Common Misstatements About the Classical Limit

The classical limit is often summarized by slogans. Some slogans are useful for orientation, but many become false when treated as explanations.

The reliable version is more precise: quantum mechanics reproduces classical descriptions only for specified states, observables, scales, environments, time intervals, and experimental resolutions. The Classical Limit is a family of controlled approximations, not one magic operation.

This page collects common misstatements and replaces each with a statement that can be used safely.

Quick Diagnostic

When a claim about the classical limit appears, ask:

1. Which quantum model is being approximated?
2. Which observables are being compared with classical variables?
3. Which states or preparations are being used?
4. What dimensionless parameter is small or large?
5. What role is played by coarse graining, measurement resolution, or decoherence?
6. Over what time interval is the approximation valid?

If those questions have no answers, the claim is probably a slogan rather than a derivation.

“Quantum Mechanics Only Matters for Small Things”

This is false. Quantum mechanics matters whenever coherence, discreteness, noncommutativity, tunneling, entanglement, or spin affect the phenomena being modeled. Size alone is not the criterion.

The better criterion is scale relative to the relevant quantum quantities. A large object can behave classically because its action scale is enormous compared with $\hbar$, its de Broglie wavelength is unresolvably small, and environmental decoherence rapidly suppresses interference between macroscopically distinct alternatives. But those are physical mechanisms, not a rule that quantum mechanics switches off above a certain size.

Macroscopic systems can also display quantum behavior when coherence is protected or collective degrees of freedom are involved. Superconductors, superfluids, interferometers, and engineered mechanical resonators are reminders that “large” and “classical” are not synonyms.

“The Classical Limit Is Just hbar Equals Zero”

The phrase "$\hbar\to0$" is shorthand. It is not a literal instruction to erase $\hbar$ from every formula.

Because $\hbar$ has units of action, its numerical size depends on units. The meaningful comparison is dimensionless:

$$\frac{S}{\hbar}\gg1,$$

where $S$ is a characteristic action of the problem. In wave mechanics this often appears as a short de Broglie wavelength relative to the scale over which the potential changes:

$$\lambda_{\mathrm{dB}} = \frac{2\pi\hbar}{p}, \qquad \lambda_{\mathrm{dB}}\ll L.$$

Even then, one must specify the approximation. Stationary phase, WKB, large quantum number, decoherence, and coarse graining are different routes to classical-looking behavior.

See Semiclassical Limit Overview for the action-phase version of this statement.

“Decoherence Solves Every Interpretation Problem”

Decoherence is essential, but the overstatement is false.

The accurate statement is that decoherence suppresses locally observable interference between alternatives by entangling them with environmental degrees of freedom. In a simple two-branch model,

$$c_1|s_1\rangle|E_1\rangle + c_2|s_2\rangle|E_2\rangle$$

has a reduced system density matrix whose off-diagonal terms are proportional to $\langle E_2|E_1\rangle$. When this overlap is tiny, local interference between $|s_1\rangle$ and $|s_2\rangle$ becomes negligible.

That explains why stable records and classical probability descriptions become effective. It does not, by itself, say why a single individual outcome is realized, whether collapse is physical, or which interpretation is correct. Those questions require additional interpretive or dynamical input.

See Decoherence Preview for the density-matrix calculation and its boundary.

“Expectation Values Always Obey Classical Equations”

Ehrenfest theorem gives an exact and important bridge, but the common slogan is too strong.

For

$$H=\frac{p^2}{2m}+V(x),$$

Ehrenfest theorem gives

$$\frac{d}{dt}\langle x\rangle = \frac{\langle p\rangle}{m}, \qquad \frac{d}{dt}\langle p\rangle = -\langle V'(x)\rangle.$$

Newton’s equation for the mean position would require

$$\langle V'(x)\rangle \approx V'(\langle x\rangle).$$

That approximation is exact for potentials at most quadratic in $x$. In nonlinear potentials it usually requires a sufficiently narrow wave packet, and the packet must remain narrow over the time interval of interest.

Expectation values can also be misleading. A state split into two well-separated wave packets may have an average position in a region where the particle is unlikely to be detected. Classical motion is therefore not just motion of means; it also needs localization, small relative fluctuations, and an appropriate measurement context.

“Macroscopic Objects Cannot Be Quantum”

This is false in two different ways.

First, macroscopic objects are described by quantum mechanics at the microscopic level. Their apparent classicality is usually an emergent approximation caused by large action scales, many degrees of freedom, coarse graining, and decoherence.

Second, carefully prepared macroscopic or mesoscopic systems can display quantum coherence. The difficulty is practical and dynamical: maintaining isolation and phase coherence becomes harder as more environmental channels become available.

The safer statement is: ordinary macroscopic objects are usually well described by classical variables for many everyday observables because interference between macroscopically distinct alternatives is extremely hard to observe and because the relevant quantum corrections are far below the measurement resolution.

“Large Quantum Number Means a Particle Has a Trajectory”

Large quantum number often helps with correspondence, but it does not automatically produce a classical trajectory.

In an infinite square well, high-$n$ energy levels have small relative spacing:

$$\frac{E_{n+1}-E_n}{E_n} \to0.$$

The high-$n$ probability density also becomes rapidly oscillatory, and a coarse-grained average approaches the classical uniform time-spent distribution. But an energy eigenstate in the box is still a standing wave, not a small object moving left and right with a definite trajectory.

Trajectories usually require wave packets or semiclassical states, not merely high energy eigenstates. Even then, the packet can spread, split, or interfere.

See Infinite Square Well for the model and Gaussian Wave Packets for a clean trajectory-like packet example.

“A Diagonal Density Matrix Is Just Ordinary Ignorance”

A diagonal density matrix can represent ordinary classical uncertainty about a preparation, but it need not.

If an ensemble prepares $|0\rangle$ with probability $1/2$ and $|1\rangle$ with probability $1/2$, then

$$\rho = \frac12|0\rangle\langle0| + \frac12|1\rangle\langle1|$$

is a proper mixture reflecting preparation ignorance.

The same reduced density matrix can arise from an entangled pure state after tracing out another system. In that case the mixed state is an improper mixture: it gives all local predictions for the subsystem, but it does not mean the subsystem was secretly prepared in one of the basis states independently of the larger system.

This distinction is central in decoherence and subsystem physics. See Reduced Density Matrices and Classical Mixtures vs Quantum Superpositions.

“The Correspondence Principle Proves Hidden Classical Paths”

The Correspondence Principle says that quantum mechanics must reproduce classical predictions in regimes where classical physics works. It does not say that every quantum state has an underlying classical path.

In semiclassical path integrals, classical paths appear as stationary-phase contributions to amplitudes. Multiple classical paths can contribute coherently and interfere. This is not the same as a classical probability distribution over hidden trajectories.

In WKB theory, the phase is related to a classical action, but the wavefunction remains a quantum amplitude with boundary conditions, phases, turning-point connection rules, and tunneling behavior.

The correction is simple: correspondence is a consistency and approximation principle, not an ontology of particles following definite classical paths.

“Quantization and the Classical Limit Are Inverses”

They point in opposite directions, but they are not inverse operations.

Quantization starts from classical data and constructs a quantum model. The classical limit starts from a quantum model and extracts a classical approximation. A classical Hamiltonian may admit more than one quantum realization because of ordering, topology, Hilbert-space, domain, and symmetry choices. A quantum model may also have several useful classical regimes.

For example, the quantum harmonic oscillator has high-$n$ spectral correspondence, coherent-state trajectory correspondence, and high-temperature thermal correspondence. These are different classical-looking statements extracted from one quantum model.

See Quantization vs Classical Limit for the detailed separation of the two arrows.

“Once Something Is Classical, Quantum Effects Are Gone”

Classical descriptions are effective descriptions. They keep some variables and ignore others. Quantum effects can be negligible for one observable and visible for another.

A beam may have a sharply defined classical path while retaining spin coherence. A macroscopic detector may have classical pointer records while its microscopic constituents remain quantum. A high-energy scattering problem may allow a semiclassical trajectory approximation while still showing phase shifts or tunneling corrections.

The right question is not “Is the system classical?” The right question is:

$$\text{Which description is accurate for which observables and resolutions?}$$

Classical behavior is therefore contextual and approximate, even when the approximation is extraordinarily good.

Common Mistakes

• Using size as the only criterion for classicality.
• Treating $\hbar\to0$ as a formal deletion rule rather than a dimensionless asymptotic limit.
• Replacing $\langle V'(x)\rangle$ by $V'(\langle x\rangle)$ without checking the wave-packet width.
• Treating decoherence as collapse or as a complete interpretation of measurement.
• Treating every diagonal reduced density matrix as ordinary ignorance.
• Assuming high quantum number creates a trajectory without constructing a localized state.
• Treating correspondence as hidden-variable ontology.
• Assuming quantization is uniquely determined by a classical Hamiltonian.

Cross-Links

• Correspondence Principle
• Classical Limit
• Ehrenfest Theorem Overview
• Semiclassical Limit Overview
• Semiclassical Limits and Correspondence
• Decoherence Preview
• Quantization vs Classical Limit
• Reduced Density Matrices
• Classical Mixtures vs Quantum Superpositions
• Gaussian Wave Packets
• Infinite Square Well

References

• N. Bohr, “On the Quantum Theory of Line-Spectra,” Det Kongelige Danske Videnskabernes Selskabs Skrifter 8, 4, 1-118, 1918.
• P. Ehrenfest, “Bemerkung über die angenäherte Gültigkeit der klassischen Mechanik innerhalb der Quantenmechanik,” Zeitschrift für Physik 45, 455-457, 1927.
• L. E. Ballentine, Quantum Mechanics: A Modern Development, 2nd ed., World Scientific, 2014.
• M. V. Berry and K. E. Mount, “Semiclassical approximations in wave mechanics,” Reports on Progress in Physics 35, 315-397, 1972.
• W. H. Zurek, “Decoherence, einselection, and the quantum origins of the classical,” Reviews of Modern Physics 75, 715-775, 2003.
• M. Schlosshauer, Decoherence and the Quantum-to-Classical Transition, 2nd ed., Springer, 2019.

Exercises

1. Explain why ”$\hbar$ is small” is not a complete classical-limit argument.

Solution

$\hbar$ has units of action, so its numerical value has no invariant meaning by itself. One needs a dimensionless comparison such as $S/\hbar\gg1$, where $S$ is a characteristic action. One must also say which states, observables, and resolutions are being approximated.

2. A wave packet in an anharmonic potential has $\Delta x$ comparable to the length scale over which $V'(x)$ changes. Why is Ehrenfest theorem not enough to justify a Newtonian trajectory for $\langle x\rangle$?

Solution

Ehrenfest theorem gives

$$m\frac{d^2}{dt^2}\langle x\rangle = -\langle V'(x)\rangle.$$

Newtonian motion for the mean would require $\langle V'(x)\rangle\approx V'(\langle x\rangle)$. If the packet is broad on the force-variation scale, higher moments of the state contribute significantly, so that approximation is not controlled.

3. Why does decoherence not imply that the global state has collapsed?

Solution

In the standard unitary model, decoherence entangles alternatives with environmental states. The reduced density matrix of the system loses off-diagonal terms when the environment is traced out, but the full system-environment state may remain a pure superposition. Decoherence suppresses local interference; it is not automatically a global nonunitary collapse.

4. Give an example where a classical-looking description is valid for one aspect of a system but not another.

Solution

A particle beam may have a narrow spatial wave packet whose center follows an approximately classical path, while its spin remains in a coherent superposition and must be treated quantum mechanically. The spatial motion can be approximated classically for the chosen resolution, but the spin degree of freedom cannot be replaced by a classical probability distribution without losing interference information.