Formal Perturbative Yang–Mills Renormalization Data

Use the adopted pure $SU(2)$ Yang–Mills model: $G=SU(2)$, $\mathfrak{g}=\mathfrak{s}\mathfrak{u}(2)$, the selected smooth right-principal bundle sector has no matter fields, and the positive invariant form is $\kappa (X,Y)=-2{\mathrm{Tr}}_2(XY)$ with $\kappa (T_a,T_b)={\delta }_{ab}$ for $T_a=-i{\sigma }_a\mathrm{/}2$, extended bilinearly to the graded coefficient algebra. Work in a nonempty local trivialization chart of the selected bundle over oriented, time-oriented Minkowski space with signature $(-,+,+,+)$ and coordinates $x^0=t$ in natural units $c=\mathrm{\hslash }=1$. Let ${\eta }_{\mu \nu }=\mathrm{diag}(-1,1,1,1)$ and define ${\mathrm{\partial }}^{\mu }:={\eta }^{\mu \nu }{\mathrm{\partial }}_{\nu }$. With the alternating wedge–Lie bracket fixed by the principal-curvature supplier, write $F_A=dA+\frac{1}{2}[A\wedge A]$. The adopted Lorentzian gauge action is $S_0[A]=-(2g_{\mathrm{Y}\mathrm{M}}^2{)}^{-1}\int \kappa (F_A\wedge \ast F_A)$, with fixed positive dimensionless $g_{\mathrm{Y}\mathrm{M}}$. In the perturbative formulas $A_{\mu }$ has canonical engineering dimension $1$ and $g_{\mathrm{Y}\mathrm{M}}$ dimension $0$ in mass units; the ghost $c$ has dimension $0$, the antighost $\bar{c}$ and auxiliary field $b$ have dimension $2$, and the BRST differential has dimension $0$. Assign each antifield ${\mathrm{\Phi }}^{\ast }$ engineering dimension $4-\dim \mathrm{\Phi }$, so its product with the corresponding BRST variation has density dimension $4$ whenever that variation is nonzero. The model postulate supplies the physical units, bundle sector, action, and boundary conditions; the classical gauge-invariance result supplies invariance under the declared bundle gauge transformations. This definition does not prove either supplier.

In the chosen local trivialization write $D_{\mu }c={\mathrm{\partial }}_{\mu }c+[A_{\mu },c]$ and choose the linear covariant gauge condition $G(A)={\mathrm{\partial }}^{\mu }{A}_{\mu }=0$ with gauge parameter $\xi >0$. The nonminimal BRST variables are an odd antighost $\bar{c}$ of ghost number $-1$ and an even auxiliary field $b$ of ghost number $0$, with $s\bar{c}=b$ and $sb=0$. The gauge-fixing fermion and gauge-fixing contribution are

${\mathrm{\Psi }}_{\xi }=\int d^{4}x\kappa (\bar{c},{\mathrm{\partial }}^{\mu }{A}_{\mu }+\frac{\xi }{2}b),S_{\mathrm{g}\mathrm{f}}:=s{\mathrm{\Psi }}_{\xi }\mathrm{.}$

Here $s$ is the odd BRST derivation on the finite-jet field algebra, extended to the nonminimal variables and their antifields using the specified BV convention. The gauge condition is a local perturbative choice; it does not assert a global slice or remove all gauge copies. For finite-dimensional regulated data as in the preceding gauge-slice definition, with a smooth right action of a $k$-dimensional Lie group $K$ on $X$, a gauge condition $\chi :U\to {\mathbb{R}}^k$, and specified positive densities on $\mathfrak{k}$ and ${\mathbb{R}}^k$, the ordinary Faddeev–Popov Jacobian at every $x\in U$ is the density determinant ${\mathrm{\Delta }}_{\mathrm{F}\mathrm{P}}(x)=\mathrm{\mid }\det (d{\chi }_x\circ R_x)\mathrm{\mid }$, where $R_x=d(k\mapsto x\cdot k{)}_e$. It is zero when $d{\chi }_x\circ R_x$ is singular; at a slice point it is positive exactly when the slice is transverse. Such a Jacobian is not a continuum functional determinant and does not define a continuum measure.

