PLB140302140302S0370-2693(26)00156-510.1016/j.physletb.2026.140302The AuthorsLetterTowards a unified viewpoint of Gribov–Zwanziger and Serreau–Tissier gauge fixingTowards a unified viewpoint of Gribov–Zwanziger and Serreau–Tissier gauge fixingRodrigoCarmo Terinrodrigo.carmo@urjc.esorganization=King Juan Carlos University, Faculty of Experimental Sciences and Technology, Department of Applied Physics, addressline=Av. del Alcalde de Móstoles, city=Madrid, postcode=28933, country=Spainorganization=King Juan Carlos University, Faculty of Experimental Sciences and Technology, Department of Applied Physics, addressline=Av. del Alcalde de Móstoles, city=Madrid, postcode=28933, country=Spainorganization=King Juan Carlos University, Faculty of Experimental Sciences and Technology, Department of Applied Physics, addressline=Av. del Alcalde de Móstoles, city=Madrid, postcode=28933, country=SpainEditor: Neil LambertAbstractWe investigate a unified Landau–gauge fixing that continuously interpolates between the viewpoints of the Serreau–Tissier (ST) copy-averaged formulation and the (Refined) Gribov–Zwanziger (RGZ) restriction to the first Gribov region. By coupling a copy-averaged ST sector and a BRST-invariant RGZ sector within a single local action; the horizon term acts uniformly on copies, and the ST weight controls their relative contributions, we obtain a single, local, BRST-invariant, power-counting renormalizable action. Algebraic renormalization shows that all counterterms are reabsorbed by a common set of field and parameter renormalizations, therefore the unification is algebraic rather than merely additive. The replica sector yields a radiatively generated gluon screening mass, while the RGZ parameters are fixed by the horizon and condensate gap equations; we also give infrared matching conditions that link both descriptions at small momentum. We present a compact BRST-superspace rewriting of the RGZ block and a simple hybrid superspace that hosts the ST replicas and RGZ side by side; these add no dynamics and organize the Ward-identity analysis. The resulting gluon propagator interpolates among the massive Faddeev–Popov–ST and the RGZ decoupling forms. This framework offers a controlled way to study how infrared Yang–Mills correlators depend on the balance between copy averaging and horizon suppression, and it suggests practical lattice tests through tunable copy weighting.KeywordsGribov copiesLandau gaugeGribov–ZwanzigerSerreau–Tissier gaugeBRST symmetryAlgebraic renormalizationData availabilityNo data was used for the research described in the article.1IntroductionQuantizing nonabelian Yang–Mills (YM) theories by means of the standard Faddeev-Popov (FP) prescription is remarkably successful in the ultraviolet (UV) but faces conceptual and practical obstacles in the infrared (IR) [1,2]. In particular, the Landau pole spoils naive perturbation theory and the gauge-fixing condition does not fully eliminate the gauge freedom because of the existence of Gribov copies, which are distinct field configurations related by gauge transformations that satisfy the same gauge condition [3,4]. The Gribov problem is intrinsic to the nontrivial geometry of gauge orbits and not to a specific gauge choice; see, for instance, the review [5].Two complementary continuum strategies have emerged to tame Gribov copies and capture YM IR dynamics. The Gribov-Zwanziger (GZ) framework restricts the functional integral to the first Gribov region Ω, where the FP operator Mab=−∂μDμab(A) is positive definite. This restriction is implemented through the (nonlocal) horizon functional H(Ah) written in terms of the transverse, gauge-field Aμh, and can be localized with the Zwanziger auxiliary fields (φ¯,φ) and (ω¯,ω) [6,7]. The RGZ extension includes dynamical dimension-two condensates, introducing mass scales (m, M) and leading to a local, renormalizable action whose gluon and ghost correlators match lattice data from the deep IR up to several GeV [8–10]. In the Ah formulation, an exact all-order BRST symmetry is retained in linear covariant gauges (LCG), which guarantees, e.g., α-independence of gauge-invariant observables and nonrenormalization of the longitudinal sector [11–13].Independently, J. Serreau and M. Tissier and further work collaborators proposed a family of Landau gauges free of Gribov ambiguities, defined by an explicit weighted average over copies [14–17]. In practice, one averages an operator O over all extrema Ui of the Landau functional f[A,U]=∫ddxtr(AμU)2 with a nonuniform weightP[A,U]=det(F[A,U]+ζ1)|detF[A,U]|e−βf[A,U],where F is the Landau-gauge FP Hessian, and β, ζ are gauge-fixing parameters of mass dimension two. A gauge-invariant observable is unaffected by the procedure thanks to the normalization by ∑iP[A,Ui]. Using the replica trick and a superspace nonlinear sigma-model representation, one obtains a local, perturbatively renormalizable field theory in d=4 which, for correlators of A and ghosts, reduces to a massive FP theory with a gluon mass term  ∼ β and a ghost mass  ∼ ζ [14,17]. The replica sector exhibits a phenomenon of radiative symmetry restoration (akin to the d=2 NLσ model), which selects a phase where a nonzero screening gluon mass survives in the replica limit n → 0 [17]. Crucially, within a range of renormalized parameters, the one-loop renormalization-group (RG) flow shows no Landau pole in the IR [14].These approaches share the goal of curing Gribov issues and at the same time providing an IR-safe perturbative description of YM. GZ/RGZ implements a hard restriction to Ω by means of the horizon functional (and dynamical condensates), whereas ST implements a soft lifting of copy degeneracy through a tunable weighted average, realized locally by means of replica nonlinear sigma-model (NLσ) fields. Both are local and renormalizable, and reproduce lattice-inspired decoupling-type gluon propagators [18]; see also the spectral and positivity analyses consistent with gluon confinement [19–21]. It is then natural to ask whether a more general gauge-fixing exists that interpolates between them.Therefore, in this work, we construct a unified gauge-fixing that treats ST and RGZ as limiting cases. At the level of the copy average, we couple a copy-averaged ST sector and a BRST-invariant RGZ sector; the horizon term acts uniformly over copies, while the ST weight controls their relative contributions,〈〈O[A]〉〉β,ζ,γ=∑is(i)O[AUi]det(F[A,Ui]+ζ1)|detF[A,Ui]|e−βf[A,Ui]−γ4H(Ah,Ui)∑is(i)det(F[A,Ui]+ζ1)|detF[A,Ui]|e−βf[A,Ui]−γ4H(Ah,Ui),so that γ=0 reproduces ST, while (β → ∞, ζ → 0) yields RGZ. We then derive a local action by mixing the ST replica/superspace localization with the Zwanziger localization in the Ah formulation [7,11–14,17]. The resulting theory is local and power-counting renormalizable, enjoys the exact BRST of the Ah sector, and continuously interpolates the structure of the gluon two-point function between a massive-FP (ST) form and the RGZ decoupling form. In this sense, our unified construction prepares a controlled framework to study how infrared YM correlators depend on the balance between (i) copy averaging and radiative restoration in the replica sector (set by β, ζ) and (ii) horizon suppression and condensates (set by γ, m, M). It connects smoothly with functional approaches (Dyson-Schwinger equations (DSEs) and the functional renormalization group (FRG)) [22–25] and with massive extensions of Faddeev–Popov theory [26,27].This paper is organized as follows. Section 2 reviews the ST and RGZ formalisms, emphasizing their gauge-fixing mechanisms and dynamical mass generation. Section 3 constructs the unified local action that interpolates between them through the replica trick and the BRST-invariant RGZ formulation. Section 4 presents the algebraic proof of renormalizability based on Slavnov–Taylor, antighost, and