\mu\nu}H(k,p)-\bar{H}(p,k)\sigma^5_{\mu\nu}S^{-1}(k)\right]\\ &+\gamma\cdot tq\cdot\Gamma(k,p)+T^2_{\mu\nu}V_{\mu\nu}, \tag*{(39)} \end{aligned}

in which we dropped color indices for simplicity’s sake. Remarkably, these two new identities involve only the vector vertex, Γμ(k,p)\Gamma_\mu(k,p), and do not explicitly depend on the quark mass mm. Likewise, information about the axialvector vertex, Γμ5(k,p)\Gamma^5_\mu(k,p), can be obtained through an analogous procedure involving the vector TSTI (32).

We have thus uncoupled the vector from the axialvector vertex and obtain, in analogy with Ref. [88], three matrix-valued equations for the scalar projections of Γμ(k,p)\Gamma_\mu(k,p), namely Eqs. (16), (38) and (39). They form a set of twelve linearly independent and coupled linear equations for the twelve unknown scalar vertex functions, λi(k,p)\lambda_i(k,p) and τi(k,p)\tau_i(k,p), which can be solved exactly provided Ha(k,p)H^a(k,p), Hˉa(p,k)\bar{H}^a(p,k) and Vνμ(k,p)V^\mu_\nu(k,p) are known. The terms Tμν1VμνT^1_{\mu\nu}V_{\mu\nu} and Tμν2VμνT^2_{\mu\nu}V_{\mu\nu} are unknown objects, yet they are Lorentz-scalar objects which, without loss of generality, can be conveniently expressed as,

iTμν1Vμν=Y1(k,p)1D+Y2(k,p)γ⋅q+Y3(k,p)γ⋅t+Y4(k,p)[γ⋅q,γ⋅t],(40)iT^1_{\mu\nu}V_{\mu\nu}=Y_1(k,p)\mathbb{1}_D+Y_2(k,p)\gamma\cdot q+Y_3(k,p)\gamma\cdot t+Y_4(k,p)[\gamma\cdot q,\gamma\cdot t], \tag*{(40)}
iTμν2Vμν=Y5(k,p)1D+Y6(k,p)γ⋅q+Y7(k,p)γ⋅t+Y8(k,p)[γ⋅q,γ⋅t].(41)iT^2_{\mu\nu}V_{\mu\nu}=Y_5(k,p)\mathbb{1}_D+Y_6(k,p)\gamma\cdot q+Y_7(k,p)\gamma\cdot t+Y_8(k,p)[\gamma\cdot q,\gamma\cdot t]. \tag*{(41)}

where Yi(k,p)Y_i(k,p) are hitherto unconstrained scalar functions.

A final ingredient we require to derive the expressions for the form factors λi(k,p)\lambda_i(k,p) and τi(k,p)\tau_i(k,p) is the quark-ghost scattering amplitude, Ha(k,p)=H(k,p)aH^a(k,p)=H(k,p)^{a}, which so far has been studied in a dressed-perturbative approach and will be discussed in more detail in § 2.4. Nonetheless, employing a similar expedient as in Eqs. (40) and (41), we can decompose the matrix-valued kernel and parameterize it as follows:

H(k,p)=X0(k,p)1D+X1(k,p) γ⋅k+X2(k,p) γ⋅p+X3(k,p) [γ⋅k,γ⋅p].(42)H(k,p)=X_0(k,p)\mathbb{1}_{D}+X_1(k,p)\,\gamma\cdot k+X_2(k,p)\,\gamma\cdot p+X_3(k,p)\,[\gamma\cdot k,\gamma\cdot p] . \tag*{(42)}
Hˉ(p,k)=X0(p,k)1D−X2(p,k) γ⋅k−X1(p,k) γ⋅p+X3(p,k) [γ⋅k,γ⋅p].(43)\bar{H}(p,k)=X_0(p,k)\mathbb{1}_{D}-X_2(p,k)\,\gamma\cdot k-X_1(p,k)\,\gamma\cdot p+X_3(p,k)\,[\gamma\cdot k,\gamma\cdot p] . \tag*{(43)}

Perturbative expressions for the form factors Xi(k,p)X_i(k,p) are known to one-loop order [17] and yield X0(k,p)=1+O(g2)X_0(k,p)=1+\mathcal{O}(g^2) and Xi=O(g2)X_i=\mathcal{O}(g^2), i=1,2,3i=1,2,3. The same diagrammatic approach based on dressed propagators and vertices that incorporate all Xi(k,p)X_i(k,p) form factors in the longitudinal vertex appears to mostly corroborate this dominance, while X3(k,p)X_3(k,p) may become sizable [92] in certain kinematic limits. For the remainder of this work, we make the simplifying assumption that the contribution of X0(k,p)X_0(k,p) dominates and limit ourselves to the kinematic configuration, k=q/2k=q/2, p=−q/2⇒H(k,p)=H(q/2,−q/2)p=-q/2\Rightarrow H(k,p)=H(q/2,-q/2), and therefore X0(k,p)≈X0(q2)X_0(k,p)\approx X_0(q^2). We neglect sub-leading form factors as their contributions are negligibly small compared to the strength of the transverse vertex, as will be shown shortly.

The decomposition in terms of Lorentz covariants of Eqs. (40), (41), (42) and (43) allows us to write the identities (16), (38) and (39) as a matrix-valued equation system that can be solved by different means. With a set of four projections, with respect to the Dirac trace, one obtains four linearly independent, coupled linear equations whose solutions yield the λi(k,p)\lambda_i(k,p) form factors and, likewise, the τi(k,p)\tau_i(k,p) can be isolated using eight different projections. Thus, with the four projections of Eq. (16) we arrive the following scalar vertex functions of the non-transverse vertex:

λ1(k,p)=12G(q2)X0(q2)[A(k2)+A(p2)],(44)\lambda_1(k,p)=\frac{1}{2}G(q^2)X_0(q^2)\left[A(k^2)+A(p^2)\right] , \tag*{(44)}
λ2(k,p)=G(q2)X0(q2)A(k2)−A(p2)k2−p2,(45)\lambda_2(k,p)=G(q^2)X_0(q^2)\frac{A(k^2)-A(p^2)}{k^2-p^2} , \tag*{(45)}
λ3(k,p)=G(q2)X0(q2)B(k2)−B(p2)k2−p2,(46)\lambda_3(k,p)=G(q^2)X_0(q^2)\frac{B(k^2)-B(p^2)}{k^2-p^2} , \tag*{(46)}
λ4(k,p)=0.(47)\lambda_4(k,p)=0 . \tag*{(47)}

Note that in the absence of the sub-leading X1,2,3(k,p)X_{1,2,3}(k,p) form factors, λ4(k,p)=0\lambda_4(k,p)=0, as in QED [16]. The transverse vertex functions are derived from four projections of each, Eqs. (38) and (39):

τ1(k,p)=−Y12(k2−p2)∇(k,p),(48)\tau_1(k,p)=-\frac{Y_1}{2(k^2-p^2)\nabla(k,p)} , \tag*{(48)}
τ2(k,p)=−Y5−3Y34(k2−p2)∇(k,p),(49)\tau_2(k,p)=-\frac{Y_5-3Y_3}{4(k^2-p^2)\nabla(k,p)} , \tag*{(49)}
τ3(k,p)=12G(q2)X0(q2)[A(k2)−A(p2)k2−p2]+Y24∇(k,p)−t2(Y3−Y5)8(k2−p2)∇(k,p),(50)\tau_3(k,p)=\frac{1}{2}G(q^2)X_0(q^2)\left[\frac{A(k^2)-A(p^2)}{k^2-p^2}\right]+\frac{Y_2}{4\nabla(k,p)}-\frac{t^2(Y_3-Y_5)}{8(k^2-p^2)\nabla(k,p)} , \tag*{(50)}
τ4(k,p)=−6Y4+Y6A8∇(k,p)−t2Y7S8(k2−p2)∇(k,p)(51)\tau_4(k,p) = -\frac{6Y_4 + Y_6^A}{8\nabla(k,p)} - \frac{t^2Y_7^S}{8(k^2-p^2)\nabla(k,p)} \tag*{(51)}
τ5(k,p)=−G(q2)X0(q2)[B(k2)−B(p2)k2−p2]−2Y4+Y6A2(k2−p2)(52)\tau_5(k,p) = -G(q^2)X_0(q^2)\left[\frac{B(k^2)-B(p^2)}{k^2-p^2}\right] - \frac{2Y_4+Y_6^A}{2(k^2-p^2)} \tag*{(52)}
τ6(k,p)=(k−p)2Y24(k2−p2)∇(k,p)−Y3−Y58∇(k,p)(53)\tau_6(k,p) = \frac{(k-p)^2Y_2}{4(k^2-p^2)\nabla(k,p)} - \frac{Y_3-Y_5}{8\nabla(k,p)} \tag*{(53)}
τ7(k,p)=q2(6Y4+Y6A)4(k2−p2)∇(k,p)+Y7S4∇(k,p)(54)\tau_7(k,p) = \frac{q^2(6Y_4+Y_6^A)}{4(k^2-p^2)\nabla(k,p)} + \frac{Y_7^S}{4\nabla(k,p)} \tag*{(54)}
τ8(k,p)=−G(q2)X0(q2)[A(k2)−A(p2)k2−p2]−2Y8Ak2−p2(55)\tau_8(k,p) = -G(q^2)X_0(q^2)\left[\frac{A(k^2)-A(p^2)}{k^2-p^2}\right] - \frac{2Y_8^A}{k^2-p^2} \tag*{(55)}

where Yi≡Yi(k,p)Y_i \equiv Y_i(k,p) and the Gram determinant is defined by:

∇(k,p)=k2p2−(k⋅p)2.(56)\nabla(k,p) = k^2p^2-(k\cdot p)^2. \tag*{(56)}

In addition, the vertex transformation properties under charge conjugation determine the symmetry properties of the YY-functions:

Yi(k,p)=Yi(p,k),i=2,6S,7S,8S,(57)Y_i(k,p)=Y_i(p,k), \qquad i=2,6^S,7^S,8^S, \tag*{(57)}
Yi(k,p)=−Yi(p,k),i=1,3,4,5,6A,7A,8A,(58)Y_i(k,p)=-Y_i(p,k), \qquad i=1,3,4,5,6^A,7^A,8^A, \tag*{(58)}

where, as in Ref. [85], we introduce the decomposition,

Yi(k,p)=YiS(k,p)+YiA(k,p),(59)Y_i(k,p)=Y_i^S(k,p)+Y_i^A(k,p), \tag*{(59)}

for i=6,7,8i=6,7,8, where the superscripts SS and AA denote the symmetric and antisymmetric parts of the corresponding YiY_i functions for k↔pk\leftrightarrow p. Note that there is no contribution of Y6SY_6^S, Y7AY_7^A and Y8SY_8^S to the form factors τi(k,p)\tau_i(k,p) in Eqs. (48) to (55) as a consequence of the symmetry properties in Eqs. (30) and (31) which imply:

Y6S(k,p)=−(k2−p2)Y1(k,p)4∇(k,p),(60)Y_6^S(k,p)=-\frac{(k^2-p^2)Y_1(k,p)}{4\nabla(k,p)}, \tag*{(60)}
Y7A(k,p)=q2Y1(k,p)4∇(k,p),(61)Y_7^A(k,p)=\frac{q^2Y_1(k,p)}{4\nabla(k,p)}, \tag*{(61)}
Y8S(k,p)=−(k−p)2Y2(k,p)8∇(k,p)+(k2−p2)Y3(k,p)8∇(k,p).(62)Y_8^S(k,p)=-\frac{(k-p)^2Y_2(k,p)}{8\nabla(k,p)}+\frac{(k^2-p^2)Y_3(k,p)}{8\nabla(k,p)}. \tag*{(62)}

We stress that while Eqs. (44) to (55) are analytic expressions, the remaining challenge stems from our ignorance of the scalar Yi(k,p)Y_i(k,p) functions. We shall constrain their analytical form with an ansatz for the transverse vertex in perturbative QCD [81] in Section 2.5 and postpone the actual calculation of the nontrivial four-point function Vμν(k,p)V_{\mu\nu}(k,p).

