QCD Fields, Scales, and the Perturbative Domain

Quantum chromodynamics is an $SU(3)$ gauge theory of gluons and six named quark fields. Its ultraviolet coupling decreases, but perturbation theory is controlled only when every relevant physical scale, mass threshold, logarithm, and measurement restriction has also been checked. The result of this page is therefore a domain test, not the slogan “large momentum means perturbative.”

Required background. Non-Abelian screening and asymptotic freedom supplies the one-loop sign and group-theory structure of the QCD beta function.

Helpful background. Effective field theory as a controlled expansion supplies the hierarchy and remainder logic used below.

QCD fields and normalizations

With fundamental quarks $q_f$ and gluons $G^a_\mu$, the renormalized Lagrangian may be organized as

$$\mathcal L_{\mathrm{QCD}} =-\frac14 G^a_{\mu\nu}G^{a\mu\nu} +\sum_f \bar q_f\left(i\gamma^\mu D_\mu-m_f\right)q_f +\frac{\theta g_s^2}{32\pi^2}G^a_{\mu\nu}\widetilde G^{a\mu\nu} +\mathcal L_{\mathrm{gf}}+\mathcal L_{\mathrm{gh}}+\mathcal L_{\mathrm{ct}}.$$

Here $D_\mu=\partial_\mu-i g_s T^aG^a_\mu$, $G^a_{\mu\nu}=\partial_\mu G^a_\nu-\partial_\nu G^a_\mu+g_sf^{abc}G^b_\mu G^c_\nu$, and $\widetilde G^{a\mu\nu}=\tfrac12\epsilon^{\mu\nu\rho\sigma}G^a_{\rho\sigma}$. Gauge fixing and ghosts define a perturbative calculation; they do not add physical external particles. The theta term is dynamically important in a different line of inquiry and is not part of ordinary CP-even perturbative predictions.

For $SU(3)$ in the fundamental representation,

Quantity	Definition	QCD value
generator trace	$\operatorname{tr}(T^aT^b)=T_F\delta^{ab}$	$T_F=\tfrac12$
quark Casimir	$T^aT^a=C_F\mathbf1$	$C_F=\tfrac43$
gluon Casimir	$f^{acd}f^{bcd}=C_A\delta^{ab}$	$C_A=3$
strong coupling	$\alpha_s=g_s^2/(4\pi)$	scheme- and scale-dependent

These conventions produce the standard color factors used in hard coefficients and splitting kernels Schwartz 2014, §§25.2 and 26.2, pp. 488–93 and 505–13.

The flavor labels are $u,d,s,c,b,t$, but the number $n_f$ appearing in a low-energy perturbative description is the number of active flavors in that description, not invariably six. A quark mass is also incomplete data unless its renormalization scheme and scale are given. The next page makes both statements operational.

In a massless flavor limit the vector current $\bar q_f\gamma^\mu q_f$ is conserved, while non-singlet axial currents have the expected chiral-limit status and the singlet axial current is anomalous. These facts constrain sum rules and operator mixing, but they do not turn confined quarks into asymptotic states.

Running and the generated scale

In a mass-independent scheme, the one-loop equation is

$$\mu\frac{d\alpha_s}{d\mu} =-\frac{\beta_0}{2\pi}\alpha_s^2+O(\alpha_s^3), \qquad \beta_0=\frac{11}{3}C_A-\frac{4}{3}T_Fn_f =11-\frac{2}{3}n_f.$$

Holding $n_f$ fixed while integrating gives

$$\alpha_s(\mu)= \frac{4\pi}{\beta_0\ln(\mu^2/\Lambda_{n_f}^2)} \left[1+O\!\left(\frac{\ln\ln(\mu^2/\Lambda_{n_f}^2)}{\ln(\mu^2/\Lambda_{n_f}^2)}\right)\right].$$

The integration constant $\Lambda_{n_f}$ depends on the scheme and active-flavor theory. Across a heavy-quark threshold, both the coupling and this constant must be related by matching; simply changing $n_f$ inside the denominator is not a consistent procedure. The derivation and its infrared limitation are developed in Schwartz 2014, §§26.4–26.6, pp. 517–28.

