Chapter 7: Running Couplings & the Renormalization Group

The Beta Function

The beta function describes how a coupling constant $g$ changes with the renormalization scale $\mu$:

The beta function is computed from the $\mu$-dependence of the counterterms. In the $\overline{\text{MS}}$ scheme, it depends only on the coefficient of the$1/\varepsilon$ pole in the renormalization constants. At one loop, the bare coupling is:

Since $g_0$ is independent of $\mu$, differentiating with respect to $\mu$ and taking $\varepsilon \to 0$ yields the beta function from the residue of the simple pole:

The sign of the beta function determines the qualitative behavior of the theory:

QED Beta Function: Derivation

In QED, the physical charge is renormalized only through the photon wavefunction renormalization $Z_3$, because the Ward identity ensures $Z_1 = Z_2$:

The one-loop vacuum polarization gives $Z_3$ in the $\overline{\text{MS}}$ scheme:

where the sum runs over all charged fermions with color multiplicity $N_c^f$ and charge $Q_f$. Differentiating $e_0$ with respect to $\mu$ and using $de_0/d\mu = 0$, we extract:

In terms of $\alpha = e^2/(4\pi)$, this becomes:

For a single electron ($N_c = 1, Q = 1$), $\beta(\alpha) = 2\alpha^2/(3\pi) > 0$. The positive sign means the QED coupling grows with energy. Each additional charged fermion makes the beta function more positive, accelerating the growth.

Running Coupling Solution

The one-loop RG equation for the inverse coupling is linear and easily solved:

The Landau Pole

The running coupling diverges when $1/\alpha(\mu) \to 0$, defining the Landau pole:

For QED with electrons only, $\Lambda_\text{Landau} \sim 10^{286}$ GeV — absurdly far above the Planck scale ($\sim 10^{19}$ GeV). Thus the Landau pole is not a practical concern for QED. However, its existence signals that QED is not a fundamental theory valid at all energies — it must be embedded in a more complete framework (the Standard Model, or beyond) well before the Landau scale is reached.

QCD and Asymptotic Freedom

For non-abelian gauge theories like QCD based on $SU(3)$, the gluon self-interactions contribute to the beta function with a sign opposite to the fermion contribution:

where $N_f$ is the number of active quark flavors. The crucial factor of 11 comes from the gluon self-coupling — the non-abelian structure of the gauge group. For $N_f \leq 16$(the Standard Model has $N_f = 6$), the beta function is negative, giving asymptotic freedom.

The discovery of asymptotic freedom by Gross, Wilczek, and Politzer (Nobel Prize 2004) was one of the pivotal moments in theoretical physics. It explained why quarks behave as nearly free particles in deep inelastic scattering at high energies, while being permanently confined inside hadrons at low energies.

The QCD coupling runs from $\alpha_s(M_Z) \approx 0.118$ down to arbitrarily small values at high energies, but grows toward $\alpha_s \sim 1$ near $\Lambda_\text{QCD} \approx 330$ MeV, where perturbation theory breaks down and confinement sets in.

Callan-Symanzik Equation

The Callan-Symanzik (CS) equation expresses the requirement that physical observables (specifically, renormalized Green's functions) are independent of the renormalization scale $\mu$. For an $n$-point correlation function:

where $\gamma(g)$ is the anomalous dimension of the field:

The anomalous dimension modifies the naive scaling dimension of the field. In a free theory, the scalar field has dimension 1 (in 4D). Interactions modify this to $1 + \gamma$, where $\gamma$ is typically a small perturbative correction. At a fixed point of the RG flow, the anomalous dimension determines the critical exponents of the theory.

Solution by Method of Characteristics

The CS equation is a first-order PDE solved by the method of characteristics. Define the running coupling $\bar{g}(\mu')$ by:

Then the Green's function at scale $\lambda\mu$ is related to that at scale $\mu$ by:

This shows that changing the energy scale is equivalent to changing the coupling constant (via the running coupling) and multiplying by the anomalous dimension factor.

RG Flow and Fixed Points

The renormalization group flow describes how the effective theory changes as we integrate out high-energy degrees of freedom. In the space of all possible coupling constants$\{g_i\}$, the RG flow is governed by:

Fixed points $g_i^*$ where all $\beta_i = 0$ correspond to scale-invariant theories (conformal field theories in many cases). The behavior near a fixed point determines whether couplings are relevant, marginal, or irrelevant:

The concept of universality emerges naturally: different microscopic theories that flow to the same IR fixed point share the same long-distance physics. This explains why widely different physical systems (fluids near critical points, magnets at phase transitions) exhibit identical critical exponents.