Quark-ghost scattering amplitude

The nonperturbative behavior of the quark-ghost scattering kernel Ha(k,p)H^{a}(k,p) introduced in the STI (16) is unknown. A tractable ansatz that allows for a connection with the perturbative result [17] is given by the one-loop dressed approximation to Ha(k,p)H^{a}(k,p) [90], whose diagrammatic representation is depicted in Figure 2. We use the tensorial decomposition of Eqs. (42) and (43) and, as mentioned previously, limit ourself to the dominant form factor X0(k,p)X_{0}(k,p) with the momentum configuration k=q/2k = q/2 and p=−q/2p = -q/2, so that X0(q/2,−q/2)≡X0(q2)X_{0}(q/2,-q/2) \equiv X_{0}(q^{2}) does not depend on angles. Applying the kinematics of the Feynman diagram in Figure 2, we project out X0(q2)X_{0}(q^{2}),

Diagrammatic representation of the unconnected and connected quark-ghost scattering amplitude

Figure 2. Diagrammatic representation of the unconnected and connected quark-ghost scattering amplitude in its dressed one-loop approximation with the simplified kinematic choice k=q/2k = q/2 and p=−q/2p = -q/2. The yellow-shaded circles denote the dressed quark, gluon and ghost propagators, while the blue- and purple-shaded circles represent the dressed ghost-gluon and quark-gluon vertices, respectively. The outgoing quark and ghost propagators are joined in a non-standard “vertex”.

X0(q2)=14Tr⁡CDH(q/2,−q/2,q)=1+CA4g2∫d4ℓ(2π)4Δμν(ℓ)D(ℓ+q)Tr⁡D[GνS(ℓ+q2)Γμ(ℓ+q2,−q2)].(63)\begin{aligned} X_{0}(q^{2}) = \frac{1}{4} \operatorname{Tr}_{\mathrm{CD}} H(q/2,-q/2,q) \\ &= 1 + \frac{C_{A}}{4}g^{2}\int\frac{d^{4}\ell}{(2\pi)^{4}} \Delta_{\mu\nu}(\ell)D(\ell+q)\operatorname{Tr}_{\mathrm{D}}\left[G_{\nu}S\left(\ell+\frac{q}{2}\right)\Gamma_{\mu}\left(\ell+\frac{q}{2},-\frac{q}{2}\right)\right] . \tag*{(63)} \end{aligned}

where CA=3C_{A}=3 is the Casimir operator in the adjoint representation,

Gν=i(ℓν+qν)H1−iℓνH2,(64)G_{\nu}=i(\ell_{\nu}+q_{\nu})H_{1}-i\ell_{\nu}H_{2}, \tag*{(64)}

is the dressed ghost-gluon vertex and ℓ+q\ell+q and qq are respectively the outgoing and incoming ghost momenta, while ℓ\ell is the gluon momentum exchanged between the quark and the ghost. Note that that the terms proportional to ℓμ\ell_{\mu} in the vertex (64) are transverse to the gluon propagator Δμν(ℓ)\Delta_{\mu\nu}(\ell) in Landau gauge; H2H_{2} does therefore not contribute to the quark-ghost kernel. Moreover, the momentum qq coincides with the gluon momentum in the gap equation (1).

The remaining form factor H1(ℓ+q,q)≃H1(ℓ+q,0)≡H1(x)H_{1}(\ell+q,q) \simeq H_{1}(\ell+q,0) \equiv H_{1}(x) is parametrized by the expression [130],

H1(x)=c(1+a2x2x4+b4)+(1−c)w4w4+x4,(65)H_{1}(x)=c\left(1+\frac{a^{2}x^{2}}{x^{4}+b^{4}}\right)+(1-c)\frac{w^{4}}{w^{4}+x^{4}}, \tag*{(65)}

fitted to lattice-QCD data: c=1.26c = 1.26, a=0.80a = 0.80 GeV, b=1.3b = 1.3 GeV and w=0.65w = 0.65 GeV. In [86] the replacement of the tree-level ghost-gluon vertex by a dressed vertex enhanced X0(q2)X_{0}(q^{2}) overall and by about 20% in the peak region q2=1q^{2}=1–44 GeV2^{2}. Adding the transverse part of the vertex in the gap equation this observation is not significantly altered. Note that Γμ(k,p)=ΓμL(k,p)+ΓμT(k,p)\Gamma_{\mu}(k,p) = \Gamma_{\mu}^{L}(k,p) + \Gamma_{\mu}^{T}(k,p) enters both the DSE (15) and the quark-ghost scattering amplitude in Eq. (63) for consistency.

Ansatz to constrain the nonlocal Lorentz scalars Tμν1VμνT_{\mu\nu}^{1}V_{\mu\nu} and Tμν2VμνT_{\mu\nu}^{2}V_{\mu\nu}

The vertex ansatz proposed in [81] is constrained by two requirements: it provides the multiplicative renormalizability of the fermion propagator and produces a gauge-independent critical coupling for DCSB. The appeal of this ansatz lies in the fact that it is merely expressed in terms of the quark propagator’s scalar and vector functions, B(p2)B(p^{2}) and A(p2)A(p^{2}), respectively. In the particular kinematical configuration k2≫p2k^{2} \gg p^{2}, the transverse form factors of this ansatz are given by:

τ1(k,p)=a1B(k2)−B(p2)(k2+p2)(k2−p2),(66)\tau_{1}(k,p) = a_{1}\frac{B(k^{2}) - B(p^{2})}{(k^{2}+p^{2})(k^{2}-p^{2})}, \tag*{(66)}
τ2(k,p)=a2A(k2)−A(p2)(k2+p2)(k2−p2),(67)\tau_{2}(k,p) = a_{2}\frac{A(k^{2}) - A(p^{2})}{(k^{2}+p^{2})(k^{2}-p^{2})}, \tag*{(67)}
τ3(k,p)=a3A(k2)−A(p2)k2−p2,(68)\tau_{3}(k,p) = a_{3}\frac{A(k^{2}) - A(p^{2})}{k^{2}-p^{2}}, \tag*{(68)}
τ4(k,p)=a4B(k2)−B(p2)4k2p2,(69)\tau_{4}(k,p) = a_{4}\frac{B(k^{2}) - B(p^{2})}{4k^{2}p^{2}}, \tag*{(69)}
τ5(k,p)=−a5B(k2)−B(p2)k2−p2,(70)\tau_{5}(k,p) = -a_{5}\frac{B(k^{2}) - B(p^{2})}{k^{2}-p^{2}}, \tag*{(70)}
τ6(k,p)=−a6k2+p2(k2−p2)2[A(k2)−A(p2)],(71)\tau_{6}(k,p) = -a_{6}\frac{k^{2}+p^{2}}{(k^{2}-p^{2})^{2}}\left[A(k^{2}) - A(p^{2})\right], \tag*{(71)}
τ7(k,p)=−[a4q22k2p2+a7k2+p2]B(k2)−B(p2)k2−p2,(72)\tau_{7}(k,p) = -\left[\frac{a_{4}q^{2}}{2k^{2}p^{2}}+\frac{a_{7}}{k^{2}+p^{2}}\right]\frac{B(k^{2}) - B(p^{2})}{k^{2}-p^{2}}, \tag*{(72)}
τ8(k,p)=a8A(k2)−A(p2)k2−p2,(73)\tau_{8}(k,p) = a_{8}\frac{A(k^{2}) - A(p^{2})}{k^{2}-p^{2}}, \tag*{(73)}

where the coefficients aia_i are unknown constants.

Inserting the above form factors in Eqs. (48) to (55), we derive the following expressions for the Yi(k,p)Y_i(k,p) which consequently depend on the aia_i coefficients:

Y1(k,p)=−2a1[B(k2)−B(p2)]∇(k,p)k2+p2,(74)Y_{1}(k,p) = -2a_{1}\left[B(k^{2}) - B(p^{2})\right]\frac{\nabla(k,p)}{k^{2}+p^{2}}, \tag*{(74)}
Y2(k,p)=12[A(k2)−A(p2)]{(k2−p2)[G(q2)X0(q2)−2a3]−2(k2+p2k2−p2)(k+p)2a6},(75)Y_{2}(k,p) = \frac{1}{2}\left[A(k^{2}) - A(p^{2})\right]\left\{(k^{2}-p^{2})\left[G(q^{2})X_{0}(q^{2}) - 2a_{3}\right] - 2\left(\frac{k^{2}+p^{2}}{k^{2}-p^{2}}\right)(k+p)^{2}a_{6}\right\}, \tag*{(75)}
Y3(k,p)=12[A(k2)−A(p2)]{−q2[G(q2)X0(q2)−2a3]+4∇(k,p)k2+p2a2+2(k2+p2)a6}(76)Y_{3}(k,p)=\frac{1}{2}\left[A(k^{2})-A(p^{2})\right]\left\{-q^{2}\left[G(q^{2})X_{0}(q^{2})-2a_{3}\right]+4\frac{\nabla(k,p)}{k^{2}+p^{2}}a_{2}+2(k^{2}+p^{2})a_{6}\right\} \tag*{(76)}
Y4(k,p)=−B(k2)−B(p2)4k2p2(k2+p2){2(k2+p2)∇(k,p)a4+2k2p2(k2+p2)[a5−G(q2)X0]+k2p2(k+p)2a7}(77)Y_{4}(k,p)=-\frac{B(k^{2})-B(p^{2})}{4k^{2}p^{2}(k^{2}+p^{2})}\left\{2(k^{2}+p^{2})\nabla(k,p)a_{4}+2k^{2}p^{2}(k^{2}+p^{2})\left[a_{5}-G(q^{2})X_{0}\right]+k^{2}p^{2}(k+p)^{2}a_{7}\right\} \tag*{(77)}
Y5(k,p)=32[A(k2)−A(p2)]{−(k−p)2[G(q2)X0−2a3]+43∇(k,p)k2+p2a2+2(k2+p2)a6}(78)Y_{5}(k,p)=\frac{3}{2}\left[A(k^{2})-A(p^{2})\right]\left\{-(k-p)^{2}\left[G(q^{2})X_{0}-2a_{3}\right]+\frac{4}{3}\frac{\nabla(k,p)}{k^{2}+p^{2}}a_{2}+2(k^{2}+p^{2})a_{6}\right\} \tag*{(78)}
Y6A(k,p)=B(k2)−B(p2)2k2p2(k2+p2){2(k2+p2)∇(k,p)a4+6k2p2(k2+p2)(a5−G(q2)X0)+k2p2(k+p)2a7}(79)Y_{6}^{A}(k,p)=\frac{B(k^{2})-B(p^{2})}{2k^{2}p^{2}(k^{2}+p^{2})}\left\{2(k^{2}+p^{2})\nabla(k,p)a_{4}+6k^{2}p^{2}(k^{2}+p^{2})\left(a_{5}-G(q^{2})X_{0}\right)+k^{2}p^{2}(k+p)^{2}a_{7}\right\} \tag*{(79)}
Y7S(k,p)=a7[B(k2)−B(p2)]k2−p2k2+p2(80)Y_{7}^{S}(k,p)=a_{7}\left[B(k^{2})-B(p^{2})\right]\frac{k^{2}-p^{2}}{k^{2}+p^{2}} \tag*{(80)}
Y8A(k,p)=−12[A(k2)−A(p2)](a8+G(q2)X0)(81)Y_{8}^{A}(k,p)=-\frac{1}{2}\left[A(k^{2})-A(p^{2})\right]\left(a_{8}+G(q^{2})X_{0}\right) \tag*{(81)}

While the ansatz given by Eqs. (66) to (73) allows us to derive analytic expressions for the Yi(k,p)Y_i(k,p) form factors, we trade our ignorance of their functional form—or of the nonlocal tensor VμνV_{\mu\nu} in general—for that of relative coefficients. The task is now to find a set of solutions a⃗:={a1,a2,a3,a4,a5,a6,a7,a8}\vec{a}:=\{a_{1},a_{2},a_{3},a_{4},a_{5},a_{6},a_{7},a_{8}\} that reproduces mass and wave-function renormalization solutions from lattice QCD or successful phenomenological models employed in calculations of hadron properties. It turns out that this set is rather small since some of the aia_i are tightly constrained due to multiplicative renormalizability, as we established with the two relations [85],

a2+2a3−2a8=0(82)a_{2}+2a_{3}-2a_{8}=0 \tag*{(82)}
a6=12(83)a_{6}=\frac{1}{2} \tag*{(83)}

As we shall see, DSE solutions are found in the parameter space defined by the intervals, −2≤ai≤+2-2\leq a_i\leq+2. Allowing for the parameters to leave this range mostly leads to unstable or non-converging iterations when solving the integral equations.