For the relative local complex used in this datum, the field coordinates are $A_{\mu }^a,c^a,{\bar{c}}^a,b^a$ and their canonical antifields $A_a^{\ast \mu },c_a^{\ast },{\bar{c}}_a^{\ast },b_a^{\ast }$. Each antifield ${\mathrm{\Phi }}^{\ast }$ has parity $\mathrm{\mid }\mathrm{\Phi }\mathrm{\mid }+1$ modulo $2$ and ghost number $-1-\mathrm{gh}(\mathrm{\Phi })$. The extended coefficient algebra ${\mathcal{F}}_{\mathrm{B}\mathrm{V},\mathrm{n}\mathrm{m}}(U)$ is $C^{\mathrm{\infty }}(U)$ tensored with the graded-commutative polynomial algebra in all even field/antifield jet generators and the exterior algebra in all odd field/antifield jet generators; each element depends on finitely many jets. For every multi-index $I\in {\mathbb{N}}^4$, total derivatives satisfy $D_{\mu }f={\mathrm{\partial }}_{\mu }f$ and $D_{\mu }{\mathrm{\Phi }}_I={\mathrm{\Phi }}_{I+e_{\mu }}$; the $D_{\mu }$ commute. Let ${\mathcal{A}}_{\mathrm{B}\mathrm{V},\mathrm{n}\mathrm{m}}^{g,p}(U)$, $0\le p\le 4$, be the finite sums of horizontal $p$-forms with coefficients in the ghost-number-$g$ part of this algebra, where the $d{x}^{\mu }$ anticommute under wedge product; set ${\mathcal{A}}_{\mathrm{B}\mathrm{V},\mathrm{n}\mathrm{m}}^{g,-1}(U)=\{0\}$ and $d_H=d{x}^{\mu }{D}_{\mu }$.

A formal perturbative renormalization datum consists of the selected pure SU(2) classical BV action ${\mathrm{\Gamma }}_0$, the above gauge-fixing choice, the dimensional-regularization prescription $D=4-2ϵ$ with a fixed positive reference mass scale $\mu$, modified minimal subtraction ($\overline{\mathrm{M}\mathrm{S}}$), and a local counterterm space ${\mathcal{C}}_{\le d_{\ast }}$. Fix the coupling convention $g_D(\mu ,ϵ)={\mu }^{ϵ}g(\mu )$ and $g_B={\mu }^{ϵ}{Z}_{g}g(\mu )$, where $g(\mu )$ is dimensionless, $g(\mu )=g_{\mathrm{Y}\mathrm{M}}$ at the selected reference scale, and $Z_g=1+O(\lambda )$ is the formal coupling-renormalization factor. Thus $g_D$ and the bare coupling $g_B$ have mass dimension $ϵ$. With the geometric connection $A$ of dimension $1$, the dimensionally continued gauge action has coefficient $1\mathrm{/}{g}_D^2={\mu }^{-2ϵ}\mathrm{/}g(\mu {)}^2$; its density has dimension $4-2ϵ=D$. The four-dimensional engineering dimensions used for local counterterms below remain those of the $ϵ\to 0$ local complex. Use the unrescaled $\mu$ convention: subtraction removes the pole parts together with the associated $-{\gamma }_E+\log (4\pi )$ constants, whose one-loop combination is ${\mathrm{\Delta }}_{ϵ}=1\mathrm{/}ϵ-{\gamma }_E+\log (4\pi )$. Equivalently, ${\bar{\mu }}^2=4\pi e^{-{\gamma }_E}{\mu }^2$ absorbs these constants into the scale; these two scale conventions are not applied simultaneously. This specifies scheme data and asserts neither a value for $Z_g$ nor a running-coupling equation. In the adopted pure-gauge sector, dimensional regularization is taken to preserve BRST in the formal analytic continuation: the regulator introduces no BRST-breaking term in the regularized gauge-fixed action or its defining rules. This is an explicit regulator property of the selected prescription, not a conclusion of the Slavnov–Taylor theorem and not an automatic claim about other matter contents or regularization schemes. The QAP and local anomaly-cancellation assumptions below remain separate. Here $d_{\ast }$ is a finite engineering-dimension bound fixed by the selected power-counting prescription; ${\mathcal{C}}_{\le d_{\ast }}$ consists of integrated Lorentz-scalar polynomials of ghost number zero and engineering dimension at most $d_{\ast }$ in $A,c,\bar{c},b$, their antifields, and finitely many spacetime derivatives, represented by elements of ${\mathcal{A}}_{\mathrm{B}\mathrm{V},\mathrm{n}\mathrm{m}}^{0,4}(U)$. The usual renormalizable four-dimensional Yang–Mills choice is $d_{\ast }=4$. With the canonical BV convention, the classical master functional is