The positive $\beta_0$ relevant to QCD gives ultraviolet asymptotic freedom. The one-loop pole near $\Lambda_{n_f}$ says that this expansion has lost control; it is not a derivation of confinement, a hadron spectrum, or a phase transition.

A perturbative-domain test

Let $Q$ denote the hard scale and let $Q_i$ include all other physical scales introduced by masses, transverse momenta, jet radii, vetoes, endpoints, or the measurement function. A fixed-order prediction is credible only if the following tests pass.

1. Weak coupling: choose natural renormalization scales for which $\alpha_s(\mu_R)$ is small enough that successive known orders are stable.
2. No untreated large logarithm: for every ratio $Q_i/Q_j$, check terms such as $\alpha_s^n\ln^m(Q_i/Q_j)$. A small coupling does not control a series when the logarithm compensates it.
3. Threshold consistency: if $Q$ is comparable to a heavy-quark mass, specify whether that quark is produced with its mass retained or absorbed into an active-flavor description, and match the two descriptions at their shared accuracy.
4. Leading-power control: estimate the first omitted terms, commonly $m^2/Q^2$, $\Lambda_{\mathrm{QCD}}^p/Q^p$, endpoint enhancements, or a factorization-breaking contribution.
5. Observable safety: sufficiently inclusive observables need real–virtual cancellation; exclusive observables need an infrared-safe measurement or an appropriate factorization theorem.

Scale pattern	Appropriate first response	What fixed order alone misses
$Q$ is the only perturbative scale	expand in $\alpha_s(Q)$	power corrections and threshold effects
$Q\gg q_T\gg\Lambda_{\mathrm{QCD}}$	resum recoil logarithms	rapidity evolution and large-impact-parameter input
$Q\gg Q\tau\gg\Lambda_{\mathrm{QCD}}$	factorize hard, jet, and soft scales	endpoint logarithms and nonperturbative soft physics
$x\ll1$ with $\alpha_s\ln(1/x)\sim1$	use matched high-energy evolution	powers of the high-energy logarithm
a relevant scale is $O(\!\left(\Lambda_{\mathrm{QCD}}\right))$	introduce nonperturbative input or change method	the long-distance matrix element itself

Checks and limitations

The domain test is auditable: list the scales; compute their ratios; state the active flavors and mass scheme; estimate the largest logarithmic counting parameter; identify the first power correction; and test the observable under an unresolved emission. Varying $\mu_R$ is then a sensitivity check inside the chosen description, not a substitute for any missing item.

Two limits are especially diagnostic. As $Q\to\infty$ at fixed dimensionless kinematics, masses and ordinary power corrections decouple and the running coupling decreases. By contrast, taking an endpoint variable to zero at fixed large $Q$ can create a soft scale that reaches the nonperturbative domain. “High energy” therefore does not uniquely select a method.

Common pitfalls

Using $Q\gg\Lambda_{\mathrm{QCD}}$ as the only criterion. A veto, small recoil, narrow jet, or endpoint can introduce a much lower scale and large logarithms. Inventory every measurement-induced scale before choosing fixed order.

Treating $n_f$ as particle census. It labels the effective theory used at a given scale. Threshold matching, not an abrupt substitution in a beta function, relates neighboring descriptions.

Reading the one-loop pole literally. It locates the failure of that perturbative solution. Nonperturbative infrared dynamics requires independent observables and methods.

Handoff

The output needed by subsequent pages is the tuple

$$\bigl\{G=SU(3),\ R_f=\mathbf3,\ \alpha_s^{(n_f)}(\mu),\ m_f^{\mathcal S}(\mu),\ Q_i, \ \text{power counting},\ \text{observable definition}\bigr\}.$$

For thermal QCD the same discipline continues, but the relevant hierarchy includes temperature-generated hard, electric, and magnetic scales; continue to scale hierarchies in hot gauge theories.

References

• Schwartz, Matthew D. Quantum Field Theory and the Standard Model. Cambridge University Press, 2014, §§25.2, 26.2, and 26.4–26.6, pp. 488–93, 505–13, and 517–28. DOI.