Gauge sector: gluon and ghost propagators from lattice QCD

We focus on the complete tensor structures of the dressed quark-gluon vertex, in particular on its contribution to DCSB in the gap equation, and, as such, we do not attempt to solve the coupled DSEs for the gluon and the ghost. This exercise is certainly most worthwhile and of particular interest to understand the effect of the vertex in the dressed gluon propagator and its feedback on the quark gap equation. We postpone this more ample and complex study and rather employ the dressed gluon and ghost propagators which have been calculated on the lattice and for which several numerical results are readily available.

We use the following three data sets for the gluon and ghost propagators in Landau gauge obtained with lattice-QCD simulations:

  • I. Quenched lattice data by Bogolubsky et al. [29] generated with β=6/q02=5.70\beta= 6/q_{0}^{2} = 5.70, a lattice spacing a=0.17a = 0.17 fm and a lattice volume of L4=(13.6 fm)4L^{4} = (13.6\ \mathrm{fm})^{4}.

  • II. Quenched lattice data for the gluon by Dudal et al. [35] generated with β=6.0\beta= 6.0, lattice spacing a=0.1016(25)a = 0.1016(25) fm and two different lattice volumes: L4=(6.57 fm)4L^{4} = (6.57\ \mathrm{fm})^{4} and L4=(8.21 fm)4L^{4} = (8.21\ \mathrm{fm})^{4}. Quenched lattice data for the ghost by Duarte et al. [36] generated with β=6.0\beta= 6.0, lattice spacing a=0.1016(25)a = 0.1016(25) fm and L4=(8.128 fm)4L^{4} = (8.128\ \mathrm{fm})^{4}.

  • III. Partially unquenched lattice data by Ayala et al. [30] from the gauge configurations generated by the European Twisted Mass Collaboration (ETMC) for Nf=2+1+1N_{f} = 2 + 1 + 1 with β=1.90\beta= 1.90, (L/a)3×T/a=323×64(L/a)^{3} \times T/a = 32^{3} \times64 and β=1.95\beta= 1.95, (L/a)3×T/a=483×96(L/a)^{3} \times T/a = 48^{3} \times96.

Each of these three data sets features distinct advantages. The propagators of the sets I and III were calculated with a large lattice volume and the lowest accessible momenta are p≈70p \approx70 MeV and p≈100p \approx100 MeV, while the largest momenta are p≈4p \approx4 GeV and p≈4.5p \approx4.5 GeV, respectively. Data set III additionally allows us to analyze unquenching effects in the solutions of the quark DSE. The lattice data of set II were obtained with a smaller lattice volume, however for a considerably smaller lattice spacing aa. Therefore, larger momenta up to p≈7.7p \approx7.7 GeV are accessible, which provides additional information on the functional behavior in the perturbative domain of these functions.

In exploiting the available lattice data we resort to analytic fits motivated by theoretical considerations. The lattice propagators, Δ(q2)\Delta(q^{2}) and G(q2)G(q^{2}), of set I and II are parametrized with an expression that combines the refined Gribov-Zwanziger tree-level gluon propagator in the infrared domain supplemented by the 1-loop renormalization group behavior at larger momenta. This amounts to a renormalization-group improved Padé approximation and can be written as [35]

Δ(q2)=Zq2+M12q4+M22q2+M34[1+ωln⁡(q2+mgl2ΛQCD2)]γgl,(84)\Delta(q^{2}) = Z\frac{q^{2} + M_{1}^{2}}{q^{4} + M_{2}^{2}q^{2} + M_{3}^{4}}\left[1 + \omega\ln\left(\frac{q^{2} + m_{\mathrm{gl}}^{2}}{\Lambda_{\mathrm{QCD}}^{2}}\right)\right]^{\gamma_{\mathrm{gl}}}, \tag*{(84)}

where ω=11Ncαs(μ)/12π\omega= 11N_{c}\alpha_{s}(\mu)/12\pi, ΛQCD=0.425\Lambda_{\mathrm{QCD}} = 0.425 GeV and γgl=−13/22\gamma_{\mathrm{gl}} = -13/22 is the 1-loop anomalous gluon dimension. We choose μ=4.3\mu= 4.3 GeV and αs=0.3\alpha_{s} = 0.3 [90, 129]. A least-squares fit yields χ2/d.o.f.=1.788\chi^{2}/\mathrm{d.o.f.} = 1.788 for the renormalized gluon-dressing function of set I [29]:

Z=1.440±0.003,M12=2.880±0.054 GeV2,M22=0.434±0.021 GeV2,M34=0.527±0.009 GeV4,mgl2=0.305±0.071 GeV2.(85)\begin{aligned} Z &= 1.440 \pm0.003, & M_{1}^{2} &= 2.880 \pm0.054\ \mathrm{GeV}^{2}, \\ M_{2}^{2} &= 0.434 \pm0.021\ \mathrm{GeV}^{2}, & M_{3}^{4} &= 0.527 \pm0.009\ \mathrm{GeV}^{4}, \\ m_{\mathrm{gl}}^{2} &= 0.305 \pm0.071\ \mathrm{GeV}^{2}. \tag*{(85)} \end{aligned}

and in case of set II [35] we obtain χ2/d.o.f.=0.031\chi^2/\mathrm{d.o.f.}=0.031 in a best fit to the bare gluon propagator with the parameters,

Z=8.421±0.040,M12=2.743±0.184 GeV2,M22=0.519±0.073 GeV2,M34=0.356±0.029 GeV4,mgl2=0.259±0.262 GeV2,(86)\begin{aligned} Z &= 8.421 \pm0.040, & M_1^2 &= 2.743 \pm0.184\ \mathrm{GeV}^2,\\ M_2^2 &= 0.519 \pm0.073\ \mathrm{GeV}^2, & M_3^4 &= 0.356 \pm0.029\ \mathrm{GeV}^4,\\ m_{\mathrm{gl}}^2 &= 0.259 \pm0.262\ \mathrm{GeV}^2, \tag*{(86)} \end{aligned}

where Δ(q2)\Delta(q^2) must be renormalized by a factor 0.3143 μ20.3143\ \mu^2. Similarly, we make use of the ghost-dressing parametrization introduced in Ref. [36],

G(q2)=Zq4+M22q2+M14q4+M42q2+M34[1+ωln⁡(q2+m14q2+m02ΛQCD2)]γgh,(87)G(q^2)=Z\frac{q^4+M_2^2q^2+M_1^4}{q^4+M_4^2q^2+M_3^4}\left[1+\omega\ln\left(\frac{q^2+\frac{m_1^4}{q^2+m_0^2}}{\Lambda_{\mathrm{QCD}}^2}\right)\right]^{\gamma_{\mathrm{gh}}}, \tag*{(87)}

with the anomalous ghost dimension γgh=−9/44\gamma_{\mathrm{gh}}=-9/44, while ω\omega and ΛQCD\Lambda_{\mathrm{QCD}} take on the same values as in Eq. (84). Fitting this expression to the ghost propagator of set I [29] renormalized at μ=4.3 GeV\mu=4.3\ \mathrm{GeV} we find χ2/d.o.f.=1.534\chi^2/\mathrm{d.o.f.}=1.534 and the following parameters:

Z=1.059±0.001,M14=35.562±0.464 GeV4,M22=33.471±0.391 GeV2,M34=14.790±0.188 GeV4,M42=29.193±0.331 GeV2,m02=0.018±0.010 GeV2,m14=0.001±0.0002 GeV4.(88)\begin{aligned} Z &= 1.059 \pm0.001, & M_1^4 &= 35.562 \pm0.464\ \mathrm{GeV}^4, & M_2^2 &= 33.471 \pm0.391\ \mathrm{GeV}^2,\\ M_3^4 &= 14.790 \pm0.188\ \mathrm{GeV}^4, & M_4^2 &= 29.193 \pm0.331\ \mathrm{GeV}^2, & m_0^2 &= 0.018 \pm0.010\ \mathrm{GeV}^2,\\ m_1^4 &= 0.001 \pm0.0002\ \mathrm{GeV}^4. \tag*{(88)} \end{aligned}

Likewise, a least-squares fit to the bare ghost propagator of Ref. [36] yields χ2/d.o.f.=0.247\chi^2/\mathrm{d.o.f.}=0.247 with the parameter set:

Z=5.068±0.086,M14=19.281±3.685 GeV4,M22=27.721±3.388 GeV2,M34=7.695±1.925 GeV4,M42=24.340±2.833 GeV2,m02=0.527±0.083 GeV2,m14=0.018±0.021 GeV4.(89)\begin{aligned} Z &= 5.068 \pm0.086, & M_1^4 &= 19.281 \pm3.685\ \mathrm{GeV}^4, & M_2^2 &= 27.721 \pm3.388\ \mathrm{GeV}^2,\\ M_3^4 &= 7.695 \pm1.925\ \mathrm{GeV}^4, & M_4^2 &= 24.340 \pm2.833\ \mathrm{GeV}^2, & m_0^2 &= 0.527 \pm0.083\ \mathrm{GeV}^2,\\ m_1^4 &= 0.018 \pm0.021\ \mathrm{GeV}^4. \tag*{(89)} \end{aligned}

Accordingly, G(q2)G(q^2) must be renormalized by a factor 4.706 at μ=4.3 GeV\mu=4.3\ \mathrm{GeV}.

On the other hand, we fit the unquenched data in Ref. [30] with a Padé approximant in the region q≤q0=3 GeVq\leq q_0=3\ \mathrm{GeV} and a renormalization-group improved continuation for q>q0=3 GeVq>q_0=3\ \mathrm{GeV} given by

Δ(q2)=α1+α2q2α3+α4q2+α5q4,q≤q0,(90)\Delta(q^2)=\frac{\alpha_1+\alpha_2q^2}{\alpha_3+\alpha_4q^2+\alpha_5q^4},\qquad q\leq q_0, \tag*{(90)}
Δ(q2)=α1+α2q02α3+α4q02+α5q04q02q2(log⁡(q0/ΛQCD)log⁡(q/ΛQCD))γgl/β,q>q0,(91)\Delta(q^2)=\frac{\alpha_1+\alpha_2q_0^2}{\alpha_3+\alpha_4q_0^2+\alpha_5q_0^4}\frac{q_0^2}{q^2}\left(\frac{\log(q_0/\Lambda_{\mathrm{QCD}})}{\log(q/\Lambda_{\mathrm{QCD}})}\right)^{\gamma_{\mathrm{gl}}/\beta},\qquad q>q_0, \tag*{(91)}

where the anomalous dimension including flavor dependence is in this case γgl=132−23Nf\gamma_{\mathrm{gl}}=\frac{13}{2}-\frac{2}{3}N_f, β=11−23Nf\beta=11-\frac{2}{3}N_f, Nf=4N_f=4 and ΛQCD=600 MeV\Lambda_{\mathrm{QCD}}=600\ \mathrm{MeV}. Analogously, the ghost propagator is fitted to the expressions,

G(q2)=β1+β2q2β3+β4q2,q≤q0,(92)G(q^2)=\frac{\beta_1+\beta_2q^2}{\beta_3+\beta_4q^2},\qquad q\leq q_0, \tag*{(92)}
G(q2)=β1+β2q02β3+β4q02(log⁡(q0/ΛQCD)log⁡(q/ΛQCD))γgh/β,q>q0.(93)G(q^2)=\frac{\beta_1+\beta_2q_0^2}{\beta_3+\beta_4q_0^2}\left(\frac{\log(q_0/\Lambda_{\mathrm{QCD}})}{\log(q/\Lambda_{\mathrm{QCD}})}\right)^{\gamma_{\mathrm{gh}}/\beta},\qquad q>q_0. \tag*{(93)}

where γgh=9/4\gamma_{\mathrm{gh}} = 9/4. The fitted parameters are [131]:

α1=4.305 GeV2,α2=0.979,α3=1 GeV4,α4=1.581 GeV2,α5=1.098,(94)\begin{aligned} \alpha_1 &= 4.305\ \mathrm{GeV}^2,\quad\alpha_2 = 0.979,\quad\alpha_3 = 1\ \mathrm{GeV}^4,\quad\alpha_4 = 1.581\ \mathrm{GeV}^2,\quad\alpha_5 = 1.098, \tag*{(94)} \end{aligned}
β1=2.750 GeV2,β2=1.855,β3=1 GeV2,β4=1.864.(95)\beta_1 = 2.750\ \mathrm{GeV}^2,\quad\beta_2 = 1.855,\quad\beta_3 = 1\ \mathrm{GeV}^2,\quad\beta_4 = 1.864. \tag*{(95)}

Results

Solving the Dyson-Schwinger equation for real space-like momenta

The DSE (15) is solved by first projecting out the scalar functions A(p2)A(p^2) and B(p2)B(p^2) of the general solution (3). This yields two coupled, nonlinear integral equations which additionally couple to the integral equation for X0(q2)X_0(q^2) (63). We solve these three equations, F(p2)=A(p2),B(p2)\mathcal{F}(p^2) = A(p^2), B(p^2) and X0(q2)X_0(q^2), simultaneously via an iterative procedure using as convergence criterium,

ϵF=∣Fn+1(p2)−Fn(p2)∣Fn(p2)=10−3,(96)\epsilon_{\mathcal{F}} = \frac{\left|\mathcal{F}^{n+1}(p^2)-\mathcal{F}^{n}(p^2)\right|}{\mathcal{F}^{n}(p^2)} = 10^{-3}, \tag*{(96)}

where n=20−30n = 20 - 30 iterations suffice. Imposing the constraints (82) and (83) we have established the following values for the coefficients aia_i:

a1a_1a2a_2a3a_3a4a_4a5a_5a6a_6a7a_7a8a_8
22−1-1−1-1112\frac{1}{2}12\frac{1}{2}0

Table 4.2.

