Semiclassical Limit Overview

A semiclassical approximation keeps the quantum amplitude structure but organizes it around classical action. It is not the same as simply replacing quantum mechanics by classical mechanics. The approximation still has phases, interference, tunneling, prefactors, and boundary conditions.

The broad Classical Limit page explains several mechanisms that make quantum predictions look classical. This page focuses on one major family: approximations controlled by action scales large compared with $\hbar$.

For the classical coordinate-momentum state space used below, see Phase Space.

Quantization vs Classical Limit explains why using a classical action to build amplitudes and using stationary phase to recover classical equations are related but opposite conceptual moves.

For common misuse of phrases such as "$\hbar\to0$" and “classical paths,” see Common Misstatements About the Classical Limit.

Semiclassical Approximations

The semiclassical idea is to treat the phase

$$\exp\left(\frac{i}{\hbar}S\right)$$

as rapidly varying when a characteristic action $S$ is large compared with $\hbar$. Contributions with rapidly changing relative phases cancel, while special regions where the phase is stationary survive.

This logic appears in several places:

• WKB wavefunctions, where the phase is an integral of the classical momentum;
• Bohr-Sommerfeld quantization, where a closed classical action is quantized;
• semiclassical propagators, where transition amplitudes are sums over classical paths;
• path integrals, where stationary action selects classical histories at leading order.

The result is often “classical plus corrections,” but the corrections are not optional decoration. They include tunneling exponents, Maslov phases, fluctuation determinants, and interference between multiple classical branches.

Action Large Compared with Hbar

The meaningful small parameter is not $\hbar$ by itself. It is a dimensionless ratio such as

$$\epsilon \sim \frac{\hbar}{S}, \qquad \epsilon\ll1.$$

For a one-dimensional particle with local classical momentum

$$p(x)=\sqrt{2m(E-V(x))},$$

the local de Broglie wavelength is

$$\lambda_{\mathrm{dB}}(x) = \frac{2\pi\hbar}{p(x)}.$$

Semiclassical wave mechanics is natural when the potential changes little over one local wavelength. Equivalently, the action accumulated over the scale of variation is large compared with $\hbar$.

This is why high-energy scattering, high-lying bound states, slowly varying potentials, and large occupation-number systems often admit semiclassical descriptions. It is also why the approximation can fail near turning points, caustics, singularities, abrupt boundaries, or long-time chaotic spreading.

WKB Preview

For a stationary one-dimensional Schrödinger equation,

$$\left[ - \frac{\hbar^2}{2m}\frac{d^2}{dx^2} + V(x) \right]\psi(x) = E\psi(x),$$

WKB uses a local phase form. In a classically allowed region, where $E>V(x)$,

$$\psi(x) \approx \frac{C_+}{\sqrt{p(x)}} \exp\left( \frac{i}{\hbar}\int^x p(x')\,dx' \right) + \frac{C_-}{\sqrt{p(x)}} \exp\left( - \frac{i}{\hbar}\int^x p(x')\,dx' \right).$$

The phase is the classical action integral. The amplitude factor $1/\sqrt{p(x)}$ expresses flux conservation: a classical particle spends more time where it moves slowly.

In a forbidden region, the local momentum becomes imaginary and WKB gives exponentially growing or decaying branches. That is the semiclassical origin of tunneling exponents.

A common validity condition in allowed regions is

$$\left\lvert \hbar\frac{p'}{p^2} \right\rvert \ll1.$$

This condition fails at a smooth turning point, where $p(x)=0$. The failure is repaired by connection formulas rather than by pretending the leading expression works everywhere.

The detailed method lives at WKB Approximation.

Stationary Phase Preview

The finite-dimensional model for many semiclassical calculations is an oscillatory integral

$$I(\hbar) = \int dq\,a(q) \exp\left( \frac{i}{\hbar}S(q) \right).$$

When $S/\hbar$ is large, the leading contributions come from points $q_\star$ satisfying

$$\left.\frac{dS}{dq}\right\rvert_{q_\star}=0.$$

Near such a point,

$$S(q) \approx S(q_\star) + \frac12S''(q_\star)(q-q_\star)^2 +\cdots .$$

The zeroth-order term gives the leading phase, while the quadratic term gives the Gaussian fluctuation prefactor and its phase. If $S''(q_\star)=0$, the ordinary stationary-phase approximation fails and a uniform approximation is needed.

This finite-dimensional picture is the seed of WKB matching, semiclassical propagators, saddle-point expansions, and many path-integral approximations.

Classical Paths in Path Integrals Preview

The real-time path integral for a propagator has the schematic form

$$K(q_b,t_b;q_a,t_a) = \int_{q(t_a)=q_a}^{q(t_b)=q_b} \mathcal Dq\, \exp\left( \frac{i}{\hbar}S[q] \right).$$

The stationary points of the action functional satisfy

$$\delta S[q_\gamma]=0,$$

which are the Euler–Lagrange equations with the endpoints fixed. Thus the leading semiclassical propagator is a sum over classical paths $\gamma$ connecting the endpoints:

$$K_{\mathrm{sc}} \sim \sum_\gamma A_\gamma \exp\left( \frac{i}{\hbar}S_\gamma \right).$$

The amplitude $A_\gamma$ is not arbitrary. It contains the fluctuation determinant, stability information, normalization, and caustic phases. In the detailed one-particle theory this is the Van Vleck-Maslov prefactor.

The important lesson is that classical paths enter as stationary contributions to quantum amplitudes. They are not classical probabilities, and multiple classical paths interfere.

Where Detailed Treatment Lives

Core Formalism owns the orientation: what the semiclassical limit is trying to do and why the ratio $S/\hbar$ matters.

Use the specialized pages for calculations:

• WKB and Semiclassical Methods for the applied chapter roadmap linking local wave mechanics, matching, quantization, tunneling, and propagators;
• Semiclassical Limit for dimensionless action-over-$\hbar$ scaling, stationary phase, Hamilton–Jacobi phases, and prefactors;
• WKB Approximation for one-dimensional wavefunctions in slowly varying potentials;
• Bohr-Sommerfeld Quantization for bound-state action quantization;
• Semiclassical Propagator for stationary-phase kernels and Van Vleck factors;
• Action Principles for fixed-endpoint variation, boundary terms, and the action functional behind path-integral saddles;
• Asymptotic Analysis for the mathematical language of big-O notation, stationary phase, and saddle points;
• Why Path Integrals? for the action-phase viewpoint;
• From Propagators to Path Integrals for the construction of the path-integral expression.

Common Mistakes

• Treating semiclassical approximations as classical probability theory.
• Saying WKB works whenever energy is high, without checking the local wavelength condition.
• Applying leading WKB directly at turning points.
• Ignoring prefactors and phases when comparing amplitudes.
• Forgetting that multiple classical paths are summed coherently.
• Treating path integrals as ordinary probability measures with positive weights.

References

• L. D. Landau and E. M. Lifshitz, Quantum Mechanics: Non-Relativistic Theory, 3rd ed., Pergamon, 1977.
• M. V. Berry and K. E. Mount, “Semiclassical approximations in wave mechanics,” Reports on Progress in Physics 35, 315-397, 1972.
• M. C. Gutzwiller, Chaos in Classical and Quantum Mechanics, Springer, 1990.
• M. Brack and R. K. Bhaduri, Semiclassical Physics, Westview Press, 2003.
• R. P. Feynman and A. R. Hibbs, Quantum Mechanics and Path Integrals, McGraw-Hill, 1965.

Exercises

1. In a region where $p(x)$ varies over length scale $L$, estimate the WKB validity condition from $\lvert \hbar p'/p^2 \rvert\ll1$.

Solution

If $p'/p\sim 1/L$, then

$$\left\lvert \hbar\frac{p'}{p^2} \right\rvert \sim \frac{\hbar}{pL}.$$

Thus WKB requires $pL/\hbar\gg1$, equivalently that the local de Broglie wavelength be small compared with the scale over which the momentum changes.

2. Why does stationary phase select $S'(q_\star)=0$ rather than points where $S(q)$ is largest?

Solution

The integrand is oscillatory, not exponentially weighted by a real positive function. Away from stationary points, neighboring phases vary rapidly and cancel. Near $S'(q_\star)=0$, the phase is locally stable to first order, so nearby contributions can add coherently.

3. In a path integral, why does the condition $\delta S=0$ produce classical equations?

Solution

The action functional is the same object used in Hamilton’s principle. Requiring its first variation to vanish under fixed-endpoint variations gives the Euler-Lagrange equations. In the path integral, those same stationary histories dominate the leading large-action approximation.

4. Why is a sum over classical paths in the semiclassical propagator not a classical mixture?

Solution

Each path contributes a complex amplitude with a phase, not a positive probability. The contributions can interfere constructively or destructively. A classical mixture would add probabilities and would not contain relative phase information.