${\mathrm{\Gamma }}_0=S_0+{\int }_U{d}^{4}x(A_a^{\ast \mu }(D_{\mu }c{)}^a+\frac{1}{2}{c}_a^{\ast }[c,c{]}^a-{\bar{c}}_a^{\ast }b),$

where the nonminimal term is omitted if that pair is not included. Let $s=B_{{\mathrm{\Gamma }}_0}=({\mathrm{\Gamma }}_0,\cdot )$ be the local evolutionary representative of the canonical BV antibracket with this master functional: on each field and antifield jet it is prolonged by $s(D_I\mathrm{\Phi })=D_I(s\mathrm{\Phi })$, and on a horizontal form $d{x}^{{\mu }_1}\wedge \cdots \wedge d{x}^{{\mu }_p}\wedge a$ it is extended by $s(d{x}^{{\mu }_1}\wedge \cdots \wedge d{x}^{{\mu }_p}\wedge a)=(-1{)}^{p}d{x}^{{\mu }_1}\wedge \cdots \wedge d{x}^{{\mu }_p}\wedge sa$, so $s{d}_H+d_{H}s=0$. Its restriction to $A,c,\bar{c},b$ is the transformations already specified. The shifted Jacobi identity for the canonical antibracket gives $s^{2}X=\frac{1}{2}(({\mathrm{\Gamma }}_0,{\mathrm{\Gamma }}_0),X)=0$ for every element $X$ of the canonical domain; the jet prolongation then gives the differential on the full finite-jet complex. The datum is admissible only when ${\mathrm{\Gamma }}_0$ satisfies the classical master equation and the chosen local BV functional domain has no nonzero central element of ghost number one. Its loop-counting variable $\lambda$ is formal and dimensionless; a formal effective action is a coefficient sequence $\mathrm{\Gamma }(\lambda )={\mathrm{\Gamma }}_0+{\sum }_{n\ge 1}{\lambda }^n{\mathrm{\Gamma }}_n$ and no convergence at any nonzero $\lambda$ is asserted.

For a local BV functional $\mathrm{\Gamma }$, define the Slavnov functional by $\mathfrak{S}(\mathrm{\Gamma }):=\frac{1}{2}(\mathrm{\Gamma },\mathrm{\Gamma })$ using the chosen odd antibracket, and define the linearized operator $B_{{\mathrm{\Gamma }}_0}X:=({\mathrm{\Gamma }}_0,X)$. For fixed allowed counterterm coefficients $C_j\in {\mathcal{C}}_{\le d_{\ast }}$, put ${\mathrm{\Gamma }}^{(<n)}:=\mathrm{\Gamma }+{\sum }_{1\le j<n}{\lambda }^j{C}_j$, with lower-order coefficients corrected so that $\mathfrak{S}({\mathrm{\Gamma }}^{(<n)})=O({\lambda }^n)$. The superscript records the corrected orders, not a truncation of the given $\mathrm{\Gamma }$. The datum has the local QAP property when, at each first order $n$ where the coefficientwise Slavnov identity fails, its defect has a finite-jet local four-form representative ${\delta }_n\in {\mathcal{A}}_{\mathrm{B}\mathrm{V},\mathrm{n}\mathrm{m}}^{1,4}(U)$ and a finite-jet local three-form ${\rho }_n\in {\mathcal{A}}_{\mathrm{B}\mathrm{V},\mathrm{n}\mathrm{m}}^{2,3}(U)$ satisfying

$\mathfrak{S}({\mathrm{\Gamma }}^{(<n)})={\lambda }^n{\int }_U{\delta }_n+O({\lambda }^{n+1}),B_{{\mathrm{\Gamma }}_0}{\delta }_n+d_H{\rho }_n=0.$

This local representative and consistency condition are part of the QAP input; take ${\delta }_n$ to have engineering dimension at most $d_{\ast }$ and ${\rho }_n$ at most $d_{\ast }-1$. The definition does not prove the QAP. Define the extended relative local cohomology by

$H_{\mathrm{B}\mathrm{V},\mathrm{n}\mathrm{m}}^{g,p}(s\mid d_H):=\frac{\{a\in {\mathcal{A}}_{\mathrm{B}\mathrm{V},\mathrm{n}\mathrm{m}}^{g,p}(U):\text{there is }b\in {\mathcal{A}}_{\mathrm{B}\mathrm{V},\mathrm{n}\mathrm{m}}^{g+1,p-1}(U)\text{ with }sa+d_{H}b=0\}}{\{sm+d_{H}n:m\in {\mathcal{A}}_{\mathrm{B}\mathrm{V},\mathrm{n}\mathrm{m}}^{g-1,p}(U),n\in {\mathcal{A}}_{\mathrm{B}\mathrm{V},\mathrm{n}\mathrm{m}}^{g,p-1}(U)\}}\mathrm{.}$