Decoupling and Matching

The Appelquist-Carazzone decoupling theorem states that heavy particles of mass $M$ decouple from low-energy physics at scales $\mu \ll M$. Their effects are suppressed by powers of $\mu/M$ and can be absorbed into the couplings of the effective low-energy theory.

In practice, when running couplings across a heavy particle threshold (e.g., the bottom quark mass $m_b \approx 4.18$ GeV), we match the coupling at $\mu = m_b$:

Above $m_b$, we use the 5-flavor beta function; below, the 4-flavor version. The matching conditions at each threshold ensure continuity of the coupling and can be computed perturbatively. At two loops, the matching includes scheme-dependent terms that must be handled carefully for precision phenomenology.

Dimensional Transmutation

A remarkable consequence of the running coupling is dimensional transmutation: a dimensionless coupling constant is traded for a dimensionful scale. In QCD, the one-loop running coupling can be written as:

The parameter $\Lambda_\text{QCD}$ is an intrinsic mass scale of QCD, generated dynamically from the dimensionless coupling. It sets the scale for all non-perturbative QCD phenomena: confinement, chiral symmetry breaking, and hadron masses. Even though the QCD Lagrangian with massless quarks contains no dimensionful parameter, the quantum theory generates a mass scale.

This is the essence of dimensional transmutation: the scale invariance of the classical theory is broken by quantum effects (the conformal anomaly or trace anomaly), generating a dynamical scale. The proton mass $m_p \approx 938$ MeV comes almost entirely from the binding energy of quarks and gluons — the quark masses contribute only about 1% of $m_p$. Thus, most of the mass of visible matter in the universe arises from dimensional transmutation in QCD.

Scheme Dependence of $\Lambda$

The numerical value of $\Lambda_\text{QCD}$ depends on the renormalization scheme and the number of active flavors. In the $\overline{\text{MS}}$ scheme with $N_f = 5$: $\Lambda_{\overline{\text{MS}}}^{(5)} \approx 213$ MeV. When matching across flavor thresholds, the coupling is continuous but $\Lambda$ jumps:

Physical observables are of course independent of both the scheme and the number of active flavors used in the calculation.

Higher-Order Beta Functions

The two-loop beta function for QCD provides important corrections:

where the two-loop coefficient is:

The first two coefficients $b_0$ and $b_1$ are scheme-independent (they are the same in $\overline{\text{MS}}$, momentum subtraction, or any other mass-independent scheme). Starting at three loops, the coefficients depend on the renormalization scheme.

The QCD beta function is currently known to five loops, with $b_4$ involving contributions from thousands of Feynman diagrams. This extraordinary perturbative accuracy is essential for precision extraction of $\alpha_s$ from collider data.

Banks-Zaks Fixed Point

For QCD-like theories with many flavors but $N_f < 33/2$ (so that $b_0 > 0$), the two-loop beta function can have a non-trivial zero at:

This is the Banks-Zaks fixed point: an infrared-stable conformal field theory where the coupling flows to a finite value. For $N_f$ just below $33/2$, the fixed-point coupling is small and perturbatively accessible. This "conformal window" is of great interest for beyond-Standard-Model physics and lattice studies.

Gauge Coupling Unification

One of the most suggestive hints for physics beyond the Standard Model comes from the running of the three gauge couplings. When extrapolated to high energies, the couplings approach each other, hinting at a unified gauge group at $\sim 10^{15-16}$ GeV.

The SM gauge couplings at $M_Z$ in the GUT normalization are:

In the Standard Model alone, the three couplings do not quite meet at a single point. However, in the Minimal Supersymmetric Standard Model (MSSM), the additional particle content modifies the running, and the three couplings unify beautifully at $M_\text{GUT} \approx 2 \times 10^{16}$ GeV with $\alpha_\text{GUT}^{-1} \approx 24$. This quantitative success of supersymmetric grand unification remains one of the strongest indirect arguments for low-energy supersymmetry.

Grand unified theories (GUTs) based on groups like $SU(5)$ or $SO(10)$embed all SM fermions of one generation into a single representation, explaining the observed quantum number patterns and anomaly cancellation. GUTs also predict proton decay at rates that are being probed by current experiments (Super-Kamiokande).

Summary: Running Couplings & RG