ghost identities, while Section 5 analyzes the coupled gap equations and their matching on the two sectors. A compact superspace formulation unifying both sectors is investigated in Section 6, and its complete derivation is collected in Appendix D. Section 7 summarizes the main results and discusses physical implications and possible extensions. Appendices A–C compile technical material, including field dimensions and ghost numbers, the linearized Slavnov operator and auxiliary Ward identities, non-renormalization relations, replica one-loop corrections, small-momentum expansions, tree-level Feynman rules, the Nielsen identity, and dimensional-regularization conventions. Finally, the complete algebraic construction of the unified BRST transformations and the corresponding gauge-fixing functional, presented for pure Yang–Mills theory, is provided in Appendix E.2Background: continuum approaches to the Gribov problemIn this section we summarize the two main continuum approaches that address the Gribov ambiguity in nonabelian gauge theories. Both share the goal of providing a consistent IR formulation of YM theories in the Landau gauge while preserving gauge invariance of physical observables. The first approach, ST, resolves the ambiguity by averaging over Gribov copies with a weighted functional measure that can be represented as a local field theory involving NLσ superfields. The second, known as the RGZ framework, restricts the domain of the functional integral to the first Gribov region through the introduction of the horizon functional and its localization through auxiliary fields. We review their main ingredients below.2.1The Serreau–Tissier gauge and radiative symmetry restorationFollowing refs. [14,15,17], the Landau gauge condition(1)∂μAμU=0,with AμU=UAμU†+igU∂μU†, can be equivalently formulated as the stationarity condition of the functional(2)f[A,U]=∫ddxtr[(AμU)2],whose extrema Ui correspond to Gribov copies. Instead of selecting one specific extremum, Serreau and Tissier proposed to perform a weighted average over all copies, defining the expectation value of an operator O as(3)〈〈O[A]〉〉=∑iO[AUi]P[A,Ui]∑iP[A,Ui],where the weight function is chosen as(4)P[A,U]=det(F[A,U]+ζ1)|detF[A,U]|e−βf[A,U],with F[A,U] the Faddeev-Popov (FP) operator in the Landau gauge, Fab[A;x,y]=−∂μ[∂μδab+gfacbAμc(x)]δ(d)(x−y), and β, ζ two gauge-fixing parameters of mass dimension two.Gauge-invariant observables are unaffected by this procedure due to the normalization in Eq. (4). The functional average can be reformulated as a local and perturbatively renormalizable field theory by introducing auxiliary ghost (c,c¯) and Nakanishi-Lautrup (b) fields, together with a group-valued matrix field U(x) integrated with the Haar measure:(5)Sgf[A,c,c¯,b]=∫ddx[∂μc¯a(∂μca+gfabcAμbcc)+ζc¯aca+iba∂μAμa+β2(Aμa)2].These fields can be unified into a superspace nonlinear sigma-model (NLσ) superfield V(x,θ,θ¯), enabling a compact representation of the gauge-fixing action:(6)Sgf[A,V]=1g2∫ddxdθdθ¯tr[(DμV)†(DμV)+2ζθ¯θ∂θ¯V†∂θV],with DμV=∂μV+igVAμ. Employing the replica trick, the authors showed that the corresponding super-NLσ model exhibits a phenomenon of radiative symmetry restoration, similar to what occurs in the two-dimensional O(N)/O(N−1) model. In the symmetric phase, a finite screening gluon mass mg2≃β emerges at tree level, consistent with lattice observations of a massive gluon propagator in the Landau gauge. This framework thus provides a gauge-fixing free of Gribov ambiguities, local and renormalizable, and reduces in practice to a massive extension of the FP action [26,27].2.2The refined Gribov–Zwanziger frameworkAn alternative approach, initiated by Gribov and developed by Zwanziger, enforces the restriction of the functional integral to the first Gribov region Ω, defined as the set of gauge fields satisfying the Landau condition ∂μAμa=0 and for which the FP operator Mab(A)=−∂μDμab(A) is positive definite [3,6,7]. This restriction can be implemented by adding to the action the nonlocal horizon functional(7)H(Ah)=g2∫d4xd4yfabcAμh,b(x)[M−1(Ah)]ad(x,y)fdecAμh,e(y),where Aμh is a gauge-invariant, transverse composite field defined through a Stueckelberg field ξ as(8)Aμh=h†Aμh+igh†∂μh,h=eigξaTa.The horizon term can be localized by introducing Zwanziger auxiliary bosonic fields (φ¯,φ) and Grassmann fields (ω¯,ω), leading to the Gribov-Zwanziger (GZ) action. The refined GZ (RGZ) model further includes dimension-two condensates ⟨A2⟩, 〈φ¯φ−ω¯ω〉, generating additional mass scales (m, M) that improve agreement with lattice data [8,9]. In the BRST-invariant Ah formulation [11–13], the complete RGZ action reads schematically(9)SRGZ=SFP−∫d4x(φ¯μacMab(Ah)φμbc−ω¯μacMab(Ah)ωμbc)−γ2∫d4xgfabc(Ah)μa(φ+φ¯)μbc+m22∫d4x(Ah)μa(Ah)μa−M2∫d4x(φ¯μabφμab−ω¯μabωμab),where γ is the Gribov parameter determined self-consistently by the horizon condition 〈H(Ah)〉=d(N2−1). The resulting gluon propagator exhibits the so-called decoupling behavior,D(p2)=p2+M2p4+(m2+M2)p2+m2M2+2g2Nγ4,remaining finite and nonzero at p2 → 0, while the ghost propagator is infrared finite. This framework preserves an exact BRST symmetry and yields correlation functions in quantitative agreement with lattice data over a wide momentum range. For later convenience, the BRST structure of the Zwanziger doublets (φ, ω) and (ω¯,φ¯) can be recast in a compact (4|1) superspace form, where the BRST operator s acts as a translation along a single Grassmann coordinate θ. This reformulation, detailed in Appendix D, makes BRST nilpotency and algebraic renormalizability manifest, without introducing any new degrees of freedom.Both the ST and RGZ formalisms thus furnish local, renormalizable and gauge-invariantly meaningful treatments of the Gribov problem. The former implements a smooth averaging over copies through replica NLσ fields and radiative restoration, while the latter imposes a hard restriction to the first Gribov region by means of the horizon condition and condensates. Their structural similarity (both leading to a massive yet confining gluon sector) motivates the unified approach developed in the next section.3A unified gauge fixing: interpolating between ST and GZ/RGZThe ST construction fixes the gauge by averaging over all Gribov copies with a nonuniform weight P[A,U] built from the Landau functional f[A, U] and the FP Hessian F[A,U] [14,17]. In parallel, the GZ framework restricts the functional measure to the first Gribov region through the horizon functional H(Ah), written in terms of the transverse, gauge-invariant composite field Aμh [6,7,11,12]. To unify both, we propose the following mixed averaging for a gauge-dependent operator O:(10)〈〈O[A]〉〉β,ζ,γ=∑is(i)O[AUi]det(F[A,Ui]+ζ1)|detF[A,Ui]|e−βf[A,Ui]−γ4H(Ah,Ui)∑is(i)det(F[A,Ui]+ζ1)|detF[A,Ui]|e−βf[A,Ui]−γ4H(Ah,Ui).Here s(i)=signdetF[A,Ui], β, ζ are the ST gauge-fixing parameters of mass dimension two controlling, respectively, a gluon mass and a ghost mass term in the local formulation, and γ is the Gribov mass of GZ. Gauge-invariant observables are unchanged by (10). The limits(γ=0)⇒ST,(β→∞,ζ→0)⇒GZ,(β,γ≠0)⇒interpolationmake the interpolation explicit. Since Ah is BRST-invariant, the horizon functional H(Ah,Ui) is identical for all copies and therefore factorizes out of the average (10). Consequently, the horizon term does not modify the relative weights of the copies; the interpolation in (10) reflects the coexistence of the ST copy-averaging mechanism and the RGZ horizon suppression.As in Serreau and Tissier [14], Serreau et al. [15], Reinosa et al. [17], the ST weight is localized by introducing a matrix field U(x) and ghost/antighost/Nakanishi-Lautrup fields (c,c¯,b) or, equivalently, the NLσ superfields V in superspace. The denominator is handled through the replica trick [28]: introduce n superfields {Vk}k=1n and take n → 0 at