The connection between Hodge theory and quantum field theory propagators is deep and multifaceted. At its core, both subjects concern the inversion of Laplacian-type operators, but the relationship extends into gauge fixing, BRST cohomology, and the geometry of field space.
The Basic Parallel
In Hodge theory on a Riemannian manifold \((M, g)\), the central object is the Hodge Laplacian acting on differential forms:
\[ \Delta = d\delta + \delta d \]
where \(d\) is the exterior derivative and \(\delta = (-1)^{n(k+1)+1} \star d \star\) is its formal adjoint (the codifferential). The Hodge decomposition theorem states that on a compact manifold, any \(k\)-form \(\omega\) decomposes uniquely as
\[ \omega = d\alpha + \delta\beta + \gamma \]
where \(\gamma\) is harmonic (\(\Delta\gamma = 0\)). Equivalently,
\[ \Omega^k(M) = \mathrm{im}(d) \oplus \mathrm{im}(\delta) \oplus \mathcal{H}^k \]
as an orthogonal direct sum with respect to the \(L^2\) inner product.
In QFT, the propagator for a free field is the Green’s function of the kinetic operator. For a scalar field with action
\[ S[\phi] = \frac{1}{2}\int \phi(-\Box + m^2)\phi \, d^nx \]
the propagator \(G(x, y)\) satisfies
\[ (-\Box_x + m^2)G(x, y) = \delta(x - y) \]
Both situations involve inverting a second-order elliptic operator. The key subtlety in both cases is the kernel: harmonic forms in Hodge theory, zero modes in QFT.
Green’s Operators and the Hodge Decomposition
On a compact Riemannian manifold, \(\Delta\) is not invertible due to the harmonic forms. However, we can define the Green’s operator \(G\) as the inverse of \(\Delta\) restricted to the orthogonal complement of \(\mathcal{H}^k\). If \(H\) denotes orthogonal projection onto harmonic forms, then
\[ \Delta G = G \Delta = I - H \]
The Hodge decomposition can then be written as
\[ \omega = d\delta G\omega + \delta d G\omega + H\omega \]
This is directly analogous to how propagators in QFT must handle zero modes. In a finite-dimensional analogy, if \(A\) is a matrix with a kernel, the “propagator” is the pseudoinverse \(A^+\) satisfying \(AA^+ = I - P_{\ker A}\).
Gauge Theory: Where the Connection Becomes Essential
The relationship becomes most concrete in gauge theories, where Hodge theory is not merely analogous but directly applicable.
Electromagnetism
Consider the free electromagnetic field on a Riemannian 4-manifold. The gauge potential \(A\) is a 1-form, the field strength is \(F = dA\), and the action is
\[ S[A] = \frac{1}{4}\int F \wedge \star F = \frac{1}{2}\int |dA|^2 \, \mathrm{dvol} \]
The equation of motion is \(\delta F = \delta dA = 0\). Combined with the Bianchi identity \(dF = 0\), this says \(F\) is harmonic as a 2-form.
The gauge symmetry \(A \mapsto A + d\lambda\) means the kinetic operator \(\delta d\) (acting on 1-forms) has a huge kernel: all exact 1-forms. To define a propagator, we must fix a gauge.
Lorenz Gauge and the Hodge Laplacian
The Lorenz gauge condition \(\delta A = 0\) restricts \(A\) to coclosed 1-forms. On this subspace, the equation of motion becomes
\[ \Delta A = (d\delta + \delta d)A = \delta d A = 0 \]
But we’ve imposed \(\delta A = 0\), so in fact \(d\delta A = 0\) automatically and we need \(\delta dA = 0\). To get a non-degenerate operator, we work with the full Laplacian \(\Delta\) on the gauge-fixed subspace.
The propagator in Lorenz gauge is essentially the Green’s operator for \(\Delta\) acting on 1-forms, projected appropriately. The Hodge decomposition
\[ \Omega^1 = d\Omega^0 \oplus \delta\Omega^2 \oplus \mathcal{H}^1 \]
separates pure gauge modes (\(d\Omega^0\)), physical transverse modes (\(\delta\Omega^2\)), and harmonic modes related to topology.
