The correspondence principle states that quantum mechanics must reduce to classical mechanics in the limit of large quantum numbers or large action (ℏ → 0). Expectation values of quantum operators should reproduce classical equations of motion; energy eigenvalues should become quasi-continuous and classically behaved. This principle constrains quantum theory and shows that classical physics is an emergent large-scale limit of quantum mechanics.
You already know from quantum operators that physical observables are represented by Hermitian operators, and that measuring an observable on a general state yields a probabilistic distribution of eigenvalues. The correspondence principle is the requirement that this quantum formalism must, in the right limit, reproduce the deterministic trajectories and continuous energy values of classical mechanics. It is not merely a consistency check — historically, it guided the construction of quantum mechanics, and it remains a useful tool for building intuition.
The clearest version is Ehrenfest's theorem: the *expectation values* of quantum operators obey the same equations as classical observables. Specifically, d⟨x⟩/dt = ⟨p⟩/m and d⟨p⟩/dt = −⟨dV/dx⟩. These look exactly like Newton's second law, with quantum expectation values in place of classical variables. For a sufficiently narrow wavepacket — one localized enough that V doesn't vary much over its width — ⟨dV/dx⟩ ≈ dV(⟨x⟩)/dx, and the wavepacket's center moves like a classical particle. The quantum "smearing" is negligible when ℏ is negligible compared to the relevant action scales. Classical mechanics is not wrong — it is the limit of quantum mechanics that applies when action ≫ ℏ.
For bound states, the classical limit appears through large quantum numbers. Consider the hydrogen atom: energy levels are E_n = −13.6 eV/n². The energy *spacing* between adjacent levels is ΔE = E_{n+1} − E_n ≈ 27.2 eV/n³, which shrinks as n → ∞. For large n, the levels are so densely packed that they appear continuous — just as classical mechanics predicts a continuous range of allowed energies. The frequency of the emitted photon (from n → n−1) approaches the classical orbital frequency of the electron. For a particle in a box, the momentum eigenvalues p_n = nπℏ/L become dense for large n, and the quantum wavefunction oscillates so rapidly that its probability density |ψ|² averages to the classical uniform distribution (equal probability everywhere). High quantum numbers erase quantum discreteness.
The deepest formulation comes from the path integral: quantum mechanics assigns probability amplitudes to