research32 min studyLesson 704 in Physics

Faddeev–Popov determinant and ghost action

Interpret gauge-fixing Jacobians and Grassmann determinants within a local perturbative patch.

Before you begin

On this page

The question for today

What you’ll understand

Construct the local Faddeev-Popov operator from an infinitesimal gauge condition.
Distinguish Grassmann determinant factors from commuting Gaussian factors.
State when zero modes and multiple gauge intersections invalidate a naive global formula.

A path integral over gauge potentials counts many descriptions of each physical configuration. Gauge fixing attempts to choose representatives, while a Jacobian compensates the distortion of the gauge-orbit coordinates. The Faddeev-Popov construction makes this precise locally when the chosen slice crosses the relevant orbits transversely. Anticommuting ghost fields represent the resulting determinant. They are essential internal bookkeeping fields in covariant perturbation theory, not additional observable scalar particles or a physical mechanism that creates gauge redundancy.

Differentiate the gauge condition along an orbit

Choose a Euclidean Lorenz condition G equal to the ordinary divergence of the connection. Under the infinitesimal transformation defined earlier, its variation is partial D acting on the gauge parameter. Define the Faddeev-Popov operator M with a conventional overall minus sign so its free part is minus the Euclidean Laplacian. A gauge-independent overall determinant sign can be absorbed into normalization in a chosen local patch; field-dependent sign changes or zeros require more care.

The Jacobian is invertible only after removing allowed residual gauge zero modes, such as constant transformations in an appropriate finite-volume setting. With that qualification, the implicit-function picture supports a nearby representative. It does not establish that no other representative lies elsewhere on the same orbit. Boundary conditions determine the operator domain and spectrum, so merely writing a differential expression is not yet a determinant definition.

G^a[A]=\partial_\mu A_\mu^a,\qquad \delta G=\partial_\mu D_\mu\alpha,\qquad M=-\partial_\mu D_\mu

This operator convention is Euclidean, with adjoint D inherited from the declared connection. M has inverse-length-squared units; residual modes and determinant regularization must be specified.↔ Scroll wide equations horizontally.

Represent a determinant with anticommuting fields

Introduce independent Grassmann-valued ghost and antighost fields in the adjoint representation. Their Gaussian integral gives a determinant of M, with an overall normalization fixed by an integration-order convention. In contrast, a normalized complex commuting Gaussian gives the reciprocal determinant when its matrix is positive. The different powers explain why replacing ghosts by ordinary scalar integration would change the gauge-field weight.

In an Abelian linear covariant gauge, M is independent of the gauge field and its determinant can be absorbed into normalization. In a non-Abelian theory, D contains A, generating a ghost-gauge interaction. Ghosts are Lorentz scalars with anticommuting statistics because they represent a Jacobian, not asymptotic physical particles to which an ordinary spin-statistics interpretation is applied. Closed ghost loops acquire the corresponding Grassmann sign in perturbative diagrams.

\det M=\int\mathcal D\bar c\,\mathcal Dc\,e^{-S_{\rm gh}},\qquad S_{\rm gh}=\int d^4x\,\bar c^a M^{ab}c^b

The equality is a regulated formal identity up to field-independent normalization and Grassmann measure convention. Four-dimensional canonical ghost fields can be assigned inverse-length units so the action is dimensionless.↔ Scroll wide equations horizontally.

Separate local validity from global gauge fixing

In a finite-dimensional analogy, a nonzero Jacobian proves only that a coordinate map is locally invertible. Multiple distant roots can each have a nonzero derivative. Gauge-orbit intersections have the same logical distinction. A zero eigenvalue of M identifies an infinitesimal direction not fixed by the condition, while more distant copies can exist even where the local determinant is nonzero. The determinant alone cannot certify a globally unique representative.

Perturbative calculations usually work in a specified neighborhood and treat the determinant as an oriented local Jacobian with a compatible prescription. A fully global change of variables would require accounting for multiple intersections and possible absolute-value issues. Gribov copies make this a substantial nonperturbative question. One should therefore state the patch, removed zero modes, and regulator instead of claiming that the ghost action has solved every global counting problem.