the end. Independently, the horizon functional is localized à la Zwanziger with (φ¯,φ) and (ω¯,ω) fields and the transversality of Ah is implemented with the Stueckelberg field h=eigξ, Lagrange multiplier τ and the pair (η¯,η), see Sorella et al. [11,12], Terin et al. [13].Choosing one replica to factor out the group volume and denoting by SST[A;Vk] the local ST gauge-fixing sector with parameters (β, ζ), the unified local action reads(11)Sunif=SYM[A]+SFP[A,c,c¯,b]+∑k=2nSST[A;Vk]−∫d4x(φ¯μacMab(Ah)φμbc−ω¯μacMab(Ah)ωμbc)−γ2∫d4xgfabc(Ah)μa(φ+φ¯)μbc+∫d4x(τa∂μ(Ah)μa−η¯aMab(Ah)ηb)+m22∫d4x(Ah)2−M2∫d4x(φ¯φ−ω¯ω),The ST block and the GZ/RGZ block each admit a nilpotent BRST symmetry. In the Ah formulation one has the exact BRST (to all orders)sAμ=−Dμc,sc=g2[c,c],sc¯=ib,sAh=0,s(φ¯,φ,ω¯,ω,η,η¯,τ)=0.and s2=0 [11,12]. The ST replica sector enjoys supersymmetries in superspace for ζ=0; for ζ ≠ 0 these are softly broken but the local theory remains perturbatively renormalizable in d=4 [14,17]. Because the two sectors couple only through A/Ah, the combined action (11) is local and power-counting renormalizable; the proof follows by merging the Ward identities of Serreau and Tissier [14], Reinosa et al. [17] with the algebraic renormalization in Dudal et al. [9], Sorella et al. [11],12], Terin et al. [13].For γ=0 and M=m=0 we recover the ST massive FP theory with parameters (β, ζ) and n replicas. The replica sector behaves as a constrained NLσ model; for ζ=0 closed replica loops vanish (topological sector), while ζ ≠ 0 allows nontrivial loops and a phenomenon of radiative symmetry restoration in the replica sector [17]. In that symmetric phase, only the singled replica contributes to the tree-level gluon mass, yielding a finite screening mass mg2=β in the limit n → 0; in the broken phase mg2∼nβ→0.Taking β → ∞ (which selects absolute minima of f[A, U]) and ζ → 0 while keeping γ ≠ 0 switches off the replica dynamics and reproduces the RGZ local action with exact BRST and parameters (γ, m, M) fixed by their gap equations [7,9,11]. With both sectors active, the tree-level transverse gluon two-point function reads schematically(12)ΔT−1(p2)=p2Z(p2;α)+βΞrep(ζ;phase)︸STreplicamassblock+MRGZ(p2;γ,m,M)︸RGZpolestructure,where Ξrep=1 in the replica-symmetric phase (and 0 in the broken phase, in the limit n → 0) as found in Reinosa et al. [17], and MRGZ(p2) represents the usual RGZ refinement. Thus, the unified propagator interpolates continuously between a pure ST massive FP and an RGZ-type decoupling curve. The nonperturbative parameters satisfy coupled gap equations:(13)∂Γ∂γ=0⇒〈H(Ah)〉=4V(N2−1),∂Γ∂m2=0,∂Γ∂M2=0,as in RGZ [9], and for the replica sector [17](14)2βrg¯r2=T(χ^r)−T(χ^r+ζr),8π2βrg¯r2=(χ^r+ζr)lnχ^r+ζrμ¯2−χ^rlnχ^rμ¯2,with χ^r≥0 the replica mass scale in the symmetric phase, T(m2) the (dimensionally reduced) tadpole integral, and bars denote renormalized quantities in the scheme of Reinosa et al. [17]. BRST/Nielsen identities of the Ah formulation ensure that physical quantities do not depend on the linear-covariant gauge parameter α and that the longitudinal sector is not renormalized, to all orders [11,12].Varying (β, ζ) at fixed (γ, m, M) deforms the IR mass scale in (12). The replica-symmetric domain (radiative restoration) provides a natural way to generate a Curci-Ferrari-like gluon mass from gauge fixing [14,17], while RGZ controls the nonlocal horizon physics and condensates. Implementing a weighted-copy average on the lattice (tunable β, ζ) and comparing the measured gluon/ghost correlators with (12) can test whether data prefer predominantly ST-like, RGZ-like, or mixed regimes. The unified action (11) reproduces (i) ST for γ=0, including the phase diagram in (β, ζ); (ii) RGZ for β → ∞, ζ → 0; and (iii) preserves exact BRST in the RGZ sector and perturbative renormalizability in the ST sectorThen, the gauge fixing (10) and its local version (11) furnish a concrete bridge between the ST copy-average mechanism (with its replica NLσ phenomenology and radiative symmetry restoration) and the GZ/RGZ restriction to the first Gribov region (with horizon suppression and condensates).4RenormalizabilityThe proof of renormalizability follows the algebraic approach based on the identification of all Ward identities and the characterization of the most general counterterm compatible with them, according to the procedure of algebraic renormalization [29]. Therefore, we work in the Landau gauge, ∂μAμ=0, and use the BRST-invariant formulation of RGZ based on the composite transverse, gauge-invariant field Aμh. We, thus, unify the ST replicated gauge-fixing with the RGZ sector:(15)Stot=SYM[A]+SFP[A,c,c¯,b]+SST[A;{Vk},β,ζ]+SGZ[Ah,φ,φ¯,ω,ω¯;γ]+Sref[Ah,φ,φ¯,ω,ω¯;m2,M2]+Stransv[Ah,τ,η,η¯],where Aμh≡h†Aμh+igh†∂μh, h=eigξ, and Vk(x,θ,θ¯) are the ST replica superfields (k=2,⋯,n; the limit n → 0 is taken at the end). The parameters β, ζ are the ST gauge-fixing masses. The BRST operator s is nilpotent, s2=0, and acts as(16)sAμa=−Dμab(A)cb,sca=g2fabccbcc,sc¯a=ba,sba=0,sh=−igch,sAμh,a=0,sφμab=sφ¯μab=sωμab=sω¯μab=0,sτa=sηa=sη¯a=0,sVk=igVkc(k=2,⋯,n).To control non-linear variations we couple external sources (antifields)(17)Sext=∫d4x(KμasAμa+Lasca),with engineering dimensions [K]=3, [L]=4 and ghost numbers gh(K)=−1, gh(L)=−2. We define Σ≡Stot+Sext. The Slavnov functional reads(18)S(Σ)=∫d4x(δΣδKμaδΣδAμa+δΣδLaδΣδca+baδΣδc¯a)=0.Linearizing around Σ gives the nilpotent operator(19)BΣX=∫d4x(δΣδKμaδXδAμa+δΣδAμaδXδKμa+δΣδLaδXδca+δΣδcaδXδLa+baδXδc¯a),BΣ2=0.Besides (18), the following hold:(20)Gaugecondition(b-equation):δΣδba=∂μAμa,(21)[2pt]Antighostequation(Landau):δΣδc¯a+∂μδΣδKμa=0,Ghostequation:GaΣ≡∫d4x(δδca+gfabcc¯bδδbc)Σ=Δcla.with a linear classical breaking Δcla (non-renormalized). For Ah we impose transversality with a Lagrange multiplier:(22)δΣδτa=∂μAμh,a=0.The replica sector enjoys (i) permutation symmetry Sn−1 among Vk (k ≥ 2); (ii) for ζ=0, topological supersymmetries (which suppress replica loops); for ζ ≠ 0, Sn−1 and the assumed gauge covariance of Vk remain.Now, consider a local perturbation Σ→Σ+εΔΣ with dim(ΔΣ) ≤ 4 and gh=0. Symmetries imply(23)BΣΔΣ=0,δΔΣδb=0,δΔΣδc¯+∂·δΔΣδK=0,δΔΣδτ=0,plus the replica symmetries. The cohomology of BΣ at ghost number 0 is generated by gauge-invariant polynomials in Fμν and Ah, together with the BRST-invariant, dimension-two operators (Ah)2 and (φ¯φ−ω¯ω). The replica sector is cohomologically trivial for local counterterms, and Sn−1 forbids terms that distinguish replicas (k ≥ 2). Up to BΣ-exact terms (field/source renormalizations), the most general local solution is(24)ΔΣ=∫d4x[a014Fμν2+a1(∂μc¯a)Dμabcb+a2iba∂·Aa+a3β2(Aμa)2+a4ζc¯aca+a5(Aμh,a)2+a6(φ¯φ−ω¯ω)]+BΣ(…).In particular, the superspace replica superfields Vk do not require independent wave-function renormalization factors: any possible divergence in the Vk sector can be absorbed into the renormalization of (Aμ,c,c¯) and the gauge-fixing parameters (β, ζ), preserving the unitarity constraint V†V=1. No new structures beyond those already in Σ appear: the theory is stable under renormalization. Although the unified action can be decomposed into an ST-like and an RGZ-like block, algebraic renormalization acts on the total functional as a single BRST complex. The cohomology of BΣ contains only gauge-invariant representatives built from Ah, Fμν, and the composite replica operators, so that all allowed counterterms are reabsorbed by a single set of renormalization constants. Hence, the unification is algebraic rather than additive, for more details the reader can see the Appendix E. Therefore, the multiplicative renormalizations are written as(25)Aμ=ZA1/2Aμ,R,c=Zc1/2cR,c¯=Zc1/2c¯R,γ2=Zγ2γR2,m2=Zm2mR2,M2=ZM2MR2,g=ZggR,β=ZββR,ζ=ZζζR,and analogously for all auxiliary fields φ,φ¯,ω,ω¯,h,….From (18), (21) and (21) in the Landau gauge (Taylor scheme) one obtains(26)ZgZcZA1/2=1,Zβ=ZA−1,Zζ=Zc−1.In