For the counterterm-relevant restriction, let ${\mathcal{L}}_{\le d_{\ast }}^{g,p}$ be the subspace of ${\mathcal{A}}_{\mathrm{B}\mathrm{V},\mathrm{n}\mathrm{m}}^{g,p}(U)$ consisting of Lorentz-covariant local horizontal forms whose coefficient engineering dimension is at most $d_{\ast }-4+p$. Here covariance means that under a simultaneous proper orthochronous Lorentz change of coordinates, fields, antifields and jet indices, the expression transforms as a horizontal $p$-form, with no fixed noncovariant background tensor; this is a local covariance requirement on charts and does not require $U$ to be Lorentz invariant. Antifields transform in the dual tensor representations needed to make the displayed master density a scalar. In degree four this says precisely that the density multiplying the oriented volume form is a Lorentz scalar. In degree three it includes the covariant current forms whose horizontal derivative is a scalar density. The four-dimensional weights count $D_{\mu }$ as dimension $1$ and $s$ as dimension $0$; they are coefficient weights, so no dimension is assigned to the displayed $d{x}^{\mu }$ for this bound. Put ${\mathcal{L}}_{\le d_{\ast }}^{g,-1}=0$. The scalar classical master action and the tensorial total derivatives preserve covariance; their dimension rules give $s{\mathcal{L}}_{\le d_{\ast }}^{g,p}\subseteq {\mathcal{L}}_{\le d_{\ast }}^{g+1,p}$ and $d_H{\mathcal{L}}_{\le d_{\ast }}^{g,p}\subseteq {\mathcal{L}}_{\le d_{\ast }}^{g,p+1}$. Define the restricted quotient for $g=0,1$ by

$H_{\mathrm{L}\mathrm{o}\mathrm{r},\le d_{\ast }}^{g,4}(s\mid d_H):=\frac{\{a\in {\mathcal{L}}_{\le d_{\ast }}^{g,4}:sa+d_{H}b=0\text{ for some }b\in {\mathcal{L}}_{\le d_{\ast }}^{g+1,3}\}}{\{sm+d_{H}n:m\in {\mathcal{L}}_{\le d_{\ast }}^{g-1,4},n\in {\mathcal{L}}_{\le d_{\ast }}^{g,3}\}}\mathrm{.}$

Both representatives and equivalence primitives are restricted: $m$ has scalar top-form density of dimension at most $d_{\ast }$, and $n$ is a covariant current three-form of coefficient dimension at most $d_{\ast }-1$. This denominator is contained in the numerator, since for $a=sm+d_{H}n$ the allowed witness $b=sn$ gives $sa+d_{H}b=0$. This restricted cohomology is not asserted equal to a subspace of unrestricted cohomology, nor to the minimal-only cohomology. The counterterm quotient means the infinitesimally Slavnov-invariant part of ${\mathcal{C}}_{\le d_{\ast }}$ modulo precisely these allowed trivial counterterms and currents. For the restricted QAP input require ${\delta }_n\in {\mathcal{L}}_{\le d_{\ast }}^{1,4}$ and ${\rho }_n\in {\mathcal{L}}_{\le d_{\ast }}^{2,3}$.

The datum satisfies the local anomaly-cancellation hypothesis when a sequence of pairs ${\mathrm{\Xi }}_n\in {\mathcal{L}}_{\le d_{\ast }}^{0,4}$ and ${\mathrm{\Theta }}_n\in {\mathcal{L}}_{\le d_{\ast }}^{1,3}$ is supplied, with

${\delta }_n=B_{{\mathrm{\Gamma }}_0}{\mathrm{\Xi }}_n+d_H{\mathrm{\Theta }}_n$

at every first failing order of the recursive partial actions with $C_j=-{\int }_U{\mathrm{\Xi }}_j$; at nonfailing orders set ${\mathrm{\Xi }}_n={\mathrm{\Theta }}_n=0$. Thus each encountered obstruction class $[{\delta }_n]$ vanishes in $H_{\mathrm{L}\mathrm{o}\mathrm{r},\le d_{\ast }}^{1,4}(s\mid d_H)$, and the sequence supplies explicit primitives rather than requiring an additional choice construction. This is a separate assumption: the QAP consistency equation alone does not imply it, and no vanishing of the whole ghost-number-one cohomology is asserted. The supplied primitives obey the covariance and dimension restrictions above; unrestricted exactness does not establish this hypothesis. A $d_H$-exact top form represents zero as a local functional only under the declared compact-support or boundary condition. The coefficientwise Slavnov–Taylor identity is $\mathfrak{S}(\mathrm{\Gamma }(\lambda ))=0$ as an equality of formal power series.

Barnich, Brandt, and Henneaux, §§2.2–2.3 and 12.2–12.3

Maas, §§4.7.2 and 5.2.7–5.2.8

Adopted Yang–Mills action, field content, units, and scope

Classical gauge invariance of the selected action

Minimal non-Abelian BRST field and sign conventions

Minimal finite-jet BRST nilpotence

Finite-jet local cohomology modulo total derivatives

BV antibracket, classical master action, and nonminimal variables

Finite-dimensional gauge-slice and Faddeev–Popov Jacobian data

Right-action principal connection and local gauge potential

Curvature formula and local transformation convention