The Faddeev-Popov Procedure
The Faddeev-Popov method for gauge fixing in the path integral has a natural Hodge-theoretic interpretation. The gauge orbit through a connection \(A\) is
\[ \mathcal{O}_A = \{A + d\lambda : \lambda \in \Omega^0\} \]
The gauge-fixing condition \(\delta A = 0\) picks out a representative in each orbit (at least locally). The Faddeev-Popov determinant
\[ \det(\delta d) \big|_{\Omega^0} \]
is the Jacobian for this slice and equals the determinant of the Laplacian on 0-forms. This is computable via the zeta function regularisation or heat kernel methods—both deeply connected to Hodge theory.
Heat Kernels: The Bridge
The heat kernel \(K_t(x, y)\) satisfying
\[ \left(\frac{\partial}{\partial t} + \Delta_x\right)K_t(x, y) = 0, \quad K_0(x,y) = \delta(x,y) \]
connects the two perspectives beautifully. The Green’s operator has the representation
\[ G = \int_0^\infty (e^{-t\Delta} - H) \, dt \]
where \(H\) is the projection onto harmonic forms (the \(t \to \infty\) limit of \(e^{-t\Delta}\)).
In QFT, this becomes Schwinger’s proper time representation of the propagator:
\[ \frac{1}{-\Box + m^2} = \int_0^\infty e^{-t(-\Box + m^2)} \, dt \]
The heat kernel \(e^{t\Box}\) has an asymptotic expansion whose coefficients are local geometric invariants. This is the basis for computing anomalies, one-loop effective actions, and index theorems in QFT.
BRST Cohomology
The most sophisticated connection comes through BRST quantisation. The BRST operator \(Q\) satisfies \(Q^2 = 0\) and acts on an extended space including ghosts. Physical states are defined as \(Q\)-cohomology classes:
\[ \mathcal{H}_{\text{phys}} = \frac{\ker Q}{\mathrm{im}\, Q} \]
This is directly analogous to de Rham cohomology, where \(H^k(M) = \ker d / \mathrm{im}\, d\).
Hodge theory enters by introducing a “co-BRST” operator \(Q^\dagger\) and forming the BRST Laplacian \(\{Q, Q^\dagger\}\). Physical states can be identified with harmonic representatives in each cohomology class. The propagator in the extended (ghost + field) space involves the Green’s operator for this BRST Laplacian.
Non-Abelian Gauge Theory
For Yang-Mills theory, the story is richer. The connection \(A\) is now Lie algebra-valued, and gauge transformations act as
\[ A \mapsto g^{-1}Ag + g^{-1}dg \]
The space of connections \(\mathcal{A}\) is an affine space, and the gauge group \(\mathcal{G}\) acts on it. The physical configuration space is the quotient \(\mathcal{A}/\mathcal{G}\), which generally has singularities (at reducible connections) and nontrivial topology.
Hodge theory on this infinite-dimensional space—while not rigorous in general—provides the conceptual framework. The gauge-fixing condition \(\delta A = 0\) (in a background field) defines a slice transverse to gauge orbits. The Faddeev-Popov operator
\[ \Delta_{\text{FP}} = \delta d_A : \Omega^0(\mathrm{ad}\, P) \to \Omega^0(\mathrm{ad}\, P) \]
where \(d_A = d + [A, \cdot]\) is the covariant exterior derivative, generalises the Hodge Laplacian on 0-forms. Its determinant appears in the path integral measure, and its zero modes (related to global gauge transformations and Gribov copies) create subtleties.
Topological Field Theory
In topological QFTs, the connection becomes an identity. Consider Witten-type TQFTs where the action is \(Q\)-exact:
\[ S = \{Q, V\} \]
for some functional \(V\). The partition function localises onto \(Q\)-fixed points, and correlation functions depend only on cohomological data.
The canonical example is Witten’s topological Yang-Mills, where the path integral computes Donaldson invariants—intersection numbers on the moduli space of anti-self-dual connections. The entire theory is essentially Hodge theory on field space, with the propagator determining how fluctuations around classical solutions contribute.
Summary
The relationship is not merely an analogy: in gauge theories, Hodge theory provides the mathematical framework for understanding gauge fixing, the structure of the propagator, and the cohomological nature of physical observables. The infinite-dimensional and analytical subtleties of QFT mean the correspondence isn’t always rigorous, but it remains the guiding geometric principle.