the BRST-invariant RGZ sector (based on Ah), known non-renormalization relations reduce the number of independent Z’s to ZA and Zc; the renormalizations of γ2, m2, M2 are then fixed functions of (ZA, Zc, Zg). Thus, all admissible counterterms are reabsorbed by renormalizations of fields, couplings, and mass parameters already present and therefore the unified action Stot is renormalizable to all orders in the algebraic sense.5Gap system, suppression mechanisms, and matching (ST ↔ RGZ)The ST and RGZ formalisms implement different but complementary mechanisms to deal with Gribov copies. In RGZ, the functional integral is restricted to the first Gribov region Ω={Aμ|∂μAμ=0,Mab(A)>0} by the horizon condition 〈H(Ah)〉=4V(N2−1), introducing the Gribov parameter γ and suppressing configurations outside Ω. In contrast, the ST replica weight statistically averages over copies inside Ω through the Landau functional F(A), exponentially damping higher-F configurations as exp[−βΔF]. Our unified construction combines these two effects: the RGZ horizon term restricts the measure to Ω, while the replica weight concentrates it near the minima of F within Ω, providing a smooth, BRST-consistent interpolation between averaging and suppression. In this setting, the dynamically generated gluon mass of the ST sector, mg2=βR, can be matched to the RGZ mass combination mAA2=m2+M2+λ4/M2, which anchors the two gap systems to the same infrared scale.We now make this correspondence explicit by analyzing the two families of gap equations and showing how the replica-generated scale βR can be matched onto the RGZ parameters (m2, M2, λ4) under a common renormalization scheme. On the ST side, the gap equation follows from the stationary condition of the replica effective potential with respect to the Lagrange multiplier χ enforcing the supersymmetric constraint Nk2=1. After integrating out the (n−1) replicas in dimensional regularization, one obtains in d=4 dimensions (DR+MS‾ scheme) the renormalized relation(27)8π2g¯R2βRμ¯2=(χ^R,sym+ζR)lnχ^R,sym+ζRμ¯2−χ^R,symlnχ^R,symμ¯2with a unique solution χ^R,sym≥0 in the symmetric phase. The matching conditions discussed later are implemented in this symmetric phase (ζ ≠ 0), requiring χ^R,sym≥0; accordingly, the parameters (β, ζ) are assumed to lie within the domain allowed by the gap inequality. In this regime the gluon screening mass is mg2=βR, which plays the same role as the dynamical mass emerging from the dimension-two condensates in the RGZ framework. On the RGZ side, the corresponding dynamical parameters are fixed by the horizon condition and the refinement gap equations,(28)Horizon:∂Γvac∂γ2=0,(29)Refinement:∂Γvac∂m2=0,∂Γvac∂M2=0.whose solutions determine (γ, m2, M2, λ4). Matching the infrared limits of the two propagators then yields the relations summarized below.To make this correspondence explicit, we expand both descriptions around small external momentum and identify the infrared matching conditions between the replica-generated scale βR and the RGZ parameters. At tree level in the RGZ sector, the transverse gluon propagator reads(30)DAARGZ,tree(p2)=p2+M2(p2+M2)(p2+m2)+λ4,λ4=2g2Nγ4.Hence the inverse propagator and its infrared limit are(31)ΓTRGZ,tree(p2)≡(DAARGZ,tree(p2))−1=(p2+M2)(p2+m2)+λ4p2+M2,ΓTRGZ,tree(0)=m2M2+λ4M2.On the ST side, in the replica-symmetric phase (ζ ≠ 0) one has a radiatively generated screening mass(32)ΓTST(0)=mg2≡βR.A working hypothesis that guides our matching is that the ST screening mass and the RGZ dimension-two mass have a common origin in the IR: the replica-induced scale βR generated by copy averaging (and radiative restoration) accounts for the very same A2-type contribution that, in the RGZ language, is represented by the refined mass combination mAA2 in (9). Operationally, we implement this by imposing the IR matching ΓT(0)=βR,(33)ΓT(0)=!βR⟹m2M2+λ4M2=βR⟺m2M2+λ4=M2βR.Next, expand ΓT(p2) around p2=0 in the unified theory,(34)ΓT(p2)=βR︸fromST+Z−1p2+ΠTrep(p2)+ΠTRGZ(p2)+O(p4),and impose the slope matching at the origin,(35)ΓT′(0)=!ddp2ΓTRGZ,tree(p2)|p2=0=(m2+M2)M2−(m2M2+λ4)M4=m2M2−βRM2,where we used (33) in the last equality. Therefore,(36)Z−1+ΠT′rep(0)+ΠT′RGZ(0)=m2−βRM2.It is important to mention that these infrared matching conditions are scheme dependent; in this work we adopt the Landau MOM/Taylor scheme, imposing ZgZcZA1/2=1, so that the one-loop contributions ΠTrep and ΠTRGZ consistently fix the wave-function renormalization Z. Eqs. (33) and (36) provide two relations tying the ST mass scale βR to the RGZ parameters (m2, M2, λ4) under a common renormalization scheme, while ∂Γvac/∂γ2=0 and the condensate gap equations fix γ2, m2, and M2. In this way, the replica-generated mass βR acts as the infrared anchor that determines the RGZ mass combination m2M2+λ4 and constrains the ratio (m2−βR)/M2 through the slope at the origin.Having established the infrared and slope matching conditions (33)–(36), the full gap system comprises the replica Eq. (27) together with the horizon and condensate gaps (28)–(29). Solving them within a common renormalization scheme, e.g. the infrared-safe one, will determine the RGZ parameters (γ, m2, M2, λ4) consistently with the dynamically selected βR from the replica sector.Using the infrared matching condition ΓT(0)=βR together with the RGZ tree relation λ4=2g2Nγ4, we obtain a direct link between the ST screening mass and the Gribov parameter:(37)γ4(μ)=M2(μ)βR(μ)−m2(μ)M2(μ)2g2(μ)N.Eq. (37) holds in the common renormalization scheme and scale μ used for both sectors and requires M2βR > m2M2 for positivity. In the simplified case m2=0, this reduces to(38)γ4=[M2βR]/(2g2N).Beyond tree level the numerical coefficient receives finite corrections, but the qualitative statement remains: once βR is fixed by the replica gap, γ is determined by Eq. (37) together with the refinement parameters.Moreover, Nielsen identities guarantee the gauge-parameter independence of the transverse pole mass; ζ → 0 collapses the replica loops (topological limit), while γ → 0 and M2 → 0 recover the Curci–Ferrari limit. The lattice comparison can then be organized through DAA(0) and its slope dD/dp2|p2=0, keeping all the gaps simultaneously satisfied. The corresponding analysis proceeds as:1.Fix the renormalization scheme and impose the Landau conditions for ZA, Zc; use ZgZcZA1/2=1.2.Solve the ST gap (27) to obtain χ^R,sym and the allowed phase domain; set βR=mg2.3.Compute ΠTrep(p2) (momentum-dependent part) and ΠTRGZ(p2) at one loop.4.Match ΓT(0),ΓT′(0) with (30) (or the one-loop RGZ result), and solve (28)+(29) for (m2, M2, λ4).5.Verify Slavnov/Nielsen identities and compare gluon/ghost correlators with lattice data.Last but not least, the RGZ restriction controlled by γ removes field configurations outside the first Gribov region, but degeneracies of the Landau functional along gauge orbits may still persist inside Ω (notably near the boundary of the FMR). In our unified measure, the ST weight gives a complementary suppression of such residual copies: for two copies (U1, U2) on the same orbit one has a relative weight ∝exp[−βΔF], with ΔF=F(AU2)−F(AU1). Thus, at fixed γ, increasing β exponentially favors the minima of F and lifts the degeneracy among copies that survive the horizon constraint. In the formal limit β → ∞ (with ζ → 0), the probability concentrates on the absolute minima of F, approaching the FMR selection; while for finite β the suppression is controlled and smooth, leaving gauge-invariant observables unaffected by the normalization of the copy average.To accomplish that fix a gauge orbit [A] and let Ui be the set of (Landau) copies on that orbit. In the unified average,〈〈O[A]〉〉β,ζ,γ=∑is(i)O[AUi]det(Fi+ζ1)|detFi|e−βFi−γ4H(Ah)∑is(i)det(Fi+ζ1)|detFi|e−βFi−γ4H(Ah)Fi:=F[A,Ui],F=12∫(∂·A)2.one has, for fixed ζ ≥ 0 and γ ≥ 0:limβ→∞〈〈O[A]〉〉β,ζ,γ=∑i∈Mwi(ζ)O[AUi]∑i∈Mwi(ζ),wi(ζ):=limζ′→ζs(i)det(Fi+ζ′1)|detFi|.where M={i:Fi=minjFj} is the set of configurations that minimize F on a given gauge orbit. In particular, if the minimum is unique, thenlimβ→∞〈〈O[A]〉〉β,ζ,γ=O[AU*],F*=minjFj.