\int\frac{d^{2n}z}{\pi^n}\,e^{-z^\dagger Mz}=\frac1{\det M}\qquad(M=M^\dagger>0)

This comparison is an ordinary finite-dimensional complex Gaussian with positive Hermitian matrix M. It illustrates a determinant power and is not itself a gauge-field path integral.↔ Scroll wide equations horizontally.

Work it through

Compare two finite Gaussian weights

Use the dimensionless positive matrix $M=\operatorname{diag}(2,3)$ . Compute the normalized Grassmann Gaussian factor and normalized complex commuting Gaussian factor. What happens to these factors if the second eigenvalue tends to zero from above?

The determinant is the product of eigenvalues, $2\times3=6$ . With a consistent normalization, the Grassmann integral therefore contributes six.
The normalized commuting Gaussian contributes $1/\det M=1/6$ . Each complex Gaussian mode supplies the inverse of its positive eigenvalue, so exchanging statistics reverses the determinant power.
If the second eigenvalue is replaced by positive epsilon, the determinant becomes $2\epsilon$ . The Grassmann factor tends to zero while the commuting integral diverges. A zero mode means the corresponding naive Gaussian or coordinate construction must be reconsidered.

The result

The two factors are six and one sixth; a vanishing eigenvalue exposes a singular local construction rather than a harmless change of normalization.

The living notebook

Explore the model

Interactive lab

Interactive lab · matrix

A determinant illustration with a restricted family

Reset and set diagonal scale $a=1$ , shear $b=1$ ; predict selected final determinant one. Compare a fresh Reset run at $a=0.5,b=1$ ; predict determinant one half. At Reset the geometry is identity and Play or Step interpolates it, while readouts always describe the selected final matrix. This positive triangular family never reaches a zero determinant and is not a Faddeev-Popov operator or ghost integral.

Interpolated geometryOriginal unit squareSelected final transform

Area scale · det A: 1.4
Singular values · σmax, σmin: 1.597, 0.876
Condition number · κ₂(A): 1.823

Diagonal scale a1.4Shear b0.6

t = 0 s

The selected matrix is A = [[a,b],[0,1]]. Playback interpolates geometry between the identity and A; readouts always describe A. Its determinant is a and its singular values come from AᵀA. Positive a makes every supported matrix invertible. This two-dimensional family is not a general eigensolver or a learned network. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.

Look forA determinant belongs to a specified matrix; this restricted lab cannot discover gauge zero modes or prove global gauge uniqueness.

Pause & reason

Check your understanding

Interactive checks need JavaScript. You can still read the explanation.

Read the checkpoint explanation

The determinant identity fixes the bookkeeping statistics. Ghosts belong to the gauge-fixed internal description, so they are not ordinary observable scalar asymptotic particles.

Your turn

Try it yourself

For the dimensionless matrix $M=\begin{pmatrix}2&1\\1&2\end{pmatrix}$ , find eigenvalues, determinant, and the two normalized Gaussian factors. Does a nonzero determinant by itself prove a unique representative everywhere on a gauge orbit?

Take a moment to make a prediction or work it out on paper. Open a hint when you need one.

Give me a hint

Use the symmetric and antisymmetric vectors, then distinguish invertibility at one point from uniqueness of distant roots.

Compare with the solution

The normalized symmetric direction has eigenvalue three and the antisymmetric direction eigenvalue one. Hence $\det M=3$ , the Grassmann factor is three, and the normalized commuting factor is one third. The matrix is locally invertible. This finite statement cannot establish global uniqueness of a nonlinear gauge condition: separate orbit intersections can each have an invertible local Jacobian.

Keep the idea

Take it with you

The Faddeev-Popov operator differentiates a gauge condition along gauge orbits.
Grassmann ghosts supply a determinant with the required sign structure.
Local determinant invertibility is weaker than global gauge fixing.

Read further

Original sources and deeper explanations to follow at your own pace.

Optional progress, saved in this browser when available. Revisit any lesson at any time.

Lesson not marked complete.