Yang–Mills Color Algebra and Perturbative Vertices

Yang–Mills perturbation theory factorizes into Lorentz and color structures fixed by the curvature and the ghost determinant. For a compact simple gauge factor in a linear covariant gauge, the three-gauge, four-gauge, ghost–gauge, and minimally coupled Dirac-matter vertices follow from one coupling and the declared field ordering. Their contractions test relative signs before any loop integration begins. These are perturbative rules; colored external legs here do not assert a nonperturbative asymptotic gluon spectrum.

Required background. The Yang–Mills action and gauge self-interaction supplies the expanded action and coupling convention. Gauge-fixed perturbation rules, ghost diagrams, and identity checks supplies the general rule-extraction and symmetry-factor method.

Helpful background. Compact Lie groups, roots, weights, and Weyl structure supplies representation theory beyond the invariant identities used here.

Color invariants and reductions

For an irreducible representation $R$ of a compact simple factor, define

$$\operatorname{tr}_R(T_R^aT_R^b)=T(R)\delta^{ab}, \qquad T_R^aT_R^a=C_2(R)\mathbf1_R,$$

and

$$f^{acd}f^{bcd}=C_A\delta^{ab}.$$

Taking a trace of the Casimir definition gives the useful dimension identity

$$d_R C_2(R)=d_A T(R).$$

For the site-wide $SU(N)$ convention $T(F)=1/2$,

Representation data	Value
$d_F$, $d_A$	$N$, $N^2-1$
$T(F)$	$1/2$
$C_F$	$(N^2-1)/(2N)$
$C_A$	$N$

Two reductions recur in vertex and loop calculations:

$$T_R^bT_R^aT_R^b =\left(C_2(R)-\frac12C_A\right)T_R^a, \qquad f^{abe}f^{cde}+f^{bce}f^{ade}+f^{cae}f^{bde}=0.$$

The first follows by commuting the middle generator once and using the adjoint Casimir; the second is Jacobi. For a reducible representation, apply the scalar Casimir formula to each irreducible block. The color tensor of a gauge-invariant amplitude must also obey color conservation,

$$\left(\sum_{i=1}^n \mathbf T_i^a\right)\mathcal C=0,$$

with all legs treated as incoming and an incoming antifundamental carrying $-(T^a)^T$. These identities reduce color without choosing explicit matrices.

Momentum-space vertices

Take every displayed momentum incoming and use the inherited Fourier rule $\partial_\mu\mapsto-ip_\mu$. The vertex factor includes the factor of $i$ from $e^{iS_{\rm int}}$. The cubic term of the expanded action is

$$\mathcal L_3=-g f^{abc}(\partial_\mu A_\nu^a)A^{b\mu}A^{c\nu}.$$

For gauge fields $A_\mu^a(p)$, $A_\nu^b(q)$, and $A_\rho^c(r)$, the two assignments with the differentiated field carrying $p$ give $g f^{abc}(p_\rho\eta_{\mu\nu}-p_\nu\eta_{\mu\rho})$. Adding the four assignments with the derivative on the other two fields, with $p+q+r=0$, gives

$$\mathcal V_{\mu\nu\rho}^{abc}(p,q,r) =g f^{abc}\left[ (q-r)_\mu\eta_{\nu\rho} +(r-p)_\nu\eta_{\rho\mu} +(p-q)_\rho\eta_{\mu\nu} \right].$$

For $A_\mu^a$, $A_\nu^b$, $A_\rho^c$, and $A_\sigma^d$, the four-gauge factor is

$$\begin{aligned} \mathcal V_{\mu\nu\rho\sigma}^{abcd} =-ig^2\big[{} &f^{abe}f^{cde} (\eta_{\mu\rho}\eta_{\nu\sigma}-\eta_{\mu\sigma}\eta_{\nu\rho})\\ &+f^{ace}f^{dbe} (\eta_{\mu\sigma}\eta_{\rho\nu}-\eta_{\mu\nu}\eta_{\rho\sigma})\\ &+f^{ade}f^{bce} (\eta_{\mu\nu}\eta_{\sigma\rho}-\eta_{\mu\rho}\eta_{\sigma\nu}) \big]. \end{aligned}$$

The four-point result follows by assigning the four external labels to the quartic action: its 24 assignments combine into the three displayed color pairings. The three-vertex is real because its derivative supplies a factor of $-i$ before the usual $iS_{\rm int}$ factor; the four-vertex retains $-i$. The same gauge tensors appear in Schwartz 2014, § 26.1, pp. 509–511, Eqs. (26.8)–(26.10) and Srednicki 2006, author manuscript § 72, p. 424, Eqs. (72.5)–(72.6), PDF. Srednicki calls the complete factor $iV$; it is $\mathcal V$ here. His opposite metric and propagator conventions must also be translated before assembling diagrams.

