Renormalization

Summary

Perturbative calculations in QFT produce divergent (infinite) integrals corresponding to virtual particles with arbitrarily high momenta. Renormalization systematically removes these divergences by absorbing them into a redefinition of physical parameters (mass, charge). The result is a finite, predictive theory — and the celebrated prediction of the electron’s anomalous magnetic moment demonstrates its extraordinary accuracy.

Overview

When Dirac and others first computed higher-order corrections in QED, they found infinite results. For decades these infinities seemed to signal a breakdown of the theory. The breakthrough came around 1950: Schwinger, Feynman, Dyson, and Tomonaga showed that all infinities in QED can be absorbed into the measured values of mass and charge — a procedure called renormalization. This is now understood as reflecting the fact that physical parameters observed at low energies differ from “bare” parameters in the Lagrangian due to quantum corrections.

The Problem: UV Divergences

In perturbation theory, higher-order Feynman diagrams involve loop integrals over all internal momenta:

$${\int }_0^{\infty }{d}^{4}k(\text{propagator}{)}^n$$

These integrals often diverge as $k\to \infty$ — these are ultraviolet (UV) divergences. They arise because QFT assumes the theory is valid at all energy scales, which is physically unreasonable.

Examples of divergent quantities:

• Electron self-energy: $m_{\text{phys}}-m_{\text{bare}}\to \infty$
• Vacuum polarization (photon self-energy)
• Vertex corrections

The Solution: Renormalization

Renormalization

Renormalization is the procedure of systematically removing UV divergences by:

1. Regularization: Introduce a cutoff $\Lambda$ (or use dimensional regularization) to make integrals finite
2. Absorb divergences: Rewrite the Lagrangian as $\mathcal{L}={\mathcal{L}}_{\text{ren}}+{\mathcal{L}}_{\text{counter}}$, where counter-terms cancel the divergences
3. Renormalization conditions: Fix the finite parts by requiring physical parameters (mass, charge) match their measured values
4. $\Lambda \to \infty$: Take the limit; all physical predictions are finite

As Tomonaga explained: “The mass and charge observed in experiments are not the original [bare] mass and charge but the mass and charge as modified by field reactions, and they are finite.”

Renormalizability

Not all QFTs can be renormalized. Dyson proved in 1949:

Dyson's Renormalizability Criterion

A theory is renormalizable if all UV divergences can be absorbed into a finite number of redefinitions of physical parameters. This requires that all coupling constants have non-negative mass dimension.

• Renormalizable: QED, QCD, electroweak theory, ${\varphi }^4$ theory
• Non-renormalizable: Fermi theory of the weak interaction, general relativity as a QFT

Non-renormalizability doesn’t mean a theory is wrong — it means it is an effective field theory, valid only up to some energy scale.

Running Coupling Constant

Renormalization reveals that coupling constants depend on the energy scale $\mu$ at which they are measured. This is the running coupling:

$$\frac{d\alpha }{d\ln \mu }=\beta (\alpha )$$

where $\beta (\alpha )$ is the beta function, computed from loop diagrams.

Asymptotic Freedom

In QCD (the theory of the strong force), the coupling constant ${\alpha }_s$ decreases as energy increases:

$${\alpha }_s(\mu )\to 0\text{ as }\mu \to \infty$$

This is called asymptotic freedom, discovered by Gross, Wilczek, and Politzer (1973 Nobel Prize). At high energies, quarks interact weakly and perturbation theory is valid; at low energies, the coupling is large — quarks are confined inside hadrons.

Conversely, in QED, the fine-structure constant $\alpha$ increases at higher energies (Landau pole), but this is at energy scales far beyond any experiment.

Key Prediction: Anomalous Magnetic Moment

The most precise test of QFT: the electron’s anomalous magnetic moment $a_e=(g-2)/2$.

Order	QED prediction	Experimental value
1-loop	$\alpha /(2\pi )\approx 0.00116$	—
5-loop	$a_e\approx 0.00115965218073$	$0.00115965218059\pm 0.00000000000028$

Agreement to 12 significant figures — the most precise prediction in all of science.

The Lamb Shift

The Lamb shift (1947) was the experimental trigger for renormalization:

• Lamb and Retherford measured a tiny energy difference between $2S_{1/2}$ and $2P_{1/2}$ levels in hydrogen that should be degenerate by the Dirac equation
• Bethe estimated this shift by noting that virtual photons with energies above the electron mass contribute negligibly; Schwinger/Feynman made it exact
• This confirmed that vacuum fluctuations are real and that renormalized QFT is predictive

Connections

• Canonical Quantization of Fields: Loop integrals over virtual quanta (created and destroyed by ladder operators) produce the divergences that renormalization cures.
• Gauge Theory Overview: Ward identities in gauge theories constrain the renormalization; they ensure that gauge invariance is preserved after renormalization.
• Yang-Mills Theory and Gauge Fields: Non-abelian gauge theories are renormalizable (‘t Hooft, 1971); asymptotic freedom makes QCD tractable.

See Also

• QFT Overview — broader context for renormalization
• Canonical Quantization of Fields — source of the divergent integrals
• Yang-Mills Theory and Gauge Fields — asymptotic freedom in QCD