The crucial choice is a3≤0a_3 \leq0, as positive values of this coefficient inevitably lead to spurious or non-converging solutions of the integral-equation system. With the choice a3=−1a_3 = -1 and with a6a_6 fixed by the condition (83) we then adjust a2a_2 and a8a_8 to satisfy condition (82). It turns out that the wave-function renormalization, Z(p2)Z(p^2), is quite sensitive to a8≠0a_8 \ne0 and either positive or negative values impact its functional behavior unfavorably in the infrared domain; i.e. Z(p2)Z(p^2) turns back to its perturbative value ≈1\approx1, a feature also observed in conjunction with the Ball-Chiu vertex. A similar observation holds for a4>0a_4 > 0 and we therefore use a4=−1a_4 = -1. Thus, setting a8=0a_8 = 0 we have a2=2a_2 = 2 while a1>0a_1 > 0 leads to an increase in the mass function. The solutions exhibit little dependence on the choice for a5a_5, whereas a7<0a_7 < 0 results in a mass function that decreases too fast and yields negative masses at large momenta. We remark that this choice is not unique, but within the constraints of Eqs. (83) and (83) and with the necessary choice of a3≤0a_3 \leq0 little room for variation is left as we checked with some n>200n > 200 configurations.

Our quark-gluon vertex is now fully determined and in the following we present the solutions for M(p2)M(p^2), Z(p2)Z(p^2) and X0(q2)X_0(q^2) for the three sets of gluon and ghost propagators discussed in Section 3. Note that in what follows we determine Z4(μ,Λ)Z_4(\mu,\Lambda) by imposing m(μ)≡M(μ)m(\mu) \equiv M(\mu) at μ=4.3 GeV\mu= 4.3\ \mathrm{GeV}. Considering degenerate light-quark masses in the range m(2 GeV)≈30m(2\ \mathrm{GeV}) \approx30–50 MeV50\ \mathrm{MeV} employed in lattice simulations [30], we choose m(4.3 GeV)=25 MeVm(4.3\ \mathrm{GeV}) = 25\ \mathrm{MeV}. Along with A(4.3 GeV)=0.98A(4.3\ \mathrm{GeV})=0.98, we therefore impose the renormalization conditions:

Z2(μ,Λ)=A(μ2)1+Σv(μ,Λ)(97)Z_2(\mu,\Lambda)=\frac{A(\mu^2)}{1+\Sigma_v(\mu,\Lambda)} \tag*{(97)}
Z4(μ,Λ)=B(μ2)−Z2(μ,Λ)Σs(μ,Λ)m(μ)(98)Z_4(\mu,\Lambda)=\frac{B(\mu^2)-Z_2(\mu,\Lambda)\Sigma_s(\mu,\Lambda)}{m(\mu)} \tag*{(98)}

where Σs(μ,Λ)\Sigma_s(\mu,\Lambda) and Σv(μ,Λ)\Sigma_v(\mu,\Lambda) are the integrals after the usual scalar and vector projections of the right-hand-side of Eq. (15) with PB=1D/4P_B=\mathbb{1}_D/4 and PA=−iγ⋅p/4p2P_A=-i\gamma\cdot p/4p^2, respectively:

A(p2)=Z2(μ,Λ)+Z2(μ,Λ)Σv(p,Λ)(99)A(p^2)=Z_2(\mu,\Lambda)+Z_2(\mu,\Lambda)\Sigma_v(p,\Lambda) \tag*{(99)}
B(p2)=Z4(μ,Λ)m(μ)+Z2(μ,Λ)Σs(p,Λ)(100)B(p^2)=Z_4(\mu,\Lambda)m(\mu)+Z_2(\mu,\Lambda)\Sigma_s(p,\Lambda) \tag*{(100)}

In order to appreciate the effect of the non-transverse components of Γμ(k,p)\Gamma_\mu(k,p), we first omit the contributions of ΓμT\Gamma_\mu^T following the work of Ref. [86], however without artificially enhancing X0(q2)X_0(q^2) by an inversion procedure. The mass and wave-renormalization functions along with the form factor X0(q2)X_0(q^2) are plotted in Figure 3 for the gluon and ghost propagator sets discussed in § 3. The values for m(μ)m(\mu) and αs(μ)\alpha_s(\mu) at the renormalization scale μ\mu are detailed for each set in Figure 3 and we remark the following: the gluon- and ghost-dressing functions of sets I and II are renormalized at μ=4.3\mu= 4.3 GeV where we employ αs(μ)=0.30\alpha_s(\mu) = 0.30 at this scale following Refs. [90, 129]. We also renormalize the DSE at μ=4.3\mu= 4.3 GeV when using the unquenched propagators of set III but increase the strong coupling to αs(μ)=0.35\alpha_s(\mu) = 0.35, as otherwise only a trivial solution, M(0)≈0M(0) \approx0, is found. This is in line with the behavior of the Taylor coupling in Ref. [30].

Figure 3. Mass function, M(p2)M(p^2), wave renormalization, Z(p2)Z(p^2), and the quark-ghost kernel form factor, X0(q2)X_0(q^2), for three sets of gluon and ghost propagators, using only ΓLμ(k,p)\Gamma_L^\mu(k,p) and the form factors of Eqs. (44)–(47) in the DSE (15). Set I [29] and II [35, 36]: αs(μ)=0.30\alpha_s(\mu)=0.30, m(μ)=25m(\mu)=25 MeV, μ=4.3\mu=4.3 GeV. Set III [30]: αs(μ)=0.35\alpha_s(\mu)=0.35, m(μ)=25m(\mu)=25 MeV, μ=4.3\mu=4.3 GeV.

The mass functions in Figure 3 exhibit the well-known fast rise at about p=1p = 1 GeV and saturate at low momenta, where the mass M(0)M(0) ranges between 100 MeV and 163 MeV. While these solutions exhibit appreciable DCSB, it is much lower than that obtained with phenomenological models and does not yield the correct pion mass and weak decay constant [4]. The functional behavior of Z(p2)Z(p^2) is characteristic of that produced with a Ball-Chiu vertex and the Maris-Tandy model [42] in the gap equation and indeed, in Eqs. (44) to (47) the longitudinal form factors describe a “ghost-corrected” Ball-Chiu vertex. However, the depth of the minimum of Z(p2)Z(p^2) is less pronounced than with the phenomenological approach and the growth in the deep infrared is exacerbated with values of Z(0)>1Z(0) > 1. Clearly, limiting the vertex to its non-transverse components yields mass- and wave-renormalization functions which are unlike those required for realistic applications in hadron physics. We note that the leading form factor of the quark-ghost scattering kernel behaves similarly in all three cases with a maximum located around 2–5 GeV2^2 but different magnitudes. Notably, the more enhanced solutions for X0(q2)X_0(q^2) using set I and II are also accompanied by an increased DCSB in the corresponding mass functions. The net contribution of X0(q2)X_0(q^2) to M(p2)M(p^2) is 27% when using set I and 29% when using set II; solving the DSE with set III and keeping X0(q2)=1X_0(q^2) = 1 does not yield a mass gap.

We now introduce the transverse components of the quark-gluon vertex, Γμ(k,p)=ΓμL(k,p)+ΓμT(k,p)\Gamma_\mu(k,p) = \Gamma_\mu^L(k,p) + \Gamma_\mu^T(k,p), in the DSE and therefore all 12 form factors, Eqs. (44) to (47) and (2.48) to (2.55), come into play. The solutions for M(p2)M(p^2), Z(p2)Z(p^2) and X0(q2)X_0(q^2) are again juxtaposed for the three sets of gluon and ghost propagators in Figure 4. We immediately notice the drastic increase of the mass functions, exemplified by M(0)=500M(0) = 500 MeV for set I, M(0)=546M(0) = 546 MeV for set II and M(0)=458M(0) = 458 MeV for set III. In other words, including the transverse vertex leads to a 235%–350% increase in DCSB, while the contribution of the quark-ghost scattering amplitude X0(q2)X_0(q^2) is in the range of 14%–17%. The resulting wave renormalizations likewise exhibit a different behavior at low momenta: while they do not monotonously fall as lattice-QCD calculations indicate, albeit with larger errors at low momenta [105], they flatten out and only slightly bend over compared to the solutions obtained with merely the longitudinal vertex. The quark-ghost form factors X0(q2)X_0(q^2) in Figure 4 are similar to those in Figure 3, yet we note that overall their maximum values decrease.

Figure 4. Mass function, M(p2)M(p^2), wave renormalization, Z(p2)Z(p^2), and leading quark-ghost kernel form factor, X0(q2)X_0(q^2), obtained with three sets of gluon and ghost propagators, Γμ(k,p)=ΓμL(k,p)+ΓμT(k,p)\Gamma_\mu(k,p) = \Gamma_\mu^{\mathrm{L}}(k,p) + \Gamma_\mu^{\mathrm{T}}(k,p) and associated form factors of Eqs. (44)–(55) in the DSE (15). Current-quark masses, strong couplings and renormalization scale as in Figure 3.

A worthwhile observation is that no solutions for the coupled integral equations (99), (100) and (63) are found if we define Yi(k,p)=0Y_i(k,p) = 0, i=1,…,8i = 1, \ldots, 8, regardless of the value of the strong coupling αs(μ)\alpha_s(\mu). The same is true if we set τ4(k,p)=τ7(k,p)=0\tau_4(k,p) = \tau_7(k,p) = 0. Indeed, these two transverse form factors are responsible for the overwhelming contribution to DCSB in the gap equation. Keeping them both and discarding, one by one, the remaining τi(k,p)\tau_i(k,p), i=1,2,3,5,6,8i = 1, 2, 3, 5, 6, 8 or combinations thereof, results in very similar solutions. Depending on which form factor is discarded, variations of 5–10% of the mass functions plotted in Figure 4 are observed, some leading to a decrease while others to an increase of M(p2)M(p^2). Hence, the form factors other than τ4(k,p)\tau_4(k,p) and τ7(k,p)\tau_7(k,p) are merely responsible for the fine details of the transverse vertex. We note that τ4(k,p)\tau_4(k,p) and τ7(k,p)\tau_7(k,p) are very similar in structure to Y4(k,p)Y_4(k,p), Y6A(k,p)Y_6^A(k,p) and Y7S(k,p)Y_7^S(k,p), all of which are proportional to the difference of the mass functions, ΔB(k,p)=B(k)−B(p)\Delta B(k,p) = B(k) - B(p). On the other hand, if we retain only τ3(k,p)\tau_3(k,p) and τ5(k,p)\tau_5(k,p), which also explicitly depend on ΔB(k,p)\Delta B(k,p), the resulting mass functions exhibit modest DCSB, about 20–30% larger than when the transverse vertex is completely neglected, as depicted in Figure 3.