•Since Ah is gauge invariant, we have H(Ah,Ui)=H(Ah) for all i. Therefore, the factor e−γ4H(Ah) cancels exactly between the numerator and denominator, and the ratio of weights between different copies does not depend on γ:PiPj=s(i)s(j)det(Fi+ζ1)det(Fj+ζ1)|detFj||detFi|e−β(Fi−Fj).•For large β, by the Laplace method for discrete Gibbs measures [30], all terms with Fi>minjFj are exponentially suppressed as e−βΔF. Thus, only indices i∈M contribute in the limit β → ∞, and the remaining factors are the wi(ζ) mentioned above.•When ζ → 0, one recovers the original ST case, where the ratio det(Fi+ζ)/|detFi| regularizes zeros and signs of the FP determinant. For finite positive ζ, the wi(ζ) remain positive and of order O(1). □Corollary:For two copies U1, U2 on the same orbit,P2P1=[prefactordependingon(ζ,F1,2)]×e−βΔF,ΔF=F2−F1.Therefore, the hierarchy between copies is mainly governed by e−βΔF. The parameter β acts as a “resolution control” that, with γ kept fixed (restriction to Ω), resolves the residual degeneracies that still exist inside Ω.6Unified superspace for ST and RGZThis section puts the superspace ingredients in one place and shows how we host, side by side, the ST replica block and the RGZ block in a supersymmetric way. As previously stated the two blocks only interact through the gauge field Aμ (or the BRST-invariant composite Aμh). This makes the symmetry content transparent and supports the algebraic renormalization used above. The Zwanziger localization fields form the BRST doublets sφμ=ωμ,sωμ=0,sω¯μ=φ¯μ,sφ¯μ=0, with s2=0. Using a (4|1) superspace with Grassmann coordinate θ and s ≡ ∂θ, we package them as Φμ=φμ+θωμ,Φ¯μ=ω¯μ+θφ¯μ. With the BRST-invariant composite Aμh=h†Aμh+igh†∂μh (and ∂μAμh=0 enforced by Stransv), the localized horizon + refinement sector isSRGZ(4|1)=∫d4xdθΦ¯μM(Ah)Φμ+γ2g∫d4xdθAμh(Φμ+Φ¯μ)+M22∫d4xdθΦ¯μΦμ+m22∫d4x(Aμh)2,with M(Ah)=−∂D(Ah). It is BRST-invariant by construction and reproduces the usual component form after dθ integration. The ST replica block is written with two Grassmann coordinates (θT,θ¯T) and group-valued superfields Vk∈SU(N), k=2,⋯,n, withSST=1g2∑k=2n∫d4xdθTdθ¯Ttr[(DμVk)†(DμVk)+2ζθ¯TθT∂θ¯TVk†∂θTVk],and the singled replica is described by SFP[A,c,c¯,b]. The topological supersymmetries are softly broken by ζ; for ζ → 0 replica loops cancel. We place both blocks in a (4|3) superspace with (θB;θT,θ¯T), where θB generates the RGZ BRST and (θT,θ¯T) the ST topological pair. The symmetry algebra has sB2=Q2=Q¯2=0, [sB,Q]=[sB,Q¯]=0, and {Q,Q¯} softly broken by ζ. The total action readsShyb=SYM+SFP+Stransv[Ah]+SST[A;Vk]+SRGZ(4|1)[Ah;Φ,Φ¯],and reduces to pure ST for γ=0 (and m=M=0), and to pure RGZ for β → ∞, ζ → 0. The Slavnov identity for sB plus the topological Ward identities in the replica block imply that all allowed local counterterms are absorbed by the common renormalizations already used in the main text (including ZgZcZA1/2=1, Zβ=ZA−1, Zζ=Zc−1 in Landau gauge). For more details the reader can see the Appendix D.7Conclusions and outlookWe have built and analyzed a unified gauge fixing that continuously interpolates between the Serreau–Tissier copy-averaged Landau gauge and the (Refined) Gribov–Zwanziger restriction to the first Gribov region. At the level of the functional average, our construction combines the ST weight based on the Landau functional and the FP Hessian with a GZ-type suppression driven by the horizon functional. Using the replica trick and the Aμh localization, we derived a manifestly local, power-counting renormalizable action that couples a replica NLσ sector to the BRST-invariant RGZ sector through the gauge field only.On the structural side, our unified action enjoys (i) the exact nilpotent BRST symmetry of the Ah formulation in the RGZ block and (ii) the replica permutation symmetry (and, for ζ=0, topological supersymmetries) in the ST block. By mixing the corresponding Ward identities, we established algebraic renormalizability to all orders: the most general local counterterm compatible with the identities is a linear combination of operators already present in the classical action and can be reabsorbed by multiplicative renormalizations of fields, the coupling, and the mass parameters (β, ζ, γ, m2, M2). In the Landau scheme, the Taylor relation ZgZcZA1/2=1 holds and fixes the non-renormalization of the ghost–gluon vertex at vanishing ghost momentum; Zβ=ZA−1 and Zζ=Zc−1 follow directly. It is important to note that the unification achieved here is not merely structural but algebraic: the BRST cohomology connects the ST and RGZ sectors through the gauge field, representing a single renormalizable framework for both descriptions.Dynamically, the ST replica sector exhibits radiative symmetry restoration for ζ ≠ 0, leading (in the replica-symmetric phase) to a finite screening gluon mass equal to the renormalized parameter, mg2=βR, after the n → 0 limit. In parallel, the RGZ parameters (γ, m2, M2) are selected by the horizon condition and the two dimension-two gap equations. We formulated a consistent matching strategy in which the small-momentum expansion of the full transverse two-point function fixes (i) ΓT(0)=βR and (ii) the wave-function normalization, while the three RGZ gaps determine (γ, m2, M2) at the same renormalization scale. The resulting propagator interpolates smoothly between a massive-FP (ST) form and the RGZ decoupling form, providing a controlled node to study how infrared correlators depend on the relative weight of copy averaging versus horizon suppression and condensates.Phenomenologically, our framework suggests direct lattice tests. Implementing a tunable copy-weighting (varying β, ζ) and confronting the measured DAA(0) and its slope with the unified prediction would indicate whether present ensembles are closer to the ST-like, RGZ-like, or mixed regime. On the continuum side, the one-loop momentum-dependent piece from the replica sector is finite at p → 0 (the mass is fixed already at tree level), and the BRST/Nielsen identities secure gauge-parameter independence of the transverse pole mass.Several developments now become natural:1.completing the unified one-loop analysis of gluon and ghost propagators (and selected vertices), including the running of the couplings and masses;2.solving the coupled gap system beyond one loop and studying the renormalization-group flow on the ST–RGZ interpolation;3.extending our construction to linear covariant gauges and to finite temperature/chemical potential; and4.exploring the impact on hadronic observables through functional methods: the DSEs and the FRG.Altogether, the present work gives a concrete and flexible bridge between two IR descriptions of YM theories, clarifying their common ground and offering a single, local, BRST-consistent setting in which to quantify their interplay.AcknowledgementsR. C. Terin gratefully acknowledges some of the researchers in infrared YM theories whose guidance have shaped part of his academic trajectory.Declaration of competing interestThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Appendix AField content, Ward identities, and basic non-renormalizationWe collect the engineering (mass) dimensions and ghost numbers of all fields and external sources used in the unified ST + RGZ formulation (Euclidean d=4):Aμcc¯bφμφ¯μωμω¯μh(ξ)Aμhτdim10221111012gh0+1−100000000ηη¯KμLdim2234gh+1−1−1−2Replica superfields Vk(x,θ,θ¯) (k=2,⋯,n) are dimensionless and ghostless. The Grassmann coordinates satisfy ∫dθdθ¯θ¯θ=−1 and carry no engineering dimension. With Σ=Stot+Sext, the ST identity readsS(Σ)=∫d4x(δΣδKμaδΣδAμa+δΣδLaδΣδca+baδΣδc¯a)=0.Linearizing around Σ defines the nilpotent operator(A.1)BΣX=∫d4x[δΣδKμaδXδAμa+δΣδAμaδXδKμa+δΣδLaδXδca+δΣδcaδXδLa+baδXδc¯a],BΣ2=0.It acts consistently on the Aμh-sector because sAμh=0 and on the Stueckelberg field through sh=−igch.