---
title: Krein Regularization in QFT
url: https://www.emergentmind.com/topics/krein-regularization
type: topic
---

# Krein Regularization in QFT

Krein regularization is a quantum field theory (QFT) methodology that exploits the structure of Krein spaces—indefinite inner-product spaces combining positive- and negative-norm sectors—to achieve built-in ultraviolet (UV) and infrared (IR) finiteness. By quantizing fields in a Krein space and incorporating quantum metric fluctuations, the method automatically cancels UV divergences without introducing auxiliary heavy regulator fields or explicit counterterms. Originally motivated by technical and structural obstacles in conventional renormalization, Krein regularization has been demonstrated explicitly in scalar field theory, quantum electrodynamics (QED), and settings involving quantum gravity and curved spacetime.

## 1. Krein Space Quantization and Field Operators

In Hilbert-space quantization, the standard field operator for a real scalar field φ(x) is constructed solely from positive-frequency modes,
\[
\phi_H(x) = \int d^3k \; [a(k) u_p(k, x) + a^\dagger(k) u_p^*(k, x)],
\]
where \(u_p(k, x)\) is a positive-energy solution with norm \(+\). The associated Feynman propagator
\[
G_F(k) = \frac{1}{k^2 - m^2 + i\epsilon}
\]
exhibits singularities leading to divergent loop integrals.

By contrast, Krein space quantization extends the operator to include negative-norm (ghost) modes [1206.3934, 2112.05390]:
\[
\phi_K(x) = \phi_p(x) + \phi_n(x),
\]
with
\[
\phi_n(x) = \int d^3k \; [b(k) u_n(k, x) + b^\dagger(k) u_n^*(k, x)],
\]
where \(u_n\) encodes negative-frequency solutions, and \(b, b^\dagger\) satisfy commutators with the negative sign. The presence of a complete negative-norm sector renders Krein spaces indefinite and enables exact cancellation of certain pathological contributions in loop diagrams.

The two-point time-ordered function in Krein space is given by
\[
i G_T(x, x') = \langle 0 | T \phi_K(x) \phi_K(x') | 0 \rangle = \text{Re} \, G_F(x, x'),
\]
resulting in a propagator that is the real part of the conventional Feynman propagator. In momentum space, this yields the principal value:
\[
G_T(k) = P \frac{1}{k^2 - m^2},
\]
where \(P\) denotes Cauchy principal value. This automatically removes the imaginary-part (delta-function) singularities responsible for UV/IR divergences in loop integrals [1206.3934, 2112.05390, 1206.2796].

Upon inclusion of quantum metric fluctuations (notably, Ford’s fluctuated light-cone prescription), Krein regularization prescribes replacing the propagator with
\[
G_K(k) = m^2 P \frac{1}{k^2 (k^2 - m^2)},
\]
which further smears light-cone singularities and ensures analytic behavior on the entire mass shell [1206.3934, 1206.2796, 1306.0383].

## 2. Finiteness of Loop Integrals: Mechanism and Examples

Krein regularization is characterized by automatic cancellation of UV and IR divergences in loop integrals as a result of the negative-norm sector. Every internal line in a potentially divergent diagram is replaced by the Krein regularized propagator, either \(P 1/(k^2-m^2)\) or \(m^2 P [1/(k^2 (k^2-m^2))]\), depending on the singularity structure [2112.05390, 1206.2796].

In λφ⁴ theory, the one-loop four-point function for all channels (s, t, u) takes the schematic form
\[
T_K^{(4)}(p) = (-i\lambda)^2/2 \int \frac{d^4k}{(2\pi)^4} \left[ P\left( \frac{1}{k^2 - m^2} - \frac{1}{k^2} \right) \right] \left[ P \left(\frac{1}{(k+p)^2 - m^2} - \frac{1}{(k+p)^2} \right) \right].
\]
Expansion of the integrand reveals that all terms leading to divergent behavior at large \(k\) are cancelled [1206.3934]. Employing Feynman parametrization and Wick rotation, the integral reduces after algebraic cancellation to the fully finite, manifestly covariant form:
\[
T_K^{(4)}(p) = -\frac{3i\lambda^2}{32\pi^2} \int_0^1 dx \, \ln \left[1 - x(1-x)\frac{p^2}{m^2}\right],
\]
which is regular for all values of external momentum and mass.

Analogous cancellation occurs in QED. The vacuum polarization, electron self-energy, and vertex function all become finite due to the principal-part subtraction of UV and IR poles [1206.2796, 1202.2416]. For example, the self-energy is rendered finite without the need for counterterms:
\[
\Sigma_K(p) = \frac{e^2}{8\pi^2} \int_0^1 dx \; [ x \slashed{p} - 2m] \ln \Big(\frac{x(x-1)p^2+(1-x)m^2}{x(x-1)p^2}\Big).
\]
No explicit cutoff or dimensional regularization is necessary.

## 3. Comparison with Pauli–Villars and Conventional Regularizations

Krein regularization is mathematically analogous to Pauli–Villars regularization, with the negative-norm states in Krein space playing the role of regulator fields. However, all regulators have the same mass as the physical field, and there is no explicit mass scale separation. One can schematically write, for a Krein-subtracted propagator:
\[
G_{\text{Krein}}(k) = G_{p}(k) - G_{n}(k) = P \frac{1}{k^2 - m^2}.
\]
This structure achieves self-contained subtraction, rendering all one-loop QED amplitudes finite [1202.2416, 1206.2796].

Unlike dimensional regularization or Pauli–Villars, Krein regularization does not introduce auxiliary Lagrangian terms, alter the bare action, or require an analytic continuation in spacetime dimension. It also maintains strict four-dimensional covariance and gauge invariance [1206.2796].

A table summarizing correspondence between the approaches:

| Feature              | Pauli–Villars                  | Dimensional Regularization      | Krein Regularization                            |
|----------------------|--------------------------------|---------------------------------|-------------------------------------------------|
| Regulator Fields     | Explicit, large mass           | None                            | Intrinsic negative-norm, same mass              |
| Counterterms Needed  | Yes                            | Yes                             | No (at one loop; all divergences cancel)        |
| Covariance           | Preserved                      | Preserved                       | Preserved                                       |
| Spacetime Dim        | 4                              | d = 4-ε (analytic)              | 4                                               |

## 4. Renormalization, β-Function, and Physical Parameters

While Krein regularization produces UV-finite loop integrals, physical mass and coupling constants are still defined by renormalization conditions. One imposes the conventional normalization of two- and four-point functions at chosen renormalization points:
\[
\Gamma^{(2)}(p^2 = \mu^2) = -(p^2 - m^2), \qquad \Gamma^{(4)}(s = \mu^2, t=0, u=0) = -\lambda_\mu.
\]
Residual finite shifts are absorbed into the definition of physical parameters, and running couplings are extracted by evaluating loop amplitudes at the appropriate subtraction point [2112.05390, 1206.3934].

In λφ⁴ theory, the effective coupling is determined as
\[
\lambda_{\mu} = \lambda + \frac{3 \lambda^2}{32\pi^2} \int_0^1 dx \ln\left[1 +

Source: https://www.emergentmind.com/topics/krein-regularization