For the ghost vertex, start with the integrated-by-parts action from the gauge-fixed construction:

$$\mathcal L_{\rm gh} =(\partial^\mu\bar c^a)\partial_\mu c^a +g f^{abc}(\partial^\mu\bar c^a)A_\mu^b c^c.$$

Keep the Grassmann order $\bar c\,c$ fixed. For an incoming gauge boson $A_\mu^a(p)$, antighost $\bar c^b(q)$, and ghost $c^c(r)$, the interaction has color tensor $f^{bac}$, because the first index belongs to the differentiated antighost. There is only one assignment of these distinct fields, so $i\,g f^{bac}(-iq_\mu)=g f^{bac}q_\mu$. Therefore

$$\mathcal V_\mu^{a\bar b c}(p,q,r) =g f^{bac}q_\mu=-g f^{abc}q_\mu.$$

This is the same rule as the canonical antighost-first expression, after relabeling the external fields. The positive ghost kinetic action is explicit in Schwartz 2014, § 26.1, p. 509, Eq. (26.1). In his ghost diagram on p. 511, Eq. (26.11), the momentum on the antighost line is outgoing: reverse that momentum as well as translating the leg labels when comparing with the all-incoming rule here.

The arrow on a ghost line records the ordered contraction $\langle c\bar c\rangle$; reversing it is not an innocuous relabeling. Its free propagator is $i\delta^{ab}/(k^2+i0)$, and each closed ghost loop has the additional Grassmann minus sign. That loop sign is separate from the vertex’s color-index permutation. A Dirac field in $R$ has the gauge vertex

$$\mathcal V_\mu^a=ig\gamma_\mu T_R^a.$$

Each three-point gauge or ghost vertex has one power of momentum and one power of the dimensionless $g$; the four-gauge vertex has $g^2$ and no momentum. This dimension check catches missing derivatives in the ghost rule.

Vertex contractions and color checks

Contracting the three-gauge vertex with $p^\mu$ gives

$$p^\mu\mathcal V_{\mu\nu\rho}^{abc}(p,q,r) =g f^{abc}\left[ (r^2\eta_{\nu\rho}-r_\nu r_\rho) -(q^2\eta_{\nu\rho}-q_\nu q_\rho) \right].$$

The right-hand side is a difference of transverse momentum tensors $K^{(0)}_{\nu\rho}(k)=k^2\eta_{\nu\rho}-k_\nu k_\rho$. The quadratic action from $-F^2/4$ has kernel $-K^{(0)}$ before gauge fixing; it is important not to identify $K^{(0)}$ with the inverse raw Feynman propagator. The contraction vanishes after the adjacent external legs are put on shell and contracted with transverse polarizations, but it is not zero for a generic off-shell Green function. At loop level this simple Ward contraction is replaced by a Slavnov–Taylor relation containing ghost functions.

For four external adjoint legs, the exchange diagrams carry products such as $f^{a_1a_2e}f^{ea_3a_4}$. Choose all cubic-vertex orientations consistently; the three channel color factors then satisfy one Jacobi relation. The contact vertex supplies precisely the Lorentz and color terms needed for the sum of exchange and contact diagrams to pass the polarization replacement $\varepsilon_i\to p_i$. Testing only the contact diagram will fail.