(A.2)(i)Gaugecondition:δΣδba=∂μAμa.(A.3)(ii)Antighostequation:δΣδc¯a+∂μδΣδKμa=0.(A.4)(iii)Ghostequation:GaΣ≡∫d4x(δδca+gfabcc¯bδδbc)Σ=Δcla,(A.5)(iv)Ahtransversality:δΣδτa=∂μAμh,a=0.The replica sector enjoys (a) the permutation symmetry Sn−1 of {Vk}k≥2 and (b) for ζ=0 additional topological supersymmetries which suppress replica loops.At ghost number zero, the cohomology of BΣ is generated by gauge-invariant polynomials in Fμν and in the BRST-invariant composite Aμh, plus the dimension-two, BRST-invariant operators (Ah)2 and (φ¯φ−ω¯ω). The Stueckelberg sector and the replica supermultiplets assemble in BRST-covariant trivial pairs for local counterterms; Sn−1 forbids replica-distinguishing terms. Hence the most general local counterterm is Eq. (24) in the main text, up to BΣ-exact pieces, proving all-order stability.In Landau gauge, combining the ST identity with the antighost and ghost equations yields the Taylor relation for the ghost-gluon vertex at vanishing ghost momentum,(A.6)ZgZcZA1/2=1,Zβ=ZA−1,Zζ=Zc−1.In the BRST-invariant RGZ formulation (based on Ah), standard algebraic arguments reduce the independent Z’s to (ZA, Zc); the renormalizations of γ2, m2, M2 become fixed functions of (ZA, Zc, Zg) constrained by (A.6). Also, the above identities, together with the properties of the linearized operator BΣ, are the algebraic ingredients securing the stability of the unified action under renormalization. They give the starting point for the infrared matching analysis summarized in Section 5.Appendix BReplica one-loop tensor and small-momentum matchingIntegrating the Nk fluctuations (symmetric phase) yields, at leading order in g,(B.1)Πμνrep(q)=(n−1)βδμν+(n−1)g¯2[Iμνχ^+ζ(q)−Iμνχ^(q)],with(B.2)Iμνm2(q)=μ2ε∫ddp(2π)d(2p−q)μ(2p−q)ν(p2+m2)((p−q)2+m2)=2δμνTm2+(q2δμν−qμqν)Fm2(q2).Dimensional regularization in d=4−2ε with μ¯2=4πe−γEμ2 gives(B.3)Tm2≡μ2ε∫ddp(2π)d1p2+m2=−m216π2(1ε+1+lnμ¯2m2)+O(ε),(B.4)[4pt]Fm2(q2)=−148π2[1ε+83+lnμ¯2m2+8m2q2−2(1+4m2q2)3/2ln(q2+q2+4m22m2)]+O(ε).Using the ST gap (symmetric phase) to eliminate the momentum-independent divergence, one obtains(B.5)Πμνrep(q)=(n−1)δμν[β−2g¯2(Tχ^−Tχ^+ζ)]+(q2δμν−qμqν)π(q2),with the finite scalar(B.6)π(q2)=(n−1)g¯2[Fχ^+ζ(q2)−Fχ^(q2)]=−(n−1)g¯248π2[lnχ^+ζχ^+8ζq2+F(4χ^q2)−F(4(χ^+ζ)q2)]F(x)=2(1+1x)3/2ln(x+1+x).The 1/q2 terms cancel between the brackets, so (B.6) is regular at q2 → 0. At q=0, tree + tadpole cancel in the symmetric phase, leaving the screening mass mg2=βR (from the singled-out replica).For the full transverse inverse two-point functionΓT(p2)=DT−1(p2)=mg2+Z−1p2+ΠTrep(p2)+ΠTRGZ(p2)+⋯,expanding (B.6) at small p2 and using the ST gap gives(B.7)ΠTrep(p2)=α1p2+O(p4),α1=ddp2[(q2δμν−qμqν)π(q2)]|q2=0.Thus the replica sector only corrects the wave-function renormalization at p2=0. The IR matching conditions used in Section 5 read(B.8)ΓT(0)=βR,ΓT′(0)=Z−1+α1+dΠTRGZdp2|0,which fix βR (from ST) and relate (m2, M2, λ4) to (βR, Z) once the RGZ gaps are imposed.Appendix CPractical rules: tree-level Feynman rules, Nielsen identity, and DR conventions•Gluon (RGZ tree):Dμνab(p)=δabPμν(p)p2+M2(p2+M2)(p2+m2)+λ4,Pμν(p)=δμν−pμpνp2.•Ghost: Gab(p)=−δab1p2.•Zwanziger fields (φ,φ¯,ω,ω¯) propagate with masses set by M2 (refined sector).•Replica superfields do not propagate as ordinary local fields; their one-loop net effect is encoded by (B.1).Standard YM three-/four-gluon vertices; ghost-gluon vertex (Taylor kinematics non-renormalization); RGZ coupling gfabcAh(φ+φ¯). We impose MOM conditions at p2=μ2:DAA(p2)|p2=μ2=DAAtree(μ2),Gcc¯(p2)|p2=μ2=Gcc¯tree(μ2),fixing ZA, Zc and imposing ZgZcZA1/2=1 in Landau gauge.In linear covariant gauges introduce a BRST doublet (α, χ) with sα=χ, sχ=0 and couple χ to the BRST variation of a gauge-fixing insertion. Differentiating the ST identity with respect to χ and to the fields yields∂∂αΓT(p2)=N(p2;α)ΓT(p2),with N regular near the pole. Hence the pole position p2=mpole2 is α-independent to all orders; in Landau gauge (α=0) this reduces to the statement used in the main text.We use d=4−2ε and the MS‾ scale μ¯2=4πe−γEμ2. The basic integrals used in Eqs. (B.3)-(B.4) are(C.1)μ2ε∫ddp(2π)d1p2+m2=−m216π2(1ε+1+lnμ¯2m2)+O(ε),(C.2)μ2ε∫ddp(2π)d1(p2+m2)((p−q)2+m2)=116π2(1ε+lnμ¯2m2+2−1+4m2q2ln1+4m2q2+11+4m2q2−1)+O(ε).from which (B.3)-(B.4) follow after standard tensor reduction.Appendix DUnified superspace for ST and RGZIn this appendix, we show with more details compared to Section 6 the BRST (4|1) superspace for the RGZ sector and a simple “hybrid” superspace that hosts both the ST replica block and the RGZ block side by side. Throughout, the two blocks talk to each other only through the gauge field Aμ (or its BRST-invariant composite Aμh). The Zwanziger localizing fields come in BRST doublets(D.1)sφμab=ωμab,sωμab=0,sω¯μab=φ¯μab,sφ¯μab=0,with s2=0. This already looks like a (nilpotent) supersymmetry: each fermionic field is the BRST variation of a bosonic partner. We use a (4|1) superspace with Grassmann coordinate θ and define the BRST differential as a translation:(D.2)s≡∂∂θ,θ2=0,s2=0.In the RGZ setup, there is no independent anti-BRST, so one Grassmann direction is enough. Package the doublets into BRST superfields(D.3)Φμab(x,θ):=φμab(x)+θωμab(x),Φ¯μab(x,θ):=ω¯μab(x)+θφ¯μab(x),so that sΦ=∂θΦ and sΦ¯=∂θΦ¯. The engineering dimensions and ghost numbers follow from the components:[Φ]=[Φ¯]=1,gh(Φ)=gh(Φ¯)=0,with gh(ω)=+1 and gh(ω¯)=−1. We work with the BRST-invariant composite field(D.4)Aμh=h†Aμh+igh†∂μh,∂μAμh=0,where h=eigξ. The transversality is enforced by a Lagrange multiplier τ and an auxiliary BRST doublet (η,η¯), while sAμh=0. The localized horizon action is written as a (4|1) superspace integral(D.5)Shor=∫d4xdθ[Φ¯μac(x,θ)Mab(Ah)Φμbc(x,θ)]+γ2gfabc∫d4xdθAμh,a(x)(Φμbc+Φ¯μbc),with Mab(Ah)=−∂μDμab(Ah). Using ∫dθ(X0+θX1)=X1, one recovers the standard component form(D.6)∫d4xdθΦ¯MΦ=∫d4x(φ¯Mφ−ω¯Mω)∫d4xdθAh·(Φ+Φ¯)=∫d4xAh·(φ+φ¯).The RGZ dimension-two pieces are BRST-invariant:(D.7)SM2=M22∫d4xdθΦ¯μabΦμab=M22∫d4x(φ¯μabφμab−ω¯μabωμab),and(D.8)Sm2=m22∫d4x(Aμh,a)2,which needs no superspace lifting. Collecting terms, the BRST-superspace action for RGZ is(D.9)SRGZ(4|1)=∫d4x(SFP[A,c,c¯,b]+Stransv[Ah;τ,η,η¯])+∫d4xdθΦ¯μacMab(Ah)Φμbc+γ2gfabc∫d4xdθAμh,a(Φμbc+Φ¯μbc)+M22∫d4xdθΦ¯μabΦμab+m22∫d4x(Aμh,a)2.This is BRST-invariant by construction (translation along θ). Coupling sources to BRST variations one builds the Slavnov functional; the linearized operator includes the θ-translation and is nilpotent. The cohomology at ghost number zero is generated by BRST-invariant operators made of Fμν, Ah, and Φ¯Φ. Hence the algebraic renormalizability of the RGZ block follows: no counterterm beyond those already present is allowed by the identities. It is consistent to take (Φ¯)†=Φ¯, (Φ)†=Φ, and ∫dθθ=1, ∫dθ1=0. Since Ah is real and BRST-invariant, the mass term (D.8) is positive in Euclidean signature.Now we will place the ST replica sector and the RGZ sector in a common superspace. The ST block lives along a topological pair (θT,θ¯T); the RGZ block lives along a BRST coordinate θB. They couple only through Aμ (or Aμh). Use a (4|3) superspace with Grassmann coordinates(θB;θT,θ¯T)1(seefootnotebelow),and define three nilpotent differentials11This (4|3) notation does not refer to an N=3 supersymmetric space; it simply denotes the product of a BRST superspace (4|1) with coordinate θB and the Parisi–Sourlas topological superspace (0|2) with coordinates (θT,θ¯T).sB≡∂θB,Q≡∂θT,Q¯≡∂θ¯T,with sB2=Q2=Q¯2=0 and [sB,Q]=[sB,Q¯]=0. In the ST block, {Q,Q¯} is softly broken by ζ (topological limit at ζ=0). We keep the BRST-invariant compositeAμh=h†Aμh+igh†∂μh,∂μAμh=0,imposed by Stransv[Ah;τ,η,η¯], and