Path Integrals in Quantum Mechanics
Semiclassical approximation, instantons, and tunneling
Page 3 of 3
2.7 Semiclassical (WKB) Approximation from the Path Integral
The semiclassical limit arises naturally from the path integral by expanding around the classical solution. For a general potential V(x), we write $x(t) = x_{\text{cl}}(t) + y(t)$and expand the action to second order in fluctuations:
The linear term vanishes by the equations of motion. The second-order term defines the fluctuation operator:
The Gaussian integral over fluctuations gives:
The sum runs over all classical trajectories connecting the endpoints. This is the WKB approximation derived directly from the path integral, valid when $S_{\text{cl}} \gg \hbar$.
π‘Geometrical Meaning
The prefactor $(\det\hat{O})^{-1/2}$ measures how "focused" the bundle of nearby paths is around the classical trajectory. A large determinant means paths spread out quickly (weak focusing), giving a small amplitude. This is the path-integral version of the Van Vleck determinant in classical mechanics.
Multiple Classical Paths and Caustics
When multiple classical paths connect the same endpoints (e.g., the harmonic oscillator at times $T = n\pi/\omega$), the semiclassical propagator is a sum over all of them. Each path picks up an additional phase $e^{-i\mu\pi/2}$ where $\mu$ is the Morse index (the number of negative eigenvalues of $\hat{O}$, counting conjugate points):
2.8 Instantons in Quantum Mechanics
One of the most powerful applications of the Euclidean path integral is computing tunneling amplitudes between classically disconnected states.
The Double-Well Potential
Consider a symmetric double-well potential:
This has two degenerate minima at $x = \pm a$. Classically, a particle sitting at$x = -a$ can never reach $x = +a$ if its energy is below the barrier. Quantum mechanically, it can tunnel.
In the Euclidean path integral, the tunneling amplitude is dominated by solutions of the Euclidean equations of motion:
Note the crucial sign flip compared to Minkowski time. The Euclidean equation of motion describes a particle moving in the inverted potential $-V(x)$, where the barrier becomes a valley!
π‘Inverted Potential Trick
In real time, the particle cannot classically cross the barrier. But in Euclidean (imaginary) time, the potential is flipped: $V \to -V$. The barrier becomes a hill that the particle can roll over. This "Euclidean classical path" is the instanton.
The Instanton Solution
The instanton interpolates from $x = -a$ at $\tau = -\infty$ to $x = +a$ at$\tau = +\infty$. Using the first integral of the Euclidean equation of motion ($\frac{m}{2}\dot{x}^2 - V(x) = 0$ for zero energy):
where $\tau_0$ is the "center" of the instanton (a collective coordinate reflecting time-translation symmetry). The instanton is localized in Euclidean time with width$\sim 1/\omega$, hence the name: it happens in an "instant."
Instanton Action
The Euclidean action of the instanton is:
2.9 Functional Determinants and the Tunneling Amplitude
To compute the full tunneling amplitude, we need the Gaussian fluctuation determinant around the instanton. Expanding $x(\tau) = x_{\text{inst}}(\tau) + y(\tau)$:
The fluctuation operator has a zero mode $y_0(\tau) \propto \dot{x}_{\text{inst}}(\tau)$, corresponding to translations of the instanton center $\tau_0$. This zero eigenvalue must be treated separately by converting to a collective coordinate integral:
where the prime denotes omission of the zero mode. The factor $\sqrt{S_{\text{inst}}/(2\pi\hbar)}$comes from the Jacobian of the change of variables. The remaining determinant (with zero mode removed) gives a finite number that we call K:
Dilute Gas Approximation
For large Euclidean time $\beta\hbar$, the dominant configurations are sequences of widely separated instantons and anti-instantons. Summing over n instanton-anti-instanton pairs in the dilute gas approximation:
where the energy splitting between the symmetric and antisymmetric ground states is:
π‘Non-perturbative Physics
The tunneling splitting $\Delta E \propto e^{-S_{\text{inst}}/\hbar}$ is non-perturbative: it vanishes to all orders in perturbation theory since$e^{-1/g}$ has a zero Taylor expansion around g = 0. Instantons capture physics invisible to Feynman diagrams. This is why path integral methods are indispensable for tunneling, vacuum decay, and confinement in QCD.
2.10 Worked Example: Tunneling in the Double Well
Let us assemble the complete calculation for the quartic double well$V(x) = \lambda(x^2 - a^2)^2/4$ with $m = 1$.
Step 1: Identify the instanton
$x_{\text{inst}}(\tau) = a\tanh[\omega(\tau - \tau_0)/2]$ with $\omega = a\sqrt{2\lambda}$.
Step 2: Compute the instanton action
$S_{\text{inst}} = \int_{-a}^{a} dx\,\sqrt{2V(x)} = \frac{2\sqrt{2\lambda}\,a^3}{3} = \frac{\omega^3}{3\lambda}$
Step 3: Evaluate the fluctuation determinant
The fluctuation operator around the instanton has a Poschl-Teller potential. The ratio of determinants (with zero mode removed) gives:
Step 4: Assemble the result
Combining the zero-mode Jacobian with the fluctuation determinant:
Step 5: Energy splitting
This non-perturbative result captures the quantum tunneling between the two wells and has been verified numerically to excellent precision.
Key Concepts (This Page)
- Semiclassical approximation: expand to quadratic order around classical path
- WKB prefactor is $(\det\hat{O})^{-1/2}$, the Van Vleck determinant
- Instantons are finite-action solutions of Euclidean equations of motion
- Double-well instanton: $x(\tau) = a\tanh[\omega(\tau-\tau_0)/2]$
- Zero modes from symmetries require collective coordinate treatment
- Tunneling splitting $\Delta E \propto e^{-S_{\text{inst}}/\hbar}$ is non-perturbative
- Dilute instanton gas gives the complete tunneling amplitude