Skip to content

Quantum-mechanical instantons are finite-action Euclidean saddle points in 0+10+1 dimensions. They are paths x(τ)x(\tau) that solve the Euclidean equation of motion and connect classically distinct regions of configuration space.

In quantum field theory, the same idea becomes a finite-action Euclidean field configuration. The integration variable is no longer a path of one coordinate but a field history, such as

ϕ(τ,x)\phi(\tau,\mathbf x)

or a gauge field

Aμ(τ,x).A_\mu(\tau,\mathbf x).

Required background. Quantum-mechanical instantons supplies the finite-action saddle; fluctuation determinants explains Gaussian fluctuations and zero modes; the path-integral reference summarizes functional integration notation.

In quantum mechanics, the Euclidean action has the form

SE[x]=∫dτ[m2x˙2+V(x)].S_E[x] = \int d\tau \left[ \frac{m}{2}\dot x^2 + V(x) \right].

In a scalar field theory, the corresponding Euclidean action has the schematic form

SE[ϕ]=∫dτ ddx[12(∂μϕ)(∂μϕ)+V(ϕ)].S_E[\phi] = \int d\tau\,d^d x \left[ \frac12(\partial_\mu\phi)(\partial_\mu\phi) + V(\phi) \right].

The saddle condition is now a partial differential equation:

δSEδϕ=0.\frac{\delta S_E}{\delta\phi}=0.

The vacuum action must first be subtracted when its energy density is nonzero. Finite excess action then constrains the approach to a vacuum at Euclidean spacetime infinity. In theories with suitable vacuum manifolds or gauge structure, these boundary conditions can define topological classes; scalar bounces need not carry an integer topological charge.

The quantum-mechanical instanton connects two minima of a potential in Euclidean time. A field-theory instanton can connect distinct classical vacua or distinct topological sectors of the field configuration space.

The semiclassical contribution has the same basic structure:

contribution∼Ae−SE[ϕinst]/ℏ.\text{contribution} \sim A e^{-S_E[\phi_{\mathrm{inst}}]/\hbar}.

The action in the exponent comes from the classical Euclidean saddle. The prefactor AA comes from fluctuations, zero modes, gauge fixing when needed, and normalization.

For SU(N)SU(N) Yang–Mills theory on Euclidean R4\mathbb R^4, set ℏ=c=1\hbar=c=1 in this subsection. Use generators with tr⁡(TaTb)=δab/2\operatorname{tr}(T^aT^b)=\delta^{ab}/2, absorb the coupling into the gauge potential, and define Fμνa=∂μAνa−∂νAμa+fabcAμbAνcF_{\mu\nu}^a=\partial_\mu A_\nu^a-\partial_\nu A_\mu^a+f^{abc}A_\mu^bA_\nu^c. The action is

SE[A]=14g2∫d4x FμνaFμνa.S_E[A] = \frac{1}{4g^2} \int d^4x\, F_{\mu\nu}^aF_{\mu\nu}^a.

For smooth fields approaching a pure gauge at infinity with the usual compactifying boundary conditions, define F~μνa=12ϵμνρσFρσa\widetilde F_{\mu\nu}^a=\tfrac12\epsilon_{\mu\nu\rho\sigma}F_{\rho\sigma}^a, with ϵ1234=+1\epsilon_{1234}=+1. The topological charge is an integer:

Q=132π2∫d4x FμνaF~μνa.Q = \frac{1}{32\pi^2} \int d^4x\, F_{\mu\nu}^a\widetilde F_{\mu\nu}^a.

Self-dual or anti-self-dual configurations satisfy

Fμνa=±F~μνa,F_{\mu\nu}^a = \pm \widetilde F_{\mu\nu}^a,

and have action

SE=8π2g2∣Q∣.S_E = \frac{8\pi^2}{g^2} |Q|.

This formula is a preview of why instanton effects in weakly coupled gauge theory often carry factors such as

exp⁡(−8π2g2).\exp\left( - \frac{8\pi^2}{g^2} \right).