sBAμh=0. Next, we introduce n−1 group-valued superfieldsVk(x,θT,θ¯T)∈SU(N),Vk†Vk=1,k=2,⋯,n,where DμVk=∂μVk+igVkAμ. Their gauge-fixing action isSST[A;Vk]=1g2∑k=2n∫d4xdθTdθ¯Ttr[(DμVk)†(DμVk)+2ζθ¯TθT∂θ¯TVk†∂θTVk].The usual FP part SFP[A,c,c¯,b] plays the role of the “singled” replica. BRST covariance is sBVk=igVkc with the standard sB on (A,c,c¯,b). As in the above mentioned RGZ case, we gather Zwanziger doublets into (4|1) superfields along θB:Φμab(x,θB)=φμab(x)+θBωμab(x),Φ¯μab(x,θB)=ω¯μab(x)+θBφ¯μab(x),with sBΦ=∂θBΦ, sBΦ¯=∂θBΦ¯, and sBAμh=0. The horizon + refinement sector is(D.10)SRGZ(4|1)[Ah;Φ,Φ¯]=∫d4xdθBΦ¯μacMab(Ah)Φμbc+γ2gfabc∫d4xdθBAμh,a(Φμbc+Φ¯μbc)+M22∫d4xdθBΦ¯μabΦμab+m22∫d4x(Aμh,a)2,in which Mab(Ah)=−∂μDμab(Ah). Putting all pieces together (and imposing ∂μAμh=0):(D.11)Shyb=SYM[A]+SFP[A,c,c¯,b]+Stransv[Ah;τ,η,η¯]+SST[A;{Vk};β,ζ]+SRGZ(4|1)[Ah;Φ,Φ¯].The only link between the ST and RGZ parts is Aμ (and Aμh). The symmetry algebra is generated by (sB;Q,Q¯) withsB2=Q2=Q¯2=0,[sB,Q]=[sB,Q¯]=0,{Q,Q¯}soft∝ζ.Limits: (i) ζ=0 kills replica loops (topological limit); (ii) γ=0, m=M=0 gives back the massive FP/ST limit; (iii) β → ∞, ζ → 0 yields pure RGZ.Then, we define the Slavnov functional for sB with sources (Kμ, L) coupled to BRST variations, and the topological Ward identities for (Q,Q¯) in the replica block. The linearized operator BΣ factorizes into two commuting pieces. At ghost number zero, the cohomology is generated by BRST-invariant polynomials in Fμν and Ah, plus the dimension-two operators (Ah)2 and (Φ¯Φ). The replica block is cohomologically trivial for local counterterms under the Sn−1 permutation symmetry. Therefore all enabled counterterms are absorbed by multiplicative renormalizationsZA,Zc,Zg,Zβ,Zζ,Zγ2,Zm2,ZM2,where the Landau (Taylor) relations ZgZcZA1/2=1, Zβ=ZA−1, Zζ=Zc−1. If one prefers a single group-valued object, defineN(x,θB,θT,θ¯T)=exp{ig[θ¯TcT+c¯TθT+θ¯TθTh˜T+θBc]}U(x),with U ∈ SU(N), sBN=igNc, and (Q,Q¯) acting on the “T” components only. Then the ST functional becomesSST=1g2∫d4xdθTdθ¯Ttr[(DμN)†(DμN)+2ζθ¯TθT∂θ¯TN†∂θTN],where unitarity N†N=1 kept manifest. The Zwanziger fields remain as adjoint superfields (Φ,Φ¯) along θB.Appendix EAlgebraic unification in pure SU(N) Yang–Mills: explicit BRST for ST – RGZThe basic fields for this work are Aμa, the Faddeev–Popov triplet ca,c¯a,ba, a Stückelberg element h(x)=eigξaTa used to define the gauge-invariant projection Aμh, the Zwanziger pairs φμab,φ¯μab and ωμab,ω¯μab for the RGZ block, and n replica group elements Ui(x) ∈ SU(N) for the ST copy averaging. To keep locality and BRST exactness in the latter weights, we add two BRST doublet pairs for each replica: (χia,ρia), (χ¯ia,ρ¯ia) and (ϕia,ηia), (ϕ¯ia,η¯ia), with χ,χ¯,ϕ,ϕ¯ bosonic and ρ,ρ¯,η,η¯ fermionic.The unified BRST operator s is nilpotent, s2=0, and acts as(E.1)sAμa=−(Dμc)a,sca=−g2fabccbcc,sc¯a=ba,sba=0,sh=−igch,sAμh=0,sφμab=ωμab,sωμab=0,sω¯μab=φ¯μab,sφ¯μab=0,sUi=−igcUi,s(AμUi)=−Dμ(AUi)c,sχia=ρia,sρia=0,sρ¯ia=χ¯ia,sχ¯ia=0,sϕia=ηia,sηia=0,sη¯ia=ϕ¯ia,sϕ¯ia=0.The ST target weight in the static (copy-averaged) sector is(E.2)WST[A;Ui]∝det(F[AUi]+ζ1)|detF[AUi]|exp[−βf(AUi)],f(A)=12∫ddxAμaAμa,with the Landau FP operator Fab[A]=−∂μDμab(A) and the usual discrete sum over copies keeping s(i)=signdetF[AUi].The algebraic unification is implemented by a single gauge-fixing fermion Ψuni whose BRST variation generates both the FP Landau fixing and the ST weights. We choose(E.3)Ψuni=∫ddx[c¯a∂μAμa+α2c¯aba]+β∑i=1nVi[AUi;χi,χ¯i,ρi,ρ¯i]+ζ∑i=1nDi[AUi;ϕi,ϕ¯i,ηi,η¯i],(E.4)Sgf≡sΨuni.For the ST Landau-functional piece we take a minimal local choice(E.5)Vi=∫ddx[ρ¯ia∂μ(AUi)μa+κ2χ¯iaχ¯ia],(E.6)sVi=∫ddx[χ¯ia∂μ(AUi)μa−ρ¯ia∂μ(Dμ(AUi)c)a]+κχ¯iaρia.In the Landau limit α → 0, Gaussian completion over (χ¯i,ρi) and then ρ¯i yields effectively(E.7)exp[−β12∫ddx(AUi)μa(AUi)μa],up to BRST-exact contact terms and constants. The parameter κ > 0 is auxiliary and cancels in BRST-invariant observables.For the FP-determinant deformation we use a Zwanziger-like first-order representation,(E.8)Di=∫ddx[η¯ia(F[AUi]+ζ1)abϕib],(E.9)sDi=∫ddx[ϕ¯ia(F[AUi]+ζ)abϕib−η¯ia(F[AUi]+ζ)abηib]+η¯ia(sF[AUi])abϕib.Path integration over (ϕ,ϕ¯) and (η,η¯) gives the local, BRST-exact realization of det(F[AUi]+ζ). If one prefers to realize the ratio det(F+ζ)/det(F) directly, an unshifted companion (ϕ˜,η˜) can be added,(E.10)D˜i=∫ddxη˜¯iaFab[AUi]ϕ˜ib,(E.11)sD˜i=∫ddx[ϕ˜¯iaFab[AUi]ϕ˜ib−η˜¯iaFab[AUi]η˜ib+η˜¯ia(sF[AUi])abϕ˜ib].though in the standard ST prescription the denominator modulus is accounted for in the discrete copy sum by s(i), so this extra pair may be omitted.The RGZ sector is written entirely in terms of the BRST-invariant field Aμh, hence it is strictly invariant under s:(E.12)SRGZ[Ah,φ,ω;γ]=∫ddx[φ¯μacMab(Ah)φμbc]−∫ddx[ω¯μacMab(Ah)ωμbc]+gγ2fabc(Ah)μa(φ+φ¯)μbc+d(N2−1)γ4,(E.13)sAh=0,(E.14)sSRGZ=0.The full unified action in pure Yang–Mills is(E.15)Suni=SYM[A]+sΨuni[A,c,c¯,b;Ui,aux]+SRGZ[Ah,φ,ω;γ].With antifields Kμa and La coupled to sAμa and sca, and optional sources for the doublets, the Slavnov functional S(Σ)=0 takes the standard form and controls the cohomology relevant for renormalization.The interpolation limits are smooth and remain within pure Yang–Mills. When γ → 0 one recovers the pure ST side,(E.16)γ→0⇒Suni→SYM+sΨuni,while for β, ζ → 0 one keeps the standard FP Landau gauge together with the RGZ block,(E.17)β,ζ→0⇒Suni→SYM+s∫ddxc¯∂·+SRGZ(Ah),and the strict Landau condition is obtained by(E.18)α→0⇒∂μAμ=0.In this way, the ST average and the RGZ restriction are glued by the same nilpotent BRST operator acting on a common field space, which is exactly what is meant by an algebraic (rather than additive) unification.References[1]C.N.YangR.L.MillsConservation of isotopic spin and isotopic gauge invariancePhys. Rev.96195419119510.1103/PhysRev.96.191C. N. Yang, R. L. Mills, Conservation of isotopic spin and isotopic gauge invariance, Phys. Rev. 96 (1954) 191–195. 10.1103/PhysRev.96.191[2]L.D.FaddeevV.N.PopovFeynman diagrams for the Yang–Mills fieldPhys. Lett. B251967293010.1016/0370-2693(67)90067-6L. D. Faddeev, V. N. Popov, Feynman diagrams for the Yang–Mills field, Phys. Lett. B 25 (1967) 29–30. 10.1016/0370-2693(67)90067-6[3]V.N.GribovQuantization of nonabelian Gauge theoriesNucl. Phys. B139197811910.1016/0550-3213(78)90175-XV. N. Gribov, Quantization of nonabelian Gauge theories, Nucl. Phys. B 139 (1978) 1–19. 10.1016/0550-3213(78)90175-X[4]I.M.SingerSome remarks on the Gribov ambiguityCommun. Math. Phys.60197871210.1007/BF01609471I. M. Singer, Some remarks on the Gribov ambiguity, Commun. Math. Phys. 60 (1978) 7–12. 10.1007/BF01609471[5]N.VandersickelD.ZwanzigerThe Gribov problem and QCD dynamicsPhys. Rept.520201217525110.1016/j.physrep.2012.07.003N. Vandersickel, D. Zwanziger, The Gribov problem and QCD dynamics, Phys. Rept. 520 (2012) 175–251. 10.1016/j.physrep.2012.07.003[6]D.ZwanzigerLocal and renormalizable action from the Gribov horizonNucl. Phys. B323198951354410.1016/0550-3213(89)90122-3D. Zwanziger, Local and renormalizable action from the Gribov horizon, Nucl. Phys. B 323 (1989) 513–544. 10.1016/0550-3213(89)90122-3[7]D.ZwanzigerQuantization of Gauge fields, classical Gauge invariance, and Gluon confinementNucl. Phys. B345199046148210.1016/0550-3213(90)90396-UD. Zwanziger, Quantization of Gauge fields, classical Gauge invariance, and Gluon confinement, Nucl. Phys. B 345 (1990) 461–482. 10.1016/0550-3213(90)90396-U[8]D.DudalJ.A.GraceyS.P.SorellaN.VandersickelH.VerscheldeRefining the Gribov-Zwanziger approach in the Landau gauge: infrared propagators in harmony with the lattice resultsPhys. Rev. D78200806504710.1103/PhysRevD.78.065047D. Dudal, J. A. Gracey, S. P. Sorella, N. Vandersickel, H. Verschelde, Refining the Gribov-Zwanziger approach in the Landau gauge: infrared propagators in harmony with the lattice results, Phys. Rev. D 78 (2008) 065047. 