The mass functions in Figure 4 clearly demonstrate that the inclusion of the transverse components in the DSE has a dramatic impact on dynamical mass generation. This was already realized long ago and calculations beyond the rainbow-ladder truncation commonly include phenomenological ansätze for several transverse components, in particular those proportional to ΔB(k2,p2)=B(k2)−B(p2)\Delta B(k^2,p^2) = B(k^2) - B(p^2). The novelty in this approach is that the present transverse quark-gluon vertex, derived from gauge identities and whose unknown tensor component is constrained with a vertex ansatz motivated by perturbation theory and multiplicative renormalizability [81], generates copious DCSB without the need to resort to a phenomenological gluon model.

On the other hand, the wave renormalization Z(p2)Z(p^2) tends to increase again in the infrared region below 1 GeV in case of the gluon and ghost propagators of set I and III. One may wonder whether the strength of the transverse vertex, whose effect flattens out Z(p2)Z(p^2) at lower momenta, is compatible with what model interactions in given truncation schemes predict. The overall strength of the quark-gluon interaction, however, is controlled by the strong coupling. In using what lattice QCD offers us for the gluon and ghost propagators renormalized at 4.3 GeV, we also work with a strong coupling expected at that renormalization scale. However, given that our transverse vertex has not been determined exactly and that our approach is not self-consistent—we do not solve the gluon and ghost DSE that couple to the quark-gap equation—we may consider variations of αs\alpha_{s}. We do so in Figure 5 where we vary the coupling αs(4.3 GeV)\alpha_{s}(4.3\ \mathrm{GeV}) in the range between 0.30 and 0.38 and concentrate merely on set III of gluon and ghost propagators. Plainly, an increasing coupling strength carries along two modifications: M(p2)M(p^{2}) is enhanced, which is expected, whereas Z(p2)Z(p^{2}) is increasingly suppressed and eventually levels out below p≈1p\approx1 GeV. The form factor X0(q2)X_{0}(q^{2}) is also enhanced as a function of αs\alpha_{s} and its peak value slightly shifts to larger momenta.

Figure 5. Functional dependence of M(p2)M(p^2), Z(p2)Z(p^2) and X0(q2)X_0(q^2) on the strong coupling αs\alpha_s. The DSE (15) is solved with the vertex Γμ(k,p)=ΓμL(k,p)+ΓμT(k,p)\Gamma_\mu(k,p) = \Gamma_\mu^{\mathrm{L}}(k,p) + \Gamma_\mu^{\mathrm{T}}(k,p) and the corresponding form factors in Eqs. (44)–(55); and with the gluon and ghost propagators of set III [30]. The current-quark mass is m(4.3 GeV)=25 MeVm(4.3\ \mathrm{GeV}) = 25\ \mathrm{MeV}.

The DSE can obviously also be solved for different quark flavors and we do this exercise in Figure 6, where we make use of the gluon and ghost propagators of set II; the results for the two other sets are qualitatively and quantitatively very similar. We employ the light, strange and charm current-quark masses renormalized at 2 GeV of Ref. [30]: mu(2 GeV)=md(2 GeV)=40 MeVm_u(2\ \mathrm{GeV}) = m_d(2\ \mathrm{GeV}) = 40\ \mathrm{MeV}, ms(2 GeV)=95 MeVm_s(2\ \mathrm{GeV}) = 95\ \mathrm{MeV}, mc(2 GeV)=1.51 GeVm_c(2\ \mathrm{GeV}) = 1.51\ \mathrm{GeV} evolved to the scale μ=4.3 GeV\mu= 4.3\ \mathrm{GeV}. For the bottom quark we choose mb(4.3 GeV)=Mb(4.3 GeV)=4.70 GeVm_b(4.3\ \mathrm{GeV}) = M_b(4.3\ \mathrm{GeV}) = 4.70\ \mathrm{GeV}, where Mb(p2)M_b(p^2) is the mass function for that flavor obtained with an interaction model [46] in rainbow truncation.

Figure 6. Flavor dependence of M(p2)M(p^{2}), Z(p2)Z(p^{2}) and X0(q2)X_{0}(q^{2}) obtained with the gluon- and ghost propagators of set II [35, 36]. The DSE (15) is solved with the vertex Γμ(k,p)=ΓμL(k,p)+ΓμT(k,p)\Gamma_{\mu}(k,p)=\Gamma_{\mu}^{L}(k,p)+\Gamma_{\mu}^{T}(k,p) and the associated form factors in Eqs. (44)–(55). The strong coupling is αs(μ)=0.3\alpha_{s}(\mu)=0.3 and the current quark masses are: mu,d(μ)=25.0m_{u,d}(\mu)=25.0 MeV, ms(μ)=82.1m_{s}(\mu)=82.1 MeV, mc(μ)=1.304m_{c}(\mu)=1.304 GeV and mb(μ)=4.697m_{b}(\mu)=4.697 GeV at the renormalization scale μ=4.3\mu=4.3 GeV. For comparison, we also plot the solutions in the chiral limit which define the quark propagators in Eqs. (104) and (105).

The pattern of the mass function for increasing current-quark masses mirrors the well-known behavior predicted by phenomenological gluon and vertex models, see Refs. [62, 64] for example. At p=0 GeVp = 0\ \mathrm{GeV} we find: Mu(0)=0.55 GeVM_u(0) = 0.55\ \mathrm{GeV}, Ms(0)=0.67 GeVM_s(0) = 0.67\ \mathrm{GeV}, Mc(0)=1.92 GeVM_c(0) = 1.92\ \mathrm{GeV} and Mb(0)=4.97 GeVM_b(0) = 4.97\ \mathrm{GeV}. Thus, even for the charm quark the impact of dynamical mass generation is important. The wave renormalization function flattens out in the infrared region for heavier quarks but remains above Z(0)≈0.93Z(0) \approx0.93 when Z(4.3 GeV)=0.98Z(4.3\ \mathrm{GeV}) = 0.98 is imposed. We also read from Figure 6 that X0(q2)X_0(q^2) is very sensitive to the current-quark mass and while the function is increasingly damped, its maximum value is considerably shifted to large momenta where it contributes little to DCSB in heavy quarks.

Solving the Dyson-Schwinger equation on the complex plane

So far, we have solved the DSE on the real spacelike axis which suffices to evaluate the magnitude of DCSB and the confining properties of the solutions. However, in solving the bound-state equation in Euclidean space [52] the quark propagators,

S(p±)=−iγ⋅p±σv(p±2)+σs(p±2),(101)S(p_\pm) = -i\gamma\cdot p_\pm\sigma_v(p_\pm^2) + \sigma_s(p_\pm^2), \tag*{(101)}

are evaluated at complex-valued momenta. This is because in Euclidean space their arguments, p+=p+ηPp_+ = p + \eta P and p−=p−ηˉPp_- = p - \bar{\eta}P, define parabolas on the complex plane,

p+2=p2−η2M2+2iηM∣p∣zp,(102)p_+^2 = p^2 - \eta^2 M^2 + 2i\eta M|p|z_p, \tag*{(102)}
p−2=p2−ηˉ2M2−2iηˉM∣p∣zp,(103)p_-^2 = p^2 - \bar{\eta}^2 M^2 - 2i\bar{\eta}M|p|z_p, \tag*{(103)}

where P=(0⃗,iM)P = (\vec{0},iM) is the meson’s rest-frame momentum, η+ηˉ=1\eta+ \bar{\eta} = 1 are the momentum partition parameters with η,ηˉ∈[0,1]\eta,\bar{\eta} \in[0,1], and zp=p⋅P/∣p∣∣P∣z_p = p\cdot P/|p||P| is the angle between the relative and total meson momenta for which −1≤zp≤+1-1 \le z_p \le+1 holds.

We use Cauchy’s integral theorem to solve the quark DSE inside this parabola. At first, we parametrize the contour of the parabola in the lower and upper half of the complex plane and find solutions on this contour via an iterative procedure using the same renormalization procedure as in Eqs. (97) and (98). Then, we apply this contour solution in Cauchy’s integral formula again to find solutions for σv(p±2)\sigma_v(p_\pm^2) and σs(p±2)\sigma_s(p_\pm^2) inside the parabola. A detailed discussion with examples of parametrizations can be found in Refs. [48, 132, 133].

The real and imaginary parts of the M(p2)M(p^2) and Z(p2)Z(p^2) functions are plotted in Figure 7 and Figure 8, respectively. The real part of M(p2)M(p^2) is a smooth and monotonically decreasing function of p2≡p+2p^2 \equiv p_+^2 in real direction, whereas it is all but constant along the imaginary axis Im⁡(p2)\operatorname{Im}(p^2). Likewise, the real part of Z(p2)Z(p^2) smoothly tends towards its perturbative limit both on and off the real axis. The imaginary part of both functions is characterized by complex-conjugate extrema near the origin of the parabola, though we note that their magnitude is considerably smaller than that of the imaginary parts of solutions with model propagators [43, 46, 135].

Real part and imaginary part of the complex mass function

Figure 7. Real part (upper panel) and imaginary part (lower panel) of the complex mass function M(p+2)M(p_+^2). The DSE is solved on the complex parabola with p+p_+ defined by Eq. (102), the external mass P2=−Mπ2P^2=-M_\pi^2 (Mπ=140M_\pi=140 MeV) and η=1/2\eta=1/2. We use the gluon and ghost propagators of set II [35, 36], the vertex Γμ(k,p)=ΓμL(k,p)+ΓμT(k,p)\Gamma_\mu(k,p)=\Gamma_\mu^{\mathrm{L}}(k,p)+\Gamma_\mu^{\mathrm{T}}(k,p) with the form factors in Eqs. (44)–(47) and (2.48)–(2.55), while αs(μ)=0.3\alpha_s(\mu)=0.3. Mass units are in GeV.

Real part (upper panel) and imaginary part (lower panel) of the complex wave-renormalization function $Z(p_{+}^{2})$

Figure 8. Real part (upper panel) and imaginary part (lower panel) of the complex wave-renormalization function Z(p+2)Z(p_{+}^{2}). Details of the DSE solution as in Figure 7.

Generally speaking, in the complex momentum range considered herein, the two solutions presented in Figure 7 and Figure 8 are qualitatively very similar to those obtained with phenomenological interaction models in rainbow-ladder truncation. There are, on the other hand, distinctive quantitative and analytical differences and we here refrain from a detailed comparison. We merely note that the convergence of the complex DSE using a Cauchy integral, with the present vertex and its inherent complexities, raises subtle issues that depend on the details of the contour parametrization and its size determined by the parabola cutoff on the complex plane.

Applications

While the numerical DSE solutions we obtain with the vertex defined by Eqs. (20) and (21) along with Eqs. (44) to (55) lead to typical constituent masses and characteristic mass functions for all flavors considered in Figure 6, only the calculation of a gauge-

independent observable can tell us more about how realistic they are. Naturally, the next step consists in the construction of an antiquark-quark Bethe-Salpeter kernel that is consistent with this new vertex and must satisfy the axialvector Ward-Takahashi identities. Given the complexity of the task, we postpone it to a future work.

However, even without the knowledge of the pion’s Bethe-Salpeter amplitude we may compute its weak decay constant, which is a measure of chiral symmetry breaking. Indeed, while the pion’s mass vanishes in chiral limit, fπf_{\pi} does not. We follow Ref. [134] where the weak decay constant in the chiral limit is expressed by the integral,

(fπ0)2=Nc8π2∫0∞dp2 p2B2(p2)(σv2−2[σsσs′+p2σvσv′−p2[σsσs′′−(σs′)2]−p4[σvσv′′−(σv′)2]])(104)\left(f_{\pi}^{0}\right)^{2} = \frac{N_{c}}{8\pi^{2}} \int_{0}^{\infty} \mathrm{d}p^{2}\,p^{2} B^{2}\left(p^{2}\right) \left(\sigma_{v}^{2} - 2\left[\sigma_{s}\sigma_{s}' + p^{2}\sigma_{v}\sigma_{v}' - p^{2}\left[\sigma_{s}\sigma_{s}'' - \left(\sigma_{s}'\right)^{2}\right] - p^{4}\left[\sigma_{v}\sigma_{v}'' - \left(\sigma_{v}'\right)^{2}\right]\right]\right) \tag*{(104)}

with σs,v′≡dσs,v(p2)/dp2\sigma_{s,v}' \equiv\mathrm{d}\sigma_{s,v}\left(p^{2}\right)/\mathrm{d}p^{2}. As we are constrained by a low renormalization point due to our use of quenched lattice-QCD input for the gluon and ghost propagators, determining Z2Z_{2} and Z4Z_{4} in the chiral limit is not straightforward. We set m=0m = 0 MeV at μ=4.3\mu= 4.3 GeV which allows for a sensible approximation in using σs(p2)\sigma_{s}\left(p^{2}\right) and σv(p2)\sigma_{v}\left(p^{2}\right) to calculate the decay constant in the chiral limit.