A quantum-mechanical instanton has a translational zero mode because its center τ0\tau_0 can be shifted. A field-theory instanton usually has more collective coordinates. Depending on the theory, these may include:

  • Euclidean position,
  • size or scale,
  • internal orientation,
  • gauge orientation,

Physical bosonic zero modes are replaced by collective-coordinate integrals. Pure gauge redundancies are handled by gauge fixing and Faddeev–Popov determinants; not every gauge variation is an additional physical modulus. Normalizable fermion zero modes instead produce Grassmann integrals. Without masses, interactions, or operator insertions to saturate them, their contribution to a given correlation function vanishes.

The fluctuation operator around a field-theory instanton is an operator on functions of Euclidean spacetime. Its determinant is much more singular than the one-dimensional determinant in quantum mechanics.

Renormalization enters because short-distance fluctuations affect the determinant and the running coupling. Thus the prefactor is not just a formal product of eigenvalues. It must be defined in a scheme compatible with the rest of the quantum field theory.

The size integral can also probe an infrared region where the running coupling is large. A classical instanton and a formal factor e−8π2/g2e^{-8\pi^2/g^2} do not by themselves establish a controlled dilute expansion of four-dimensional strongly coupled gauge theory.

This is one reason the quantum-mechanical double well is pedagogically valuable: it displays the saddle, zero-mode, and determinant logic without gauge redundancy or ultraviolet renormalization.

Field theory adds new issues:

  • infinitely many spatial degrees of freedom,
  • topology of field configurations,
  • gauge redundancy,
  • fermion zero modes and selection rules,
  • ultraviolet divergences and renormalization,
  • possible infrared problems,
  • multiple competing nonperturbative mechanisms.

For that reason, one should not transfer a quantum-mechanical instanton formula to field theory by analogy alone. The saddle logic transfers; the detailed measure and interpretation must be rebuilt in the field-theory setting.

For false-vacuum decay, Bounce Solutions supplies the zero-spatial-dimensional saddle and shows how the field-theory radial equation acquires a Euclidean friction term. False Vacuum Decay in Quantum Mechanics supplies the resonance, survival-law, and rate-per-volume dictionary.

  1. Compare the saddle equations in quantum mechanics and scalar field theory.
Solution

For quantum mechanics,

SE[x]=∫dτ[m2x˙2+V(x)],S_E[x] = \int d\tau \left[ \frac{m}{2}\dot x^2+V(x) \right],

and stationarity gives an ordinary differential equation,

mx¨=V′(x).m\ddot x=V'(x).

For a scalar field,

SE[ϕ]=∫dτ ddx[12(∂μϕ)(∂μϕ)+V(ϕ)],S_E[\phi] = \int d\tau\,d^d x \left[ \frac12(\partial_\mu\phi)(\partial_\mu\phi)+V(\phi) \right],

and stationarity gives a partial differential equation,

−∂μ∂μϕ+V′(ϕ)=0.- \partial_\mu\partial_\mu\phi + V'(\phi) = 0.
  1. Complete the square in the Yang–Mills action to derive its topological lower bound.
Solution

In four Euclidean dimensions, F~~=F\widetilde{\widetilde F}=F and F~2=F2\widetilde F^2=F^2. Thus

SE=18g2∫(F∓F~)2d4x ±8π2Qg2≥8π2∣Q∣g2,S_E=\frac{1}{8g^2}\int(F\mp\widetilde F)^2d^4x \ \pm\frac{8\pi^2 Q}{g^2} \geq\frac{8\pi^2|Q|}{g^2},

choosing the sign to match QQ. Equality requires self-duality for positive QQ or anti-self-duality for negative QQ. A generic finite-action field need not saturate the bound.

  1. A fermion zero mode contributes a Grassmann coordinate η\eta. Explain the difference between ∫dη 1\int d\eta\,1 and ∫dη η\int d\eta\,\eta.
Solution

Grassmann integration gives ∫dη 1=0\int d\eta\,1=0 and ∫dη η=1\int d\eta\,\eta=1. An unsaturated zero mode makes that contribution vanish. An appropriate fermion insertion or mass term can supply the missing factor. This is a selection rule, not the divergent Gaussian integral associated with a bosonic translation mode.