10.1103/PhysRevD.78.065047[9]D.DudalS.P.SorellaN.VandersickelH.VerscheldeGribov no-pole condition, Zwanziger horizon function, Kugo-Ojima confinement criterion, boundary conditions, BRST breaking and all thatPhys. Rev. D79200906503910.1103/PhysRevD.79.121701D. Dudal, S. P. Sorella, N. Vandersickel, H. Verschelde, Gribov no-pole condition, Zwanziger horizon function, Kugo-Ojima confinement criterion, boundary conditions, BRST breaking and all that, Phys. Rev. D 79 (2009) 065039. 10.1103/PhysRevD.79.121701[10]D.DudalS.P.SorellaN.VandersickelH.VerscheldeNew features of the gluon and ghost propagator in the infrared region from the Gribov–Zwanziger approachPhys. Rev. D77200807150110.1103/PhysRevD.77.0715010711.4496D. Dudal, S. P. Sorella, N. Vandersickel, H. Verschelde, New features of the gluon and ghost propagator in the infrared region from the Gribov–Zwanziger approach, Phys. Rev. D 77 (2008) 071501. 0711.4496. 10.1103/PhysRevD.77.071501[11]S.P.SorellaExact nilpotent nonperturbative BRST symmetry for the Gribov-Zwanziger action in the linear covariant gaugePhys. Rev. D92201504503910.1103/PhysRevD.92.045039S. P. Sorella, et al., Exact nilpotent nonperturbative BRST symmetry for the Gribov-Zwanziger action in the linear covariant gauge, Phys. Rev. D 92 (2015) 045039. 10.1103/PhysRevD.92.045039[12]S.P.SorellaNonperturbative aspects of Euclidean Yang-Mills theories in linear covariant gauges: Nielsen identities and a BRST-invariant two-point correlation functionPhys. Rev. D95201704501110.1103/PhysRevD.95.045011S. P. Sorella, et al., Nonperturbative aspects of Euclidean Yang-Mills theories in linear covariant gauges: Nielsen identities and a BRST-invariant two-point correlation function, Phys. Rev. D 95 (2017) 045011. 10.1103/PhysRevD.95.045011[13]R.C.TerinAll-order renormalizable refined Gribov-Zwanziger model with BRST-invariant fermionic horizon function in linear covariant gaugesPhys. Rev. D1045202105404810.1103/PhysRevD.104.0540482107.08569R. C. Terin, et al., All-order renormalizable refined Gribov-Zwanziger model with BRST-invariant fermionic horizon function in linear covariant gauges, Phys. Rev. D 104 (5) (2021) 054048. 2107.08569. 10.1103/PhysRevD.104.054048[14]J.SerreauM.TissierLifting the Gribov ambiguity in Yang-Mills theoriesPhys. Lett. B71220129710310.1016/j.physletb.2012.04.041J. Serreau, M. Tissier, Lifting the Gribov ambiguity in Yang-Mills theories, Phys. Lett. B 712 (2012) 97–103. 10.1016/j.physletb.2012.04.041[15]J.SerreauM.TissierA.TresmontantCovariant gauges without Gribov copiesPhys. Rev. D92201510500310.1103/PhysRevD.92.1050031505.07270J. Serreau, M. Tissier, A. Tresmontant, Covariant gauges without Gribov copies, Phys. Rev. D 92 (2015) 105003. 1505.07270. 10.1103/PhysRevD.92.105003[16]M.TissierGribov copies, avalanches and dynamic generation of a gluon massPhys. Lett. B784201814615010.1016/j.physletb.2018.07.043M. Tissier, Gribov copies, avalanches and dynamic generation of a gluon mass, Phys. Lett. B 784 (2018) 146-150. 10.1016/j.physletb.2018.07.043[17]U.ReinosaJ.SerreauR.C.TerinM.TissierSymmetry restoration and the gluon mass in the Landau gaugeSciPost Phys.102202103510.21468/SciPostPhys.10.2.0352004.12413U. Reinosa, J. Serreau, R. C. Terin, M. Tissier, Symmetry restoration and the gluon mass in the Landau gauge, SciPost Phys. 10 (2) (2021) 035. 2004.12413. 10.21468/SciPostPhys.10.2.035[18]A.CucchieriT.MendesConstraints on the IR behavior of the gluon propagator in Yang-Mills theoriesPhys. Rev. Lett.100200824160110.1103/PhysRevLett.100.241601A. Cucchieri, T. Mendes, Constraints on the IR behavior of the gluon propagator in Yang-Mills theories, Phys. Rev. Lett. 100 (2008) 241601. 10.1103/PhysRevLett.100.241601[19]D.DudalO.OliveiraP.J.SilvaHigh precision statistical Landau gauge lattice gluon propagator computation vs. the Gribov–Zwanziger approachAnn. Phys. (N.Y.)397201835136410.1016/j.aop.2018.08.0191803.02281D. Dudal, O. Oliveira, P. J. Silva, High precision statistical Landau gauge lattice gluon propagator computation vs. the Gribov–Zwanziger approach, Ann. Phys. (N.Y.) 397 (2018) 351–364. 1803.02281. 10.1016/j.aop.2018.08.019[20]D.DudalO.OliveiraM.RoelfsP.J.SilvaSpectral representation for the gluon propagator and positivity violation at zero temperatureNucl. Phys. B952202011491210.1016/j.nuclphysb.2019.1149121901.05348D. Dudal, O. Oliveira, M. Roelfs, P. J. Silva, Spectral representation for the gluon propagator and positivity violation at zero temperature, Nucl. Phys. B 952 (2020) 114912. 1901.05348. 10.1016/j.nuclphysb.2019.114912[21]D.BinosiR.A.TripoltSpectral functions of confined particlesPhys. Lett. B801202013517110.1016/j.physletb.2019.1351711904.08172D. Binosi, R. A. Tripolt, Spectral functions of confined particles, Phys. Lett. B 801 (2020) 135171. 1904.08172. 10.1016/j.physletb.2019.135171[22]R.AlkoferL.von SmekalThe infrared behavior of QCD Green’s functions: confinement, dynamical symmetry breaking, and hadrons as relativistic bound statesPhys. Rept.353200128146510.1016/S0370-1573(01)00010-2R. Alkofer, L. von Smekal, The infrared behavior of QCD Green’s functions: confinement, dynamical symmetry breaking, and hadrons as relativistic bound states, Phys. Rept. 353 (2001) 281–465. 10.1016/S0370-1573(01)00010-2[23]A.C.AguilarA.A.NataleP.S.R.da SilvaRelating a Gluon mass scale to an infrared fixed point in pure Gauge QCDPhys. Rev. Lett.90200315200110.1103/PhysRevLett.90.152001A. C. Aguilar, A. A. Natale, P. S. R. da Silva, Relating a Gluon mass scale to an infrared fixed point in pure Gauge QCD, Phys. Rev. Lett. 90 (2003) 152001. 10.1103/PhysRevLett.90.152001[24]M.Q.HuberNonperturbative properties of Yang-Mills theoriesPhys. Rept.879202019210.1016/j.physrep.2020.04.004M. Q. Huber, Nonperturbative properties of Yang-Mills theories, Phys. Rept. 879 (2020) 1–92. 10.1016/j.physrep.2020.04.004[25]J.M.PawlowskiAspects of the functional renormalisation groupAnnal. Phys.32220072831291510.1016/j.aop.2007.01.007J. M. Pawlowski, Aspects of the functional renormalisation group, Annal. Phys. 322 (2007) 2831–2915. 10.1016/j.aop.2007.01.007[26]M.TissierN.WscheborInfrared propagators of Yang-Mills theory from perturbation theoryPhys. Rev. D82201010170110.1103/PhysRevD.82.101701M. Tissier, N. Wschebor, Infrared propagators of Yang-Mills theory from perturbation theory, Phys. Rev. D 82 (2010) 101701. 10.1103/PhysRevD.82.101701[27]M.TissierN.WscheborAn infrared safe perturbative approach to Yang-Mills correlatorsPhys. Rev. D84201104501810.1103/PhysRevD.84.045018M. Tissier, N. Wschebor, An infrared safe perturbative approach to Yang-Mills correlators, Phys. Rev. D 84 (2011) 045018. 10.1103/PhysRevD.84.045018[28]G.ParisiN.SourlasRandom magnetic fields, supersymmetry and negative dimensionsPhys. Rev. Lett.43197974474510.1103/PhysRevLett.43.744G. Parisi, N. Sourlas, Random magnetic fields, supersymmetry and negative dimensions, Phys. Rev. Lett. 43 (1979) 744–745. 10.1103/PhysRevLett.43.744[29]O.PiguetS.P.SorellaAlgebraic Renormalization: Perturbative Renormalization, Symmetries and AnomaliesLecture Notes in Physics Monographs281995Springer-VerlagBerlin, Heidelberg10.1007/978-3-540-49192-7O. Piguet, S. P. Sorella, Algebraic Renormalization: Perturbative Renormalization, Symmetries and Anomalies, 28 of Lecture Notes in Physics Monographs, Springer-Verlag, Berlin, Heidelberg, 1995. 10.1007/978-3-540-49192-7[30]D.García-ZeladaA large deviation principle for empirical measures on Polish spaces: application to singular Gibbs measures on manifoldsAnnales de l’Institut Henri Poincaré, Probabilités et Statistiques55320191673169510.1214/18-AIHP922D. García-Zelada, A large deviation principle for empirical measures on Polish spaces: application to singular Gibbs measures on manifolds, Annales de l’Institut Henri Poincaré, Probabilités et Statistiques 55 (3) (2019) 1673–1695. 10.1214/18-AIHP922