A trace decomposition provides a bridge to color-ordered amplitudes. For $n\geq3$ external gauge bosons at tree level in the declared $SU(N)$ Yang–Mills theory, repeated use of the generator commutator rewrites the products of structure constants in fundamental single traces:

$$\mathcal M_n^{\rm tree} =g^{n-2}\sum_{\sigma\in S_n/Z_n} \operatorname{tr}\!\left( T^{a_{\sigma(1)}}\cdots T^{a_{\sigma(n)}} \right)A_n(\sigma),$$

with any overall trace-normalization factor absorbed consistently into the partial amplitudes. Reconstructing the full four-point color tensor from the ordered pieces and reducing it with Jacobi is an independent check on the vertex-based calculation. The canonical trace-basis construction uses generators $\sqrt2\,T^a$: its length-$n$ traces are larger by $2^{n/2}$, so its partial amplitudes are smaller by the same factor at fixed $g$. Translate both factors when using those rules.

A benchmark reduction

Suppose a fermion-line correction produces $T_R^bT_R^aT_R^b$. Then

$$\begin{aligned} T_R^bT_R^aT_R^b &=C_2(R)T_R^a+if^{abc}T_R^bT_R^c\\ &=\left(C_2(R)-\frac12C_A\right)T_R^a. \end{aligned}$$

For an $SU(N)$ fundamental this becomes $-T^a/(2N)$. A second common closed adjoint contraction gives $f^{acd}f^{bcd}=N\delta^{ab}$. Together these distinguish $C_F$, $C_A$, and $T(F)$: none of the three may be replaced by “the color factor” without specifying the contraction.

Independent checks and limits

1. Permutation check: exchanging any two complete labels $(a,\mu,p)$ of the three-gauge vertex leaves the bosonic vertex invariant because both $f^{abc}$ and the kinematic bracket change sign.
2. Contraction check: $p^\mu\mathcal V_{\mu\nu\rho}$ must be the difference of the $r$ and $q$ transverse momentum tensors in the order shown.
3. Casimir check: $d_R C_2(R)=d_A T(R)$ gives $N C_F=(N^2-1)/2$ for the fundamental.
4. Abelian limit: $f^{abc}\to0$ removes all pure-gauge and ghost interactions while leaving the matter–gauge vertex for an Abelian charge.

Common pitfalls

Leaving momentum flow implicit. Reversing one incoming momentum changes derivative vertices. Redraw every diagram with all momenta incoming before comparing two rule tables.

Using $C_F$, $C_A$, and $T(F)$ interchangeably. They answer different contractions. Reduce the actual index pattern before substituting $SU(N)$ values.

Demanding an off-shell Ward contraction vanish. The contraction is a difference of inverse propagators. Zero follows only after the appropriate on-shell or full Slavnov–Taylor conditions are imposed.

Exercises

Derive the displayed contraction $p^\mu\mathcal V_{\mu\nu\rho}^{abc}$ using $p=-q-r$. Explain why it vanishes between transverse on-shell states on the other two legs, but need not vanish off shell. As a concrete off-shell check, use $SU(2)$ with $f^{123}=1$, $q=(Q,0,0,0)$, $r=(0,Q,0,0)$ and $Q>0$, and evaluate the component with $(a,b,c)=(1,2,3)$ and $\nu=\rho=2$.

Solution

The metric term uses $p\cdot(q-r)=r^2-q^2$. The two remaining terms combine as

$$p_\rho(r-p)_\nu+p_\nu(p-q)_\rho =q_\nu q_\rho-r_\nu r_\rho.$$

Together these reproduce the difference of transverse momentum tensors. If $q^2=r^2=0$, contraction with $\varepsilon^\nu(q)\varepsilon^\rho(r)$ kills the remaining terms by $q\cdot\varepsilon(q)=r\cdot\varepsilon(r)=0$. This is an algebraic external-leg check; a complete scattering Ward identity still requires all the diagrams for that process.

In the off-shell example, $q^2=Q^2$, $r^2=-Q^2$, $q_2=r_2=0$ and $\eta_{22}=-1$, giving

$$p^\mu\mathcal V_{\mu22}^{123} =g\bigl[(-Q^2)(-1)-(Q^2)(-1)\bigr] =2gQ^2\neq0.$$

The nonzero result is the required difference of momentum tensors, not a failure of gauge invariance.

References

• Schwartz, Matthew D. Quantum Field Theory and the Standard Model. Cambridge University Press, 2014. DOI.
• Srednicki, Mark. Quantum Field Theory. Prepublication author manuscript, 2006. Author’s draft and published-edition errata. Open PDF.