As another application, we consider the quark condensate which is an order parameter for DCSB. As for the pion decay constant, we do so using the three DSE solutions for the full vertex at hand, i.e. M(p2)M\left(p^{2}\right) and Z(p2)Z\left(p^{2}\right) obtained with the different ghost and gluon-dressing functions in the chiral limit. We thus calculate the integral over the trace of the quark propagator:

−⟨qˉq⟩≡Z4Nc∫Λd4k(2π)4tr⁡D[S(k)].(105)-\langle\bar{q}q\rangle\equiv Z_{4}N_{c}\int^{\Lambda}\frac{\mathrm{d}^{4}k}{(2\pi)^{4}}\operatorname{tr}_{D}\left[S(k)\right]. \tag*{(105)}

The values of the decay constant fπ0f_{\pi}^{0} and of the quark condensate are obtained for all three sets of gluon and ghost propagators and are summarized in Table 1. The three values for fπ0f_{\pi}^{0} are slightly above the experimental value, fπ±=92.2f_{\pi^{\pm}} = 92.2 MeV, while the quark condensates are in good agreement with other estimates, for instance the chiral condensate in the MS scheme using SU(2) chiral perturbation theory [136], (−⟨qˉq⟩)1/3=272(2)\left(-\langle\bar{q}q\rangle\right)^{1/3} = 272(2) MeV, or the light-quark condensate from lattice QCD [137]: (−⟨qˉq⟩)1/3=283(2)\left(-\langle\bar{q}q\rangle\right)^{1/3} = 283(2) MeV.

Set ISet IISet III
fπ0f_{\pi}^{0} [MeV]97.096.0298.67
(−⟨qˉq⟩)13\left(-\langle\bar{q}q\rangle\right)^{\frac{1}{3}} [MeV]251.16249.61255.80

Table 1. Weak decay constant in the chiral limit (104) and quark condensate (105) calculated with the solutions for M(p2)M(p^{2}) and Z(p2)Z(p^{2}) using the longitudinal (20) and transverse (21) vertex in the DSE (15) and the three sets of gluon and ghost propagators introduced in § 3.

Conclusive remarks and future developments

We have derived a novel form of the transverse quark-gluon vertex that complements the “longitudinal” components obtained as a ghost-corrected Ball-Chiu vertex in Refs. [86, 90, 92, 94] which saturates the STI [120, 121]. As for the non-transverse components, we were guided by symmetry transformations and multiplicative renormalizability encoded in the set of two TSTI [122] which couple the vector and axialvector vertices. By means of projections with two appropriate tensors one obtains an identity for each of these vertices, i.e. the TSTI have been decoupled.

Once this is realized, one can project out the eight transverse form factors for which we obtain the expressions in Eqs. (48) to (55). Notably, of the twelve form factors that describe the fully dressed quark-gluon vertex, only λ1(k,p)\lambda_{1}(k,p), λ2(k,p)\lambda_{2}(k,p), λ3(k,p)\lambda_{3}(k,p), τ3(k,p)\tau_{3}(k,p), τ5(k,p)\tau_5(k,p) and τ8(k,p)\tau_8(k,p) depend directly on the quark-ghost kernel and ghost-dressing function5. Of course, due to imposing the Bashir-Bermudez ansatz for the transverse vertex to constrain the scalar Yi(k,p)Y_i(k,p) functions, the latter are also functions of the quark-ghost kernel and ghost-dressing function. The last ingredients are the eight parameters, a⃗:={a1,a2,a3,a4,a5,a6,a7,a8}\vec{a}:=\{a_1,a_2,a_3,a_4,a_5,a_6,a_7,a_8\}, inherent to the Bashir-Bermudez vertex specified by Eqs. (66) to (73). We showed that they are far from being free parameters, as they are constrained by multiplicative renormalizability relations and limited to a rather narrow range outside of which no or very unsatisfying solutions of the DSE are found.

Nonetheless, while our dressed quark-gluon vertex in its present form is successful in producing the right amount of DCSB for hadron phenomenology, as our results for the weak decay constant of the pion and the quark condensate demonstrate, the reliance on parameters is unsatisfying and theoretically undesirable. Amongst future perspectives, we plan to obtain an integral expression for the four-point function, at the origin of the tensor elements Tμν1VμνT^1_{\mu\nu}V_{\mu\nu} and Tμν2VμνT^2_{\mu\nu}V_{\mu\nu}, along similar lines applied to the quark-ghost kernel in § 2.4. We remind that the nonlocal four-point function is related to the tensor VμνV_{\mu\nu} via the line integral,

Vμν=∫d4ℓ(2π)42ℓλϵλμνρΓρ(k,p;ℓ),(106)V_{\mu\nu}=\int\frac{\mathrm{d}^4\ell}{(2\pi)^4}2\ell^\lambda\epsilon_{\lambda\mu\nu\rho}\Gamma^\rho(k,p;\ell), \tag*{(106)}

where Γρ(k,p;ℓ)\Gamma_\rho(k,p;\ell) is the Fourier transform of the four-point function in coordinate space and which is defined in QED by [122]:

∫d4x d4x′ d4x1 d4x2 ei(k⋅x1−p⋅x2+(p−ℓ)⋅x−(k−ℓ)⋅x′)⟨0∣Tψˉ(x′)γρW(x′,x)ψ(x)ψ(x1)ψˉ(x2)∣0⟩=(2π)4δ4(k−p−q)S(k)Γρ(k,p;ℓ)S(p),(107)\int\mathrm{d}^4x\,\mathrm{d}^4x'\,\mathrm{d}^4x_1\,\mathrm{d}^4x_2\,e^{i(k\cdot x_1-p\cdot x_2+(p-\ell)\cdot x-(k-\ell)\cdot x')}\langle0|T\bar{\psi}(x')\gamma_\rho W(x',x)\psi(x)\psi(x_1)\bar{\psi}(x_2)|0\rangle = (2\pi)^4\delta^4(k-p-q)S(k)\Gamma_\rho(k,p;\ell)S(p), \tag*{(107)}

with q=(k−ℓ)−(p−ℓ)q=(k-\ell)-(p-\ell) and W(x′,x)W(x',x) is a Wilson line that ensures a gauge invariant expression. The expression in QCD is analogous, yet involves in addition color matrices and ghost fields. Therefore, expanding the Wilson line to leading order in the strong coupling gg [138], the matrix element can be expressed approximately by diagrams that describe the gluon dressing of quarks, the two-quark scattering and the interaction of a quark with a ghost via gluon exchange. However, in such an approach the propagators ought to be dressed as in the dressed approximation to Ha(k,p)H^a(k,p) discussed in § 2.4. With this ansatz, one can contract VμνV_{\mu\nu} with the tensors Tμν1T^1_{\mu\nu} and Tμν2T^2_{\mu\nu} to obtain the form factor decomposition of Eqs. (40) and (41) and subsequently project out all Yi(k,p)Y_i(k,p) functions. As a consequence, we arrive at a set of eight integral equations which are all coupled with each other as well as with the integral equations for A(p2)A(p^2), B(p2)B(p^2) and X0(q2)X_0(q^2) we considered in this work.

In future studies, the quark-ghost kernel we presented in § 2.4 ought to be calculated beyond the leading approximation that includes only X0(q2)X_0(q^2) and moreover neglects any angular dependence; see Refs. [92, 94] for detailed calculations. This is important as λ4(k,p)\lambda_4(k,p) remains otherwise zero and in the particular case of the soft-gluon limit, k=pk=p (q2=0q^2 = 0), the expressions for λ1(k,p)\lambda_1(k,p) (44) and λ2(k,p)\lambda_2(k,p) (45) hardly express any functional dependence without the full inclusion of the Ξi(k,p)\Xi_i(k,p) [97]. However, for our present purpose this simplified ansatz proves to be sufficient, as our aim was to demonstrate that an important amount of DCSB in the gap equation is due to the transverse quark-gluon vertex. We are therefore optimistic that with future refinements a complete vertex structure can be achieved that catches the dynamical subtleties much beyond the leading truncation, indispensable for the reproduction of the excited hadron spectrum and exotic states.

Acknowledgments

We kindly thank Orlando Oliveira for helpful comments about the refined Gribov-Zwanziger parametrization of the ghost- and gluon-dressing functions and for providing the corresponding lattice QCD data. B.E. acknowledges funding by FAPESP, grant no. 2018/20218-4, and by CNPq, grant no. 428003/2018-4. F.E.S. is supported by CAPES-PNPD grant no. 88882.314890/2013-01, R.C.S. by a CAPES PhD fellowship and L.A. by a FAPESP postdoctoral fellowship grant no. 2018/17643-5. E.R. acknowledges support from “Vicerrectoría de Investigaciones e Interacción Social VIIS de la Universidad de Nariño”, project numbers 1928 and 2172. This research was also partly supported by Coordinación de la Investigación Científica (CIC) of the University of Michoacan and CONACyT, Mexico, through grant nos. 4.10 and CB2014-22117, respectively. This work is part of the project “INCT-Física Nuclear e Aplicações”, no. 464898/2014-5.

References

  1. [1]S. Aoki et al. [Flavour Lattice Averaging Group], Eur. Phys. J. C 80, no.2, 113 (2020) doi:10.1140/epjc/s10052-019-7354-7DOI
  2. [2]Y. B. Yang, J. Liang, Y. J. Bi, Y. Chen, T. Draper, K. F. Liu and Z. Liu, Phys. Rev. Lett. 121, no.21, 212001 (2018) doi:10.1103/PhysRevLett.121.212001DOI
  3. [3]C. D. Roberts and A. G. Williams, Prog. Part. Nucl. Phys. 33, 477-575 (1994) doi:10.1016/0146-6410(94)90049-3DOI
  4. [4]A. Bashir, L. Chang, I. C. Cloët, B. El-Bennich, Y. X. Liu, C. D. Roberts and P. C. Tandy, Commun. Theor. Phys. 58, 79-134 (2012) doi:10.1088/0253-6102/58/1/16DOI
  5. [5]I. C. Cloët, G. Eichmann, B. El-Bennich, T. Klähn and C. D. Roberts, Few Body Syst. 46, 1-36 (2009) doi:10.1007/s00601-009-0015-xDOI
  6. [6]G. Eichmann, R. Alkofer, A. Krassnigg and D. Nicmorus, Phys. Rev. Lett. 104, 201601 (2010) doi:10.1103/PhysRevLett.104.201601DOI
  7. [7]I. G. Aznauryan, A. Bashir, V. Braun, S. J. Brodsky, V. D. Burkert, L. Chang, C. Chen, B. El-Bennich, I. C. Cloët and P. L. Cole, et al. Int. J. Mod. Phys. E 22, 1330015 (2013) doi:10.1142/S0218301313300154DOI
  8. [8]J. Segovia, B. El-Bennich, E. Rojas, I. C. Cloët, C. D. Roberts, S. S. Xu and H. S. Zong, Phys. Rev. Lett. 115, no.17, 171801 (2015) doi:10.1103/PhysRevLett.115.171801DOI
  9. [9]G. Eichmann, C. S. Fischer and H. Sanchis-Alepuz, Phys. Rev. D 94, no.9, 094033 (2016) doi:10.1103/PhysRevD.94.094033DOI
  10. [10]G. Eichmann, H. Sanchis-Alepuz, R. Williams, R. Alkofer and C. S. Fischer, Prog. Part. Nucl. Phys. 91, 1-100 (2016) doi:10.1016/j.ppnp.2016.07.001DOI
  11. [11]H. Sanchis-Alepuz and R. Williams, Comput. Phys. Commun. 232, 1-21 (2018) doi:10.1016/j.cpc.2018.05.020
  12. [12]C. Chen, B. El-Bennich, C. D. Roberts, S. M. Schmidt, J. Segovia and S. Wan, Phys. Rev. D 97, no.3, 034016 (2018) doi:10.1103/PhysRevD.97.034016DOI
  13. [13]C. Chen, Y. Lu, D. Bin​​osi, C. D. Roberts, J. Rodríguez-Quintero and J. Segovia, Phys. Rev. D 99, no.3, 034013 (2019) doi:10.1103/PhysRevD.99.034013DOI
  14. [14]K. D. Bednar, I. C. Cloët and P. C. Tandy, Phys. Lett. B 782, 675-681 (2018) doi:10.1016/j.physletb.2018.06.020DOI
  15. [15]V. V. Flambaum, A. Holl, P. Jaikumar, C. D. Roberts and S. V. Wright, Few Body Syst. 38, 31-51 (2006) doi:10.1007/s00601-005-0123-1DOI
  16. [16]J. S. Ball and T. W. Chiu, Phys. Rev. D 22, 2542 (1980) doi:10.1103/PhysRevD.22.2542DOI
  17. [17]A. I. Davydychev, P. Osland and L. Saks, Phys. Rev. D 63, 014022 (2001) doi:10.1103/PhysRevD.63.014022DOI
  18. [18]C. S. Fischer, A. Maas and J. M. Pawlowski, Annals Phys. 324, 2408-2437 (2009) doi:10.1016/j.aop.2009.07.009DOI
  19. [19]R. Alkofer, M. Q. Huber and K. Schwenzer, Phys. Rev. D 81, 105010 (2010) doi:10.1103/PhysRevD.81.105010DOI
  20. [20]D. Dudal, J. A. Gracey, S. P. Sorella, N. Vandersickel and H. Verschelde, Phys. Rev. D 78, 065047 (2008) doi:10.1103/PhysRevD.78.065047DOI
  21. [21]A. C. Aguilar and A. A. Natale, JHEP 08, 057 (2004) doi:10.1088/1126-6708/2004/08/057DOI
  22. [22]A. C. Aguilar, D. Binosi and J. Papavassiliou, Phys. Rev. D 78, 025010 (2008) doi:10.1103/PhysRevD.78.025010DOI
  23. [23]A. C. Aguilar, D. Binosi and J. Papavassiliou, Phys. Rev. D 86, 014032 (2012) doi:10.1103/PhysRevD.86.014032DOI
  24. [24]A. Cucchieri and T. Mendes, PoS LATTICE2007, 297 (2007) doi:10.22323/1.042.0297
  25. [25]A. Cucchieri and T. Mendes, Phys. Rev. Lett. 100, 241601 (2008) doi:10.1103/PhysRevLett.100.241601DOI
  26. [26]O. Oliveira and P. J. Silva, Phys. Rev. D 79, 031501 (2009) doi:10.1103/PhysRevD.79.031501DOI
  27. [27]M. R. Pennington and D. J. Wilson, Phys. Rev. D 84, 119901 (2011) doi:10.1103/PhysRevD.84.094028DOI
  28. [28]O. Oliveira and P. J. Silva, Phys. Rev. D 86, 114513 (2012) doi:10.1103/PhysRevD.86.114513DOI
  29. [29]I. L. Bogolubsky, E. M. Ilgenfritz, M. Müller-Preussker and A. Sternbeck, Phys. Lett. B 676, 69-73 (2009) doi:10.1016/j.physletb.2009.04.076DOI
  30. [30]A. Ayala, A. Bashir, D. Binosi, M. Cristoforetti and J. Rodríguez-Quintero, Phys. Rev. D 86, 074512 (2012) doi:10.1103/PhysRevD.86.074512DOI
  31. [31]S. Strauss, C. S. Fischer and C. Kellermann, Phys. Rev. Lett. 109, 252001 (2012) doi:10.1103/PhysRevLett.109.252001DOI
  32. [32]A. K. Cyrol, L. Fister, M. Mitter, J. M. Pawlowski and N. Strothoff, Phys. Rev. D 94, no.5, 054005 (2016) doi:10.1103/PhysRevD.94.054005DOI
  33. [33]P. Boucaud, F. De Soto, K. Raya, J. Rodriguez-Quintero and S. Zafeiropoulos, Phys. Rev. D 98, no.11, 114515 (2018) doi:10.1103/PhysRevD.98.114515DOI
  34. [34]B. W. Mintz, L. F. Palhares, G. Peruzzo and S. P. Sorella, Phys. Rev. D 99, no.3, 034002 (2019) doi:10.1103/PhysRevD.99.034002DOI
  35. [35]D. Dudal, O. Oliveira and P. J. Silva, Annals Phys. 397, 351-364 (2018) doi:10.1016/j.aop.2018.08.019DOI
  36. [36]A. G. Duarte, O. Oliveira and P. J. Silva, Phys. Rev. D 94 (2016) no.1, 014502 doi:10.1103/PhysRevD.94.014502DOI
  37. [37]A. C. Aguilar, F. De Soto, M. N. Ferreira, J. Papavassiliou, J. Rodríguez-Quintero and S. Zafeiropoulos, Eur. Phys. J. C 80, no.2, 154 (2020) doi:10.1140/epjc/s10052-020-7741-0DOI
  38. [38]M. Q. Huber, Phys. Rev. D 91, no.8, 085018 (2015) doi:10.1103/PhysRevD.91.085018DOI
  39. [39]M. Q. Huber, Phys. Rev. D 101, 114009 (2020) doi:10.1103/PhysRevD.101.114009DOI
  40. [40]C. S. Fischer and M. Q. Huber, Phys. Rev. D 102, no.9, 094005 (2020) doi:10.1103/PhysRevD.102.094005DOI
  41. [41]A. F. Falcão, O. Oliveira and P. J. Silva, Phys. Rev. D 102, no.11, 114518 (2020) doi:10.1103/PhysRevD.102.114518DOI
  42. [42]P. Maris and C. D. Roberts, Phys. Rev. C 56, 3369 (1997) doi:10.1103/PhysRevC.56.3369DOI
  43. [43]P. Maris, C. D. Roberts and P. C. Tandy, Phys. Lett. B 420, 267 (1998) doi:10.1016/S0370-2693(97)01535-9DOI
  44. [44]P. Maris and P. C. Tandy, Phys. Rev. C 60, 055214 (1999) doi:10.1103/PhysRevC.60.055214DOI
  45. [45]R. Alkofer, P. Watson and H. Weigel, Phys. Rev. D 65 (2002), 094026 doi:10.1103/PhysRevD.65.094026
  46. [46]S. x. Qin, L. Chang, Y. x. Liu, C. D. Roberts and D. J. Wilson, Phys. Rev. C 84, 042202 (2011) doi:10.1103/PhysRevC.84.042202DOI
  47. [47]L. Chang and C. D. Roberts, Phys. Rev. C 85, 052201 (2012) doi:10.1103/PhysRevC.85.052201DOI
  48. [48]E. Rojas, B. El-Bennich and J. P. B. C. de Melo, Phys. Rev. D 90, 074025 (2014) doi:10.1103/PhysRevD.90.074025DOI
  49. [49]K. Raya, L. Chang, A. Bashir, J. J. Cobos-Martínez, L. X. Gutiérrez-Guerrero, C. D. Roberts and P. C. Tandy, Phys. Rev. D 93, no.7, 074017 (2016) doi:10.1103/PhysRevD.93.074017DOI
  50. [50]B. El-Bennich, G. Krein, E. Rojas and F. E. Serna, Few Body Syst. 57, no.10, 955-963 (2016) doi:10.1007/s00601-016-1133-x
  51. [51]F. F. Mojica, C. E. Vera, E. Rojas and B. El-Bennich, Phys. Rev. D 96, no.1, 014012 (2017) doi:10.1103/PhysRevD.96.014012DOI
  52. [52]F. E. Serna, B. El-Bennich and G. Krein, Phys. Rev. D 96 (2017) no.1, 014013 doi:10.1103/PhysRevD.96.014013DOI
  53. [53]M. A. Bedolla, J. J. Cobos-Martínez and A. Bashir, Phys. Rev. D 92 (2015) no.5, 054031 doi:10.1103/PhysRevD.92.054031DOI
  54. [54]M. A. Bedolla, K. Raya, J. J. Cobos-Martínez and A. Bashir, Phys. Rev. D 93 (2016) no.9, 094025 doi:10.1103/PhysRevD.93.094025DOI
  55. [55]K. Raya, M. A. Bedolla, J. J. Cobos-Martínez and A. Bashir, Few Body Syst. 59 (2018) no.6, 133 doi:10.1007/s00601-018-1455-yDOI
  56. [56]C. S. Fischer, S. Kubrak and R. Williams, Eur. Phys. J. A 51 (2015), 10 doi:10.1140/epja/i2015-15010-7DOI
  57. [57]T. Hilger, M. Gómez-Rocha, A. Krassnigg and W. Lucha, Eur. Phys. J. A 53 (2017) no.10, 213 doi:10.1140/epja/i2017-12384-4DOI
  58. [58]P. J. Gunkel, C. S. Fischer and P. Isserstedt, Eur. Phys. J. A 55, no.9, 169 (2019) doi:10.1140/epja/i2019-12868-1DOI
  59. [59]B. El-Bennich, J. P. B. C. de Melo, B. Loiseau, J. P. Dedonder and T. Frederico, Braz. J. Phys. 38 (2008), 465-471 doi:10.1590/S0103-97332008000400016DOI
  60. [60]B. El-Bennich, M. A. Ivanov and C. D. Roberts, Nucl. Phys. B Proc. Suppl. 199 (2010), 184-190 doi:10.1016/j.nuclphysbps.2010.02.026DOI
  61. [61]B. El-Bennich, EPJ Web Conf. 172, 02005 (2018) doi:10.1051/epjconf/201817202005
  62. [62]B. El-Bennich, C. D. Roberts and M. A. Ivanov, PoS QCD-TNT-II, 018 (2012) doi:10.22323/1.136.0018
  63. [63]M. Atif Sultan, K. Raya, F. Akram, A. Bashir and B. Masud, Phys. Rev. D 103 (2021) no.5, 054036 doi:10.1103/PhysRevD.103.054036DOI
  64. [64]F. E. Serna, C. Chen and B. El-Bennich, Phys. Rev. D 99, no.9, 094027 (2019) doi:10.1103/PhysRevD.99.094027DOI
  65. [65]F. E. Serna, R. C. da Silveira, J. J. Cobos-Martínez, B. El-Bennich and E. Rojas, Eur. Phys. J. C 80 (2020) no.10, 955 doi:10.1140/epjc/s10052-020-08517-3DOI
  66. [66]M. Gómez-Rocha, T. Hilger and A. Krassnigg, Few Body Syst. 56 (2015) no.6-9, 475-480 doi:10.1007/s00601-014-0938-8
  67. [67]M. Gómez-Rocha, T. Hilger and A. Krassnigg, Phys. Rev. D 92 (2015) no.5, 054030 doi:10.1103/PhysRevD.92.054030
  68. [68]M. Gómez-Rocha, T. Hilger and A. Krassnigg, Phys. Rev. D 93 (2016) no.7, 074010 doi:10.1103/PhysRevD.93.074010
  69. [69]C. S. Fischer and R. Alkofer, Phys. Rev. D 67, 094020 (2003) doi:10.1103/PhysRevD.67.094020DOI
  70. [70]R. Alkofer, C. S. Fischer, F. J. Llanes-Estrada and K. Schwenzer, Annals Phys. 324, 106-172 (2009) doi:10.1016/j.aop.2008.07.001DOI
  71. [71]M. Hopfer, A. Windisch and R. Alkofer, PoS ConfinementX, 073 (2012) doi:10.22323/1.171.0073DOI
  72. [72]R. Williams, Eur. Phys. J. A 51, no.5, 57 (2015) doi:10.1140/epja/i2015-15057-4DOI
  73. [73]R. Williams, C. S. Fischer and W. Heupel, Phys. Rev. D 93, no.3, 034026 (2016) doi:10.1103/PhysRevD.93.034026DOI
  74. [74]D. C. Curtis and M. R. Pennington, Phys. Rev. D 42 (1990), 4165-4169 doi:10.1103/PhysRevD.42.4165DOI
  75. [75]A. Bashir and M. R. Pennington, Phys. Rev. D 50 (1994), 7679-7689 doi:10.1103/PhysRevD.50.7679DOI
  76. [76]A. Bashir and M. R. Pennington, Phys. Rev. D 53 (1996), 4694-4697 doi:10.1103/PhysRevD.53.4694DOI
  77. [77]A. Bashir and A. Raya, Nucl. Phys. B 709 (2005), 307-328 doi:10.1016/j.nuclphysb.2004.12.010DOI
  78. [78]A. Bashir and A. Raya, [arXiv:hep-ph/0411310 [hep-ph]].DOI
  79. [79]A. Kızılersü and M. R. Pennington, Phys. Rev. D 79, 125020 (2009) doi:10.1103/PhysRevD.79.125020DOI
  80. [80]A. Bashir, A. Raya and S. Sánchez-Madrigal, Phys. Rev. D 84, 036013 (2011) doi:10.1103/PhysRevD.84.036013DOI
  81. [81]A. Bashir, R. Bermudez, L. Chang and C. D. Roberts, Phys. Rev. C 85, 045205 (2012) doi:10.1103/PhysRevC.85.045205DOI
  82. [82]M. J. Aslam, A. Bashir and L. X. Gutiérrez-Guerrero, Phys. Rev. D 93, no.7, 076001 (2016) doi:10.1103/PhysRevD.93.076001DOI
  83. [83]L. A. Fernández-Rangel, A. Bashir, L. X. Gutiérrez-Guerrero and Y. Concha-Sanchez, Phys. Rev. D 93, no.6, 065022 (2016) doi:10.1103/PhysRevD.93.065022DOI
  84. [84]R. Bermudez, L. Albino, L. X. Gutiérrez-Guerrero, M. E. Tejeda-Yeomans and A. Bashir, Phys. Rev. D 95, no.3, 034041 (2017) doi:10.1103/PhysRevD.95.034041DOI
  85. [85]L. Albino, A. Bashir, L. X. Gutiérrez-Guerrero, B. El-Bennich and E. Rojas, Phys. Rev. D 100, no.5, 054028 (2019) doi:10.1103/PhysRevD.100.054028DOI
  86. [86]E. Rojas, J. P. B. C. de Melo, B. El-Bennich, O. Oliveira and T. Frederico, JHEP 10, 193 (2013) doi:10.1007/JHEP10(2013)193DOI
  87. [87]E. Rojas, B. El-Bennich, J. P. B. C. De Melo and M. A. Paracha, Few Body Syst. 56, no. 6-9, 639 (2015) doi:10.1007/s00601-015-1020-xDOI
  88. [88]S. X. Qin, L. Chang, Y. X. Liu, C. D. Roberts and S. M. Schmidt, Phys. Lett. B 722, 384-388 (2013) doi:10.1016/j.physletb.2013.04.034DOI
  89. [89]D. Binosi, L. Chang, J. Papavassiliou, S. X. Qin and C. D. Roberts, Phys. Rev. D 95, no. 3, 031501 (2017). doi:10.1103/PhysRevD.95.031501DOI
  90. [90]A. C. Aguilar and J. Papavassiliou, Phys. Rev. D 83, 014013 (2011) doi:10.1103/PhysRevD.83.014013DOI
  91. [91]A. C. Aguilar, D. Binosi, D. Ibañez and J. Papavassiliou, Phys. Rev. D 90, no.6, 065027 (2014) doi:10.1103/PhysRevD.90.065027DOI
  92. [92]A. C. Aguilar, J. C. Cardona, M. N. Ferreira and J. Papavassiliou, Phys. Rev. D 96, no.1, 014029 (2017) doi:10.1103/PhysRevD.96.014029DOI
  93. [93]A. C. Aguilar, J. C. Cardona, M. N. Ferreira and J. Papavassiliou, Phys. Rev. D 98, no.1, 014002 (2018) doi:10.1103/PhysRevD.98.014002DOI
  94. [94]A. C. Aguilar, M. N. Ferreira, C. T. Figueiredo and J. Papavassiliou, Phys. Rev. D 99, no.3, 034026 (2019) doi:10.1103/PhysRevD.99.034026DOI
  95. [95]S. Jia and M. R. Pennington, Phys. Rev. D 94, no.11, 116004 (2016) doi:10.1103/PhysRevD.94.116004DOI
  96. [96]O. Oliveira, W. de Paula, T. Frederico and J. P. B. C. de Melo, Eur. Phys. J. C 79, no.2, 116 (2019) doi:10.1140/epjc/s10052-019-6617-7DOI
  97. [97]O. Oliveira, T. Frederico and W. de Paula, Eur. Phys. J. C 80, no.5, 484 (2020) doi:10.1140/epjc/s10052-020-8037-0DOI
  98. [98]A. Bashir, A. Kızılersü and M. R. Pennington, Phys. Rev. D 57, 1242-1249 (1998) doi:10.1103/PhysRevD.57.1242DOI
  99. [99]A. Bashir, A. Kızılersü and M. R. Pennington, [arXiv:hep-ph/9907418 [hep-ph]].DOI
  100. [100]A. Bashir, A. Kızılersü and M. R. Pennington, Phys. Rev. D 62, 085002 (2000) doi:10.1103/PhysRevD.62.085002DOI
  101. [101]M. Peláez, M. Tissier and N. Wschebor, Phys. Rev. D 92, no.4, 045012 (2015) doi:10.1103/PhysRevD.92.045012DOI
  102. [102]J. Skullerud and A. Kızılersü, JHEP 09, 013 (2002) doi:10.1088/1126-6708/2002/09/013DOI
  103. [103]J. I. Skullerud, P. O. Bowman, A. Kızılersü, D. B. Leinweber and A. G. Williams, JHEP 04, 047 (2003) doi:10.1088/1126-6708/2003/04/047DOI
  104. [104]A. Kızılersü, D. B. Leinweber, J. I. Skullerud and A. G. Williams, Eur. Phys. J. C 50, 871-875 (2007) doi:10.1140/epjc/s10052-007-0250-6DOI
  105. [105]O. Oliveira, A. Kızılersü, P. J. Silva, J. I. Skullerud, A. Sternbeck and A. G. Williams, Acta Phys. Polon. Supp. 9, 363-368 (2016) doi:10.5506/APhysPolBSupp.9.363DOI
  106. [106]A. Sternbeck, P. H. Balduf, A. Kızılersü, O. Oliveira, P. J. Silva, J. I. Skullerud and A. G. Williams, PoS LATTICE2016, 349 (2017) doi:10.22323/1.256.0349DOI
  107. [107]A. Kızılersü, O. Oliveira, P. J. Silva, J. I. Skullerud and A. Sternbeck, Phys. Rev. D 103, no.11, 114515 (2021) doi:10.1103/PhysRevD.103.114515DOI
  108. [108]J. C. Ward, Phys. Rev. 78, 182 (1950) doi:10.1103/PhysRev.78.182DOI
  109. [109]E. S. Fradkin, Zh. Eksp. Teor. Fiz. 29, 258-261 (1955)
  110. [110]H. S. Green, Proc. Phys. Soc. A 66, 873-880 (1953) doi:10.1088/0370-1298/66/10/303DOI
  111. [111]Y. Takahashi, Nuovo Cim. 6, 371 (1957) doi:10.1007/BF02832514DOI
  112. [112]Y. Takahashi, Print-85-0421 (ALBERTA).
  113. [113]K. I. Kondo, Int. J. Mod. Phys. A 12, 5651-5686 (1997) doi:10.1142/S0217751X97002978DOI
  114. [114]H. X. He, F. C. Khanna and Y. Takahashi, Phys. Lett. B 480, 222-228 (2000) doi:10.1016/S0370-2693(00)00353-1DOI
  115. [115]H. X. He, Commun. Theor. Phys. 46, 109-112 (2006) doi:10.1088/0253-6102/46/1/025DOI
  116. [116]H. X. He, Int. J. Mod. Phys. A 22, 2119-2132 (2007) doi:10.1142/S0217751X07036257DOI
  117. [117]M. R. Pennington and R. Williams, J. Phys. G 32, 2219-2234 (2006) doi:10.1088/0954-3899/32/11/014DOI
  118. [118]H. x. He and F. C. Khanna, Int. J. Mod. Phys. A 21, 2541-2551 (2006) doi:10.1142/S0217751X06029533DOI
  119. [119]L. D. Landau and I. M. Khalatnikov, Sov. Phys. JETP 2, 69 (1956)
  120. [120]A. A. Slavnov, Theor. Math. Phys. 10, 99 (1972) [Teor. Mat. Fiz. 10, 153 (1972)]. doi:10.1007/BF01090719DOI
  121. [121]J. C. Taylor, Nucl. Phys. B 33, 436 (1971) doi:10.1016/0550-3213(71)90297-5DOI
  122. [122]H. x. He, Phys. Rev. D 80, 016004 (2009) doi:10.1103/PhysRevD.80.016004.DOI
  123. [123]T. De Meerleer, D. Dudal, S. P. Sorella, P. Dall’Olio and A. Bashir, Phys. Rev. D 97 (2018) no.7, 074017 doi:10.1103/PhysRevD.97.074017DOI
  124. [124]T. De Meerleer, D. Dudal, S. P. Sorella, P. Dall’Olio and A. Bashir, Phys. Rev. D 101 (2020) no.8, 085005 doi:10.1103/PhysRevD.101.085005DOI
  125. [125]A. Kizilersu, M. Reenders and M. R. Pennington, Phys. Rev. D 52 (1995), 1242-1259 doi:10.1103/PhysRevD.52.1242DOI
  126. [126]C. Itzykson and J. B. Zuber, “Quantum Field Theory”, New York, USA: Mcgraw-Hill (1980) 705 p. (International Series In Pure and Applied Physics).
  127. [127]W. J. Marciano and H. Pagels, Phys. Rept. 36 (1978), 137 doi:10.1016/0370-1573(78)90208-9DOI
  128. [128]P. Boucaud, F. De Soto, J. P. Leroy, A. Le Yaouanc, J. Micheli, O. Pene and J. Rodríguez-Quintero, Phys. Rev. D 79, 014508 (2009) doi:10.1103/PhysRevD.79.014508DOI
  129. [129]A. C. Aguilar, D. Binosi and J. Papavassiliou, JHEP 07, 002 (2010) doi:10.1007/JHEP07(2010)002DOI
  130. [130]D. Dudal, O. Oliveira and J. Rodríguez-Quintero, Phys. Rev. D 86 (2012), 105005 doi:10.1103/PhysRevD.86.105005DOI
  131. [131]J. Rodríguez-Quintero, private communication.
  132. [132]C. S. Fischer, P. Watson and W. Cassing, Phys. Rev. D 72 (2005), 094025 doi:10.1103/PhysRevD.72.094025DOI
  133. [133]A. Krassnigg, PoS CONFINEMENT8 (2008), 075 doi:10.22323/1.077.0075
  134. [134]C. D. Roberts, Nucl. Phys. A 605, 475-495 (1996) doi:10.1016/0375-9474(96)00174-1DOI
  135. [135]B. El-Bennich, G. Ramos-Zambrano and E. Rojas, Phys. Rev. D 103, no.7, 076008 (2021) doi:10.1103/PhysRevD.103.076008
  136. [136]S. Borsanyi, S. Durr, Z. Fodor, S. Krieg, A. Schäfer, E. E. Scholz and K. K. Szabo, Phys. Rev. D 88, 014513 (2013) doi:10.1103/PhysRevD.88.014513DOI
  137. [137]C. McNeile, A. Bazavov, C. T. H. Davies, R. J. Dowdall, K. Hornbostel, G. P. Lepage and H. D. Trottier, Phys. Rev. D 87, no.3, 034503 (2013) doi:10.1103/PhysRevD.87.034503DOI
  138. [138]C. S. R. Costa, A. Freese, I. C. Cloët, B. El-Bennich, G. Krein and P. C. Tandy, Phys. Rev. C 104, no.4, 045201 (2021) doi:10.1103/PhysRevC.104.045201DOI

Paper details

vlnc:W_sydr53q6
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents