Classical Limit and the Correspondence Principle

The chapter has spent eleven modules building classical field theory toward quantum field theory; this article checks the bridge from the far side. If quantization is legitimate, the quantum theory must hand classical field theory back in the appropriate limit — Bohr's correspondence principle, upgraded from atomic orbits to fields. The article's three derivations are the three faces of that guarantee. In the path-integral face, quantum mechanics weights every field history by $e^{iS[\phi]/\hbar}$ ; when the action dwarfs $\hbar$ , neighbouring histories cancel in a fury of spinning phases except near the stationary point $\delta S = 0$ — the classical solution survives as the lone constructive interference, and Hamilton's principle, the variational bedrock of this entire course, is revealed as a stationary-phase approximation. In the operator face, Ehrenfest's theorem makes expectation values obey Hamilton's field equations, exactly for free fields and up to fluctuation corrections otherwise. In the wavefunctional face, the WKB phase obeys the field-theoretic Hamilton-Jacobi equation — closing the loop the Hamilton-Jacobi chapter opened when it derived Schrödinger's equation as waves whose rays are trajectories. Running through all three is Dirac's dictionary read in reverse: $\frac{1}{i\hbar}[\hat A, \hat B] \to \{A, B\}$ — the commutator's rate of failure is the Poisson bracket, so the entire canonical skeleton of this chapter (brackets, charges, generators, the Poincaré algebra) is what survives when $\hbar$ becomes negligible. The operational meaning of "negligible" — large occupation numbers, coherent states, actions of $10^{31}\hbar$ — is quantified honestly, along with where correspondence fails: nonlinear fluctuations, and the quantum states no classical field can imitate. The capstone article walks the bridge forward; this one certifies it holds weight in both directions.

Key takeaways

Correspondence principle for fields: quantum dynamics must reduce to the classical Euler-Lagrange equations in the appropriate limit — and it does, three ways.
Path integral: histories weighted by $e^{iS/\hbar}$ interfere destructively except near $\delta S = 0$ ; the classical solution is the stationary-phase survivor, and Hamilton's principle is its shadow.
Ehrenfest: $\partial_t^2\langle\hat\phi\rangle = \nabla^2\langle\hat\phi\rangle - \langle V'(\hat\phi)\rangle$ — exact; it closes into the classical equation when $\langle V'\rangle \approx V'(\langle\phi\rangle)$ : exact for free fields, fluctuation-corrected otherwise.
WKB wavefunctional $\Psi = A\,e^{iS/\hbar}$ : order $\hbar^0$ is the field Hamilton-Jacobi equation; order $\hbar^1$ a continuity equation along classical flow.
Dirac's dictionary in reverse: $\frac{1}{i\hbar}[\hat A, \hat B] \to \{A, B\}$ — the Poisson-bracket algebra, charges and generators included, is the classical skeleton of the quantum theory.
" $\hbar \to 0$ " means dimensionless ratios: $S/\hbar \gg 1$ , mode occupation $N \gg 1$ ; coherent states with $\Delta\phi/\langle\phi\rangle \sim 1/\sqrt{N}$ are the operational classical states.
Correspondence fails informatively: nonlinear fluctuation corrections (loop expansion), chaotic systems (log- $\hbar$ Ehrenfest time), and intrinsically quantum states (entangled, sub-Poissonian) with no classical counterpart.
Key idea: classical field theory is the stationary-phase, large-occupation skeleton of quantum field theory — everything this chapter built is what survives the limit.

The core insight

Replace the quantum amplitude $e^{iS[\phi]/\hbar}$ by its stationary-phase value as $\hbar \to 0$ : only configurations with $\delta S = 0$ — the classical solutions — survive, while everything else interferes destructively. Equivalently, $\frac{1}{i\hbar}[\hat A, \hat B] \to \{A, B\}_{\mathrm{PB}}$ , so quantum Heisenberg evolution becomes classical Hamiltonian flow.

Where this lives today

The classical limit stopped being philosophy and became engineering specification. Gravitational-wave astronomy lives at the boundary: LIGO's mirrors are 40 kg classical objects whose positions are read near the standard quantum limit, and the injected squeezed light that now improves every observing run is a deliberate reshaping of the $\sqrt{\hbar}$ coherent-state blob this article used to define classicality. Quantum optics manufactures the correspondence dial at will — from single-photon states (maximally non-classical) through lasers ( $N \sim 10^{10}$ , classical to $10^{-5}$ ) — and cavity optomechanics has placed micromechanical drums and levitated nanospheres into superpositions, pushing the measured classical-quantum frontier to objects of $10^{9}$ atoms. Decoherence theory, the modern completion of the story (see FAQ), is quantitatively confirmed in these platforms: environmental monitoring rates, not any new law, decide where the stationary-phase world begins. On the theory side, the semiclassical expansion is a precision industry: instantons — complex saddles of $e^{iS/\hbar}$ — compute tunnelling and vacuum decay (the metastability analysis of the Higgs vacuum is a one-loop-around-a-saddle calculation of exactly this article's Step 4 type); the classical-field approximation of heavy-ion physics (color-glass condensate) and of early-universe preheating exploits high occupation $N \gg 1$ where tree-level field theory is quantitatively exact; and inflationary cosmology runs the correspondence in reverse — quantum fluctuations of order $\sqrt{\hbar}$ , stretched to cosmic size and decohered, become the classical density perturbations in the CMB: the largest classical structures in existence carry quantum birthmarks, priced by this article's ratios. Within the chapter: the bridge is certified from both ends, and one article remains — the capstone, which walks it forward: from the action principle, through the brackets and charges this chapter built, into quantum field theory.

The physical picture

Imagine summing arrows — one unit phasor $e^{iS[\phi]/\hbar}$ for every conceivable field history, solutions and absurdities alike. The action assigns each history a phase, and $\hbar$ sets the exchange rate between action and angle. When $S \gg \hbar$ , two histories differing by even a whisper of action have phases differing by many full turns: their arrows point at random, and summing them is a drunken walk going nowhere. The exception is a history where the phase is stationary — where first-order changes of the history change $S$ not at all. Near such a history, neighbours share their phase, arrows align, and the sum is dominated by one coherent bundle. But " $\delta S = 0$ " is exactly Hamilton's principle: the classical solution is where quantum interference fails to destroy. Smaller $\hbar$ narrows the constructive window; the classical world is the limit in which only the stationary history's bundle survives.

Figure

Why the classical path exists — phasor sums over histories far from and near the stationary point of the action

Two companion pictures complete the intuition. The coherent state: a quantum field mode in such a state is a Gaussian blob in the $(\phi_k, \pi_k)$ phase plane of the Hamiltonian article, riding the classical ellipse with fixed shape. Its spread is order $\sqrt{\hbar}$ ; the orbit's amplitude, for a mode holding $N$ quanta, is order $\sqrt{N\hbar}$ . Relative uncertainty $\sim 1/\sqrt{N}$ : a bright laser mode with $N \sim 10^{10}$ is a classical field to ten parts per million, and nothing but photon-counting statistics betrays the graininess. "Classical" is not a different theory; it is quantum mechanics at high occupation, too smooth to notice its own pixels. And the dictionary: the canonical commutator $[\hat\phi, \hat\pi] = i\hbar\,\delta^3$ is $\hbar$ times the fundamental bracket of the Poisson-brackets article. As $\hbar \to 0$ the operators commute — they become ordinary functions — but the rate at which commutativity fails, commutator over $i\hbar$ , tends to the Poisson bracket. The classical world is not the zeroth-order world where everything is plain numbers; it is the leading non-commuting skeleton — brackets, flows, charges, the whole canonical apparatus — with the higher-order quantum flesh stripped off. That is why every structure of this chapter reappears intact inside QFT: it was the load-bearing frame all along.

Figure

The coherent state — a blob of width square-root-of-ħ riding the classical ellipse of amplitude square-root-of-N

Three faces of the limit

Derivation AThe classical field equation as the stationary phase of the path integral

We derive the chapter's founding axiom — Hamilton's principle for fields — as the $\hbar \to 0$ limit of the quantum sum over histories. This is the field-theoretic generalization of the stationary-phase logic the WKB article of the Hamilton-Jacobi chapter used for particles, with the functional calculus of this chapter supplying the machinery. Strategy: expand the action functionally about a trial history, demand the expansion have no linear term (that defines the saddle), and watch the fluctuation integral decouple at leading order in $\hbar$ .

Step 1. The quantum amplitude sums all field histories with the action as phase:

Z = \int \mathcal{D}\phi\;\, e^{\tfrac{i}{\hbar}S[\phi]}

where $\mathcal{D}\phi$ ranges over every configuration history — no equations of motion imposed. We seek the $\hbar \to 0$ behaviour.

Step 2. Write $\phi = \phi_{\text{cl}} + \eta$ and Taylor-expand the action functionally (the functional-derivatives article's toolkit):

S[\phi] = S[\phi_{\text{cl}}] + \int \frac{\delta S}{\delta\phi}\bigg|_{\phi_{\text{cl}}}\eta + \frac{1}{2}\iint \eta\,\frac{\delta^2 S}{\delta\phi\,\delta\phi}\bigg|_{\phi_{\text{cl}}}\eta + \cdots

Step 3. Choose the expansion point to kill the linear term:

\frac{\delta S}{\delta\phi(x)}\bigg|_{\phi_{\text{cl}}} = \frac{\partial\mathcal{L}}{\partial\phi} - \partial_\mu\frac{\partial\mathcal{L}}{\partial(\partial_\mu\phi)} = 0

— the Euler-Lagrange equation. Why this is forced rather than chosen: with a linear term present, the phase varies at first order in $\eta$ and the oscillations $e^{i(\text{linear})/\hbar}$ average the integral to zero as $\hbar \to 0$ ; only expansion points with no linear term — genuine saddles — contribute at leading order. The classical solutions are not selected by decree but by the destructive interference of everything else.

Step 4. At the saddle, the leading factorization is

Z \simeq e^{\tfrac{i}{\hbar}S[\phi_{\text{cl}}]} \int \mathcal{D}\eta\;\exp\!\left(\frac{i}{2\hbar}\iint \eta\,S''\,\eta\right)

The Gaussian fluctuation integral contributes $[\det S'']^{-1/2}$ — the one-loop correction, suppressed relative to the classical phase — so $\lim_{\hbar\to 0} \frac{\hbar}{i}\ln Z = S[\phi_{\text{cl}}]$ : the classical action governs the leading behaviour, and classical dynamics is the tree of a systematic expansion in $\hbar$ . Hamilton's principle — postulated in the action-functional article as mechanics' deepest axiom — is derived: it is where quantum interference is constructive. Verified: standard saddle-point/stationary-phase reduction (any path-integral text); consistent with Hamilton's principle per Goldstein ch. 2; the particle-mechanics shadow is the WKB article's eikonal argument, of which this is the functional transcription.

Go deeper · Derivation B · 1 minEhrenfest's theorem for fields — and exactly when it closes

Derivation BEhrenfest's theorem for fields — and exactly when it closes

We derive the equation of motion for the expectation value of the quantum field and identify precisely when it coincides with the classical field equation. The result is the operator-language face of correspondence and the honest statement of its limits: exact closure for free (quadratic) theories, fluctuation-corrected closure otherwise. Strategy: Heisenberg evolution with the canonical commutator, twice; then one inspection of where averaging and differentiating fail to commute.

Step 1. The quantum Hamiltonian is the classical functional with hats, $\hat H = \int\mathrm{d}^3x\,[\tfrac12\hat\pi^2 + \tfrac12(\nabla\hat\phi)^2 + V(\hat\phi)]$ , and Heisenberg evolution is $\mathrm{d}\hat A/\mathrm{d}t = \frac{1}{i\hbar}[\hat A, \hat H]$ — Dirac's image of $\dot A = \{A, H\}$ .

Step 2. Evolve the field: using $[\hat\phi(\mathbf{x}), \hat\pi(\mathbf{y})] = i\hbar\,\delta^3(\mathbf{x} - \mathbf{y})$ ,

\dot{\hat\phi} = \frac{1}{i\hbar}[\hat\phi, \hat H] = \hat\pi

Note the $\hbar$ 's fate: one factor from the commutator cancels the $1/i\hbar$ of the evolution law, leaving an $\hbar$ -free equation — the structural reason quantum expectation values can obey classical-looking dynamics at all. This is the reverse of the bracket article's quantization map, executed.

Step 3. Evolve the momentum the same way: $\dot{\hat\pi} = \nabla^2\hat\phi - V'(\hat\phi)$ . Combine and average over any state:

\partial_t^2\,\langle\hat\phi\rangle = \nabla^2\langle\hat\phi\rangle - \big\langle V'(\hat\phi)\big\rangle

— an exact operator identity, no approximation yet.

Step 4. Inspect the closure. The classical equation needs $\langle V'(\hat\phi)\rangle = V'(\langle\hat\phi\rangle)$ — the average of the force equal to the force of the average. For $V = \tfrac12 m^2\phi^2$ , $V'$ is linear and the identity is exact: the mean of a free quantum field obeys the Klein-Gordon equation with no corrections whatsoever, in any state. For nonlinear $V$ , expand about the mean with $\hat\phi = \bar\phi + \delta\hat\phi$ :

\langle V'(\hat\phi)\rangle = V'(\bar\phi) + \tfrac12 V'''(\bar\phi)\,\langle\delta\hat\phi^2\rangle + \cdots

The classical equation is the tree-level truncation; fluctuation corrections enter through the variance, order $\hbar$ (a coherent state's $\langle\delta\phi^2\rangle \sim \hbar/2\omega$ ), and grow with the potential's curvature — quantified in the worked example. Mean-field theory, the effective action, and the whole loop expansion of QFT live in the correction terms. Verified: commutators recomputed against the bracket article's classical counterparts; free-field exactness checked (linear $V'$ commutes with averaging); the fluctuation expansion is the standard mean-field/one-loop structure.

Go deeper · Derivation C · 1 minThe WKB wavefunctional and the field Hamilton-Jacobi equation

Derivation CThe WKB wavefunctional and the field Hamilton-Jacobi equation

We take the Schrödinger-picture quantum field theory — a wavefunctional $\Psi[\phi]$ over field configurations — insert the WKB ansatz, and recover at leading order the Hamilton-Jacobi equation for fields, with the next order transporting probability along classical flow lines. This closes the grand circle of the course: the Hamilton-Jacobi chapter derived Schrödinger's equation by making mechanics wave-like; here the same asymptotics runs in reverse at the field level. Strategy: functional Schrödinger equation, ansatz $\Psi = A\,e^{iS/\hbar}$ , collect powers of $\hbar$ .

Step 1. In the functional Schrödinger picture, states are functionals $\Psi[\phi, t]$ and the momentum acts as $\hat\pi(\mathbf{x}) = -i\hbar\,\delta/\delta\phi(\mathbf{x})$ — the field version of $\hat p = -i\hbar\,\partial_x$ , with the functional derivative of this chapter in place of the ordinary one:

i\hbar\,\partial_t\Psi = \int\mathrm{d}^3x\left[ -\frac{\hbar^2}{2}\frac{\delta^2}{\delta\phi(\mathbf{x})^2} + \frac{1}{2}(\nabla\phi)^2 + V(\phi) \right]\Psi

Step 2. Insert $\Psi = A[\phi, t]\,e^{iS[\phi, t]/\hbar}$ with $A, S$ real functionals. The functional Laplacian produces three kinds of terms — $(\delta S/\delta\phi)^2$ at order $\hbar^0$ , mixed $A$ - $S$ terms at order $\hbar^1$ , and $\delta^2 A/\delta\phi^2$ at order $\hbar^2$ .

Step 3. Order $\hbar^0$ (the real part):

-\,\frac{\partial S}{\partial t} = \int\mathrm{d}^3x\left[ \frac{1}{2}\left(\frac{\delta S}{\delta\phi}\right)^2 + \frac{1}{2}(\nabla\phi)^2 + V(\phi) \right] = H\!\left[\phi,\, \frac{\delta S}{\delta\phi}\right]

— the Hamilton-Jacobi equation for fields: the phase $S$ is a classical action functional, its functional gradient the momentum density $\pi = \delta S/\delta\phi$ , exactly as the Hamilton-Jacobi chapter's $p = \partial S/\partial q$ promised, with the dictionary $\partial \to \delta$ applied.

Step 4. Order $\hbar^1$ (the imaginary part):

\partial_t |A|^2 + \int\mathrm{d}^3x\,\frac{\delta}{\delta\phi}\!\left( |A|^2\,\frac{\delta S}{\delta\phi} \right) = 0

— a continuity equation on configuration space: probability density $|A|^2$ is transported along the classical Hamiltonian flow whose velocity field is $\delta S/\delta\phi$ . Constant- $S$ surfaces propagate as wavefronts whose rays are classical field histories — the optical-mechanical analogy of the Hamilton-Jacobi chapter, promoted to infinite dimensions. Quantum mechanics at this order is classical mechanics plus a conserved probability riding it; genuine quantum novelty (tunnelling between saddles, interference of separated bundles) begins at the orders beyond. Verified: orders collected as in the standard WKB reduction (Landau-Lifshitz QM §6 for the particle case); the field HJ equation matches the canonical $H[\phi, \pi]$ of the Hamiltonian article with $\pi = \delta S/\delta\phi$ ; consistent with the HJ-to-Schrödinger article run backwards.

Limiting cases

Case	Result	Meaning
free field ( $V$ quadratic)	Ehrenfest exact: $(\Box + m^2)\langle\hat\phi\rangle = 0$	means of free quantum fields are classical solutions, period
coherent state, $N$ quanta	$\Delta\phi/\langle\phi\rangle \sim 1/\sqrt{N}$	laser limit: classical to fluctuations of $1/\sqrt{N}$
$S/\hbar \to \infty$	stationary phase exact; one-loop $\det S''$ correction	macroscopic mechanics: $10^{31}\hbar$ for a pendulum
single mode, $\hbar\to 0$ at fixed energy	$[\,,]/i\hbar \to \{,\}$ on that oscillator	the bracket article's quantization map, inverted
nonlinear $V$ , small fluctuations	classical equation + $\tfrac12 V'''\langle\delta\phi^2\rangle$ correction	tree level + loops; mean-field theory
chaotic classical dynamics	correspondence holds only to $t_E \sim \ln(1/\hbar)/\lambda$	Ehrenfest time; see FAQ
squeezed/entangled states	no classical field imitates the statistics	correspondence bounds, not bans, quantum novelty
$\hbar \to 0$ in the Noether sector	charges, algebra, generators survive as brackets	this chapter is the surviving skeleton

Common trapThere is no dimensionful limit ħ → 0 — and the mean of a quantum field is not automatically a classical field

Two precision errors, both live in research settings. First: $\hbar$ is a dimensionful constant of nature ( $1.055\times10^{-34}$ J·s); "sending it to zero" is meaningless except through dimensionless ratios. The honest statements are $S/\hbar \gg 1$ (actions large in quantum units), $N \gg 1$ (occupations large), $k_BT \gg \hbar\omega$ (thermal occupation large — the Hamiltonian article's criterion). Systems with small action are not "slightly quantum"; they are fully quantum, and no formal limit rescues classicality for them. The same system can be classical in one variable and quantum in another (a laser's field amplitude versus its photon-number parity), because different observables involve different ratios. Second: Ehrenfest's theorem does not say "the expectation value is a classical field." It says the mean obeys the classical equation if $\langle V'(\hat\phi)\rangle = V'(\langle\hat\phi\rangle)$ — exact only for linear $V'$ ; otherwise the mean's motion is dragged by the fluctuations ( $\tfrac12 V'''\langle\delta\phi^2\rangle$ and up), and in strongly fluctuating states the mean can behave like no classical solution at all (a symmetric superposition of two lumps has $\langle\hat\phi\rangle$ halfway between — a configuration the classical dynamics might never visit). The clean hierarchy to keep: operator identity (always) → mean-field equation (small fluctuations) → classical field (small fluctuations and a state that stays localized in configuration space). The third condition is decoherence's job — see the FAQ.

Worked examples

1 · How classical is classical? Two audits (confidence builder)

(i) A pendulum: $m = 0.100$ kg, $\ell = 1.00$ m, amplitude $0.100$ rad, period $T \approx 2.0$ s. Estimate its action per cycle in units of $\hbar$ . (ii) A 1.0 mW He-Ne laser ( $\lambda = 633$ nm): how many photons occupy one coherence time ( $\tau = 1.0\ \mu$ s), and what is the relative field uncertainty of the coherent state?

Step 1. Pendulum energy: $E \approx \tfrac12 mg\ell\,\theta_0^2 = \tfrac12(0.100)(9.81)(1.00)(0.0100) = 4.9\times10^{-3}$ J. Action per cycle $\sim E\,T \approx 9.8\times10^{-3}$ J·s.

Step 2. In quantum units: $S/\hbar \approx 9.8\times10^{-3}/1.055\times10^{-34} = \mathbf{9\times10^{31}}$ . The stationary-phase window is fantastically narrow: histories deviating by even one part in $10^{31}$ of the action decohere from the bundle. This number — thirty-two orders of magnitude — is why classical mechanics is exact for pendulums and why no laboratory pendulum will ever tunnel.

Step 3. Laser photons: each carries $E_\gamma = hc/\lambda = (6.63\times10^{-34})(3.0\times10^8)/(633\times10^{-9}) = 3.1\times10^{-19}$ J. Flux: $10^{-3}/3.1\times10^{-19} = 3.2\times10^{15}$ photons/s. In one coherence time: $N = 3.2\times10^{15}\times10^{-6} = \mathbf{3.2\times10^{9}}$ .

Step 4. Coherent-state fluctuation: $\Delta\phi/\langle\phi\rangle \sim 1/\sqrt{N} = 1/\sqrt{3.2\times10^9} = \mathbf{1.8\times10^{-5}}$ . The beam is a classical electromagnetic wave to eighteen parts per million — and photon-counting experiments see exactly the residual $\sqrt{N}$ shot noise this estimate predicts, the graininess a classical field cannot have. Classicality is not a yes/no property; it is a measured number, and here it is $10^{-5}$ .

2 · When Ehrenfest fails: pricing the fluctuation correction (exam level)

A scalar field with quartic interaction $V = \tfrac12 m^2\phi^2 + \tfrac{\lambda}{4!}\phi^4$ is in a state with mean field $\bar\phi = 0.50$ and Gaussian fluctuations of variance $\langle\delta\phi^2\rangle = 0.010$ (both in natural units, $m = 1$ , $\lambda = 6.0$ ). (i) Compute the exact $\langle V'\rangle$ and the classical (tree) value $V'(\bar\phi)$ . (ii) Quantify the correction, and identify which classical parameter it mimics.

Step 1. For a Gaussian state, odd moments of $\delta\hat\phi$ vanish and $\langle\hat\phi^3\rangle = \bar\phi^3 + 3\bar\phi\,\langle\delta\phi^2\rangle$ . So with $V' = m^2\phi + \tfrac{\lambda}{6}\phi^3$ :

\langle V'\rangle = m^2\bar\phi + \frac{\lambda}{6}\left(\bar\phi^3 + 3\bar\phi\,\langle\delta\phi^2\rangle\right)

Step 2. Numbers. Tree: $V'(\bar\phi) = (1)(0.50) + (1.0)(0.125) = \mathbf{0.625}$ (using $\lambda/6 = 1.0$ ). Exact: add $\tfrac{\lambda}{6}\cdot 3\bar\phi\langle\delta\phi^2\rangle = (1.0)(3)(0.50)(0.010) = 0.015$ : $\langle V'\rangle = \mathbf{0.640}$ .

Step 3. The correction is $0.015/0.625 = \mathbf{2.4\%}$ — the mean field feels a force $2.4\%$ stiffer than the classical equation predicts. Note its structure: $\tfrac{\lambda}{2}\langle\delta\phi^2\rangle\,\bar\phi = (3.0)(0.010)\,\bar\phi = 0.030\,\bar\phi$ — linear in $\bar\phi$ , i.e. the fluctuations act as a mass-squared shift $\delta m^2 = \tfrac{\lambda}{2}\langle\delta\phi^2\rangle = 0.030$ . The quantum correction masquerades as a change of a classical parameter.

Step 4. Read the moral at full depth: this is the seed of renormalization. The variance $\langle\delta\phi^2\rangle$ of a quantum field diverges with the mode cutoff (the Hamiltonian article's zero-point ledger), so the mass shift is cutoff-dependent — and the resolution, matching the measured mass to $m^2 + \tfrac{\lambda}{2}\langle\delta\phi^2\rangle$ rather than the bare parameter, is the first renormalization every QFT student performs. Correspondence fails by exactly computable amounts, and the failures are not noise but the loop expansion: tree = classical field theory (this chapter), one loop = the Gaussian fluctuations just priced, and upward into QFT proper.

Origins · 1 minNiels Bohr and how this was discovered

Niels Bohr

1885–1962 · Copenhagen

The correspondence principle is Bohr's signature move, and it built quantum mechanics before quantum mechanics existed. From his 1913 atom onward, Bohr's method was to demand that quantum rules merge into classical physics in the limit of large quantum numbers — high orbits radiating at their classical revolution frequencies — and through the 1910s he and his Copenhagen circle wielded the principle as a crowbar: prying selection rules, intensities and polarizations out of classical radiation theory when no quantum dynamics yet existed to compute them. Heisenberg's 1925 matrix mechanics was born directly of this program — an attempt to keep only relations between observables that survive the correspondence limit — and Dirac's bracket-commutator dictionary, the spine of this chapter's quantization story, is the principle made algebra. Bohr's institute on Blegdamsvej became quantum theory's capital, and his later years produced the complementarity doctrine, the Bohr-Einstein debates (whose 1935 EPR round seeded today's quantum-information science), the liquid-drop model of nuclei, and wartime advocacy for openness about the bomb. His coat of arms, chosen for Denmark's Order of the Elephant, bears the taijitu and the motto contraria sunt complementa — opposites are complementary — a fair summary of an article in which the classical world is quantum interference viewed from far away.

History

1913

— Bohr's atom: quantum rules constrained to merge with classical radiation at large quantum numbers — correspondence as method.

1925

— Heisenberg builds matrix mechanics from correspondence-surviving observables; Dirac reads the commutator as

i\hbar

times the Poisson bracket.

1926

— Wentzel, Kramers and Brillouin systematize the

\hbar

-expansion; Ehrenfest proves the mean obeys Newton.

1933

— Dirac's "The Lagrangian in Quantum Mechanics" proposes

e^{iS/\hbar}

as the quantum weight of a history.

1948

— Feynman's path integral makes the sum over histories exact calculus; stationary phase = classical mechanics.

1963

— Glauber's coherent states give "classical light" a precise quantum definition — the operational classical limit.

Going further

References. Dirac, The Principles of Quantum Mechanics, §21 and §31–32 · Feynman & Hibbs, Quantum Mechanics and Path Integrals, Ch. 2–3 · Landau & Lifshitz, Quantum Mechanics, §6–7 (quasi-classical case) · Goldstein, Poole & Safko, Classical Mechanics, 3rd ed., Ch. 10 (Hamilton-Jacobi) and Ch. 13 · Zurek, "Decoherence, einselection, and the quantum origins of the classical," Rev. Mod. Phys. 75 (2003) 715

Provenance. A verifier should: (1) confirm the canonical commutator $[\hat\phi,\hat\pi]=i\hbar\delta^{(3)}$ and check that dividing by $i\hbar$ reproduces the field Poisson bracket (Dirac, Principles of QM §21; the field-Poisson-brackets article); (2) reproduce the stationary-phase reduction of the path integral to $\delta S/\delta\phi=0$ (standard saddle-point; consistent with Hamilton's principle in Goldstein Ch. 2, Morin Ch. 6); (3) verify the Ehrenfest derivation by computing $[\hat\phi,\hat H]$ and $[\hat\pi,\hat H]$ for the scalar Hamiltonian and confirming the free-field case yields $(\Box+m^2)\langle\hat\phi\rangle=0$ ; (4) check the WKB orders against the particle Hamilton–Jacobi equation in Landau–Lifshitz QM §6. Additionally for this upgrade: worked-example numbers recomputed directly (pendulum $E = 4.9\times10^{-3}$ J, $S/\hbar \approx 9\times10^{31}$ ; laser $E_\gamma = 3.1\times10^{-19}$ J, $N = 3.2\times10^9$ , $1/\sqrt{N} = 1.8\times10^{-5}$ ; Gaussian moment $\langle\phi^3\rangle = \bar\phi^3 + 3\bar\phi\sigma^2$ giving correction $0.015$ on $0.625$ = $2.4\%$ , mass shift $\delta m^2 = \lambda\sigma^2/2 = 0.030$ ). Scope note: the particle WKB/eikonal story is owned by the Hamilton-Jacobi chapter's WKB and Schrödinger-connection articles; the bracket algebra by the Poisson-brackets article; the oscillator bank and occupation criterion by the Hamiltonian article; the forward quantization by the capstone.

Builds on: The Lagrangian Density and Field Action · Covariant Euler-Lagrange Equations for Fields · Hamiltonian Formulation for Fields · Poisson Brackets for Fields · Functional Derivatives and Field Variations Leads to: From the Action Principle to Quantum Field Theory · Noether's Theorem for Continuous Symmetries · The Canonical Stress-Energy Tensor

If ħ never actually changes, why do we see a classical world at all?

Because the relevant ratios are astronomically large for macroscopic variables — and because of decoherence, which the bare ratios do not capture. The ratio story: a pendulum at $S/\hbar \sim 10^{31}$ has a stationary-phase window so narrow that quantum spreading of its center of mass would take longer than the age of the universe. But ratios alone leave a loophole: quantum mechanics permits superpositions of two macroscopically distinct near-classical bundles (a field lump here and there), each internally classical, jointly nothing of the sort. Decoherence closes the loophole: macroscopic variables are relentlessly monitored by their environments (photons scatter off them, gas molecules collide, phonons couple), and the environment's records destroy interference between distinct bundles at rates that grow ferociously with separation and size — for a dust grain superposed over a micron, air-molecule scattering kills coherence in about $10^{-31}$ s. What survives environmental monitoring is precisely the set of localized, coherent-state-like configurations riding classical trajectories — the "pointer states." So the honest modern answer has two layers: stationary phase explains why classical dynamics is accurate; decoherence explains why classical states are the only ones we encounter. Neither invokes any physics beyond quantum mechanics itself.

Is the classical limit of quantum field theory classical particles or classical fields?

Both — through two different limits, and distinguishing them dissolves a century of confusion. Classical field theory emerges at large occupation: many quanta piling coherently into the same modes ( $N \gg 1$ , coherent states), fluctuations $1/\sqrt{N}$ — the laser, the radio wave, this chapter's subject matter. Classical particle mechanics emerges from the opposite corner: few quanta, but with well-localized wave packets whose actions dwarf $\hbar$ — a single electron in a bubble chamber follows a classical trajectory not because its field is highly occupied (it is one quantum!) but because its packet's phase varies by many turns over its width, and stationary phase pins it to the classical path (the WKB/eikonal story of the Hamilton-Jacobi chapter). The two limits are inequivalent and can exclude each other: light in a single-photon state has no classical field limit (its field fluctuations dominate its mean) yet exhibits particle-like trajectories; a Bose-Einstein condensate is the field limit of atoms usually treated as particles. Planck's blackbody puzzle sat exactly at the crossroads — photons are bosons, so both limits are available to light, and which one describes an experiment depends on occupation per mode ( $N \gg 1$ : Rayleigh-Jeans, classical field; $N \ll 1$ : Wien, particle-like). That the same theory contains both classical worlds, selected by state rather than by law, is among QFT's quiet triumphs.

Do the Noether charges and the symmetry structure survive the classical limit intact?

Yes — and more precisely than "survive": they are the invariant skeleton the limit is organized around. Dirac's dictionary maps the quantum charge algebra $[\hat Q^A, \hat Q^B] = i\hbar f^{ABC}\hat Q^C$ onto the classical $\{Q^A, Q^B\} = f^{ABC}Q^C$ of the Poisson-brackets article with the same structure constants — the $\hbar$ 's cancel in the correspondence, exactly as they did in Ehrenfest's Step 2, so the symmetry algebra is $\hbar$ -independent and passes through the limit untouched. Conserved quantities remain conserved ( $[\hat Q, \hat H] = 0 \leftrightarrow \{Q, H\} = 0$ ), generators keep generating, and the Poincaré algebra certifying Lorentz covariance is the same algebra on both sides. The exceptions are the celebrated ones flagged in the bracket article's FAQ: anomalies, where the quantum theory cannot preserve all classical brackets simultaneously and a symmetry of the action dies at order $\hbar$ (the axial current's divergence picks up an $\hbar$ -proportional term). Anomalies are the precise, computable failures of this FAQ's "yes" — and their existence sharpens the general lesson: the classical limit preserves the symmetry skeleton except where quantization itself obstructs, and those obstructions are physics (the $\pi^0$ decay, anomaly-cancellation constraints on the Standard Model), not bookkeeping accidents.

How long does correspondence last for a chaotic system?

Strikingly briefly — logarithmically in $\hbar$ — and this is the sharpest known limitation of the classical limit. The Ehrenfest picture requires the quantum packet to stay narrow relative to the scales over which forces vary. In a regular system, packets spread polynomially and the classical description of means holds for enormous times. In a chaotic system (the M10 chapter's subject), neighbouring trajectories separate as $e^{\lambda t}$ with $\lambda$ the Lyapunov exponent — and the packet's width inherits the exponential stretching. Starting from the minimal quantum width $\sim\sqrt{\hbar}$ , the packet reaches macroscopic scale $L$ at the Ehrenfest time $t_E \sim \lambda^{-1}\ln(L^2/\hbar)$ : the logarithm converts thirty orders of magnitude of $S/\hbar$ into a mere factor of seventy in time. For the chaotic tumbling of Saturn's moon Hyperion, naive estimates give $t_E$ of order decades — after which the moon's orientation "should" be a grotesque quantum superposition. The resolution is the decoherence layer again: sunlight scattering off Hyperion monitors its orientation and re-localizes it continuously, restoring classicality on timescales absurdly shorter than $t_E$ . The pairing is general and beautiful: chaos degrades isolated-system correspondence exponentially fast, while the same sensitivity makes chaotic systems hyper-vulnerable to decoherence, which rescues classicality — quantum-classical correspondence in a chaotic world is a race the environment always wins.

The path integral sums non-differentiable, wildly non-classical histories. How can such an object know about smooth classical solutions?

Through phase statistics, not through any preference in the measure — and the mechanism repays a careful look. Typical histories in the functional integral are indeed jagged (in the rigorous construction, almost none are differentiable), and each contributes an arrow of unit length; smooth classical solutions are an infinitely thin subset carrying zero weight by counting. What distinguishes them is organizational: around a generic jagged history, the action varies linearly under small deformations, so arrows from its neighbourhood spin through all angles and annihilate; around a stationary history, the action varies only quadratically, so an anomalously large neighbourhood — of thickness $\sim\sqrt{\hbar}$ in field-deformation norm — shares nearly one phase and votes as a bloc. Classicality is the coordination of votes, not the number. Two corollaries sharpen the picture. The $\sqrt{\hbar}$ -thick winning coalition is the quantum fluctuation band: the "classical solution" of the limit is really a tube of histories, which is why one-loop determinants (the tube's width bookkeeping) accompany every semiclassical amplitude. And when no real stationary point connects the boundary conditions — tunnelling — the dominant coordination happens at complex saddles, giving amplitudes $\sim e^{-S_E/\hbar}$ : exponentially small, nonzero, and computed by exactly the same principle. The classical world is the real-saddle sector of a democracy of histories; the quantum corrections are the minority reports.

Where does this article leave the chapter — what exactly does the capstone still have to do?

This article certified the return direction: given quantum field theory, classical field theory emerges at stationary phase and large occupation, with every canonical structure surviving as the $\hbar$ -independent skeleton. The capstone — the action-principle-to-QFT article — walks the forward direction as a construction: take this chapter's finished classical apparatus (the action and its functional calculus; the canonical pairs and equal-time brackets; the mode-diagonalized oscillator bank; the Noether charges and their algebra) and perform Dirac's substitution $\{,\} \to [,]/i\hbar$ once, promoting mode amplitudes to ladder operators and mode quanta to particles. What makes the forward walk more than a formality is everything this chapter proved in advance: covariance (the Poincaré algebra will be represented on the quantum states), positivity (the Hamiltonian article's $\mathcal{H} \geq 0$ becomes a well-defined vacuum), completeness (the two frequency branches become particles and antiparticles per the Klein-Gordon article), and the charge ledger (the internal-symmetries article's signed count becomes the particle-number operators). This article guarantees the round trip closes: quantize, then take $\hbar \to 0$ through the ratios above, and you land exactly back on these eleven modules. The bridge is load-tested in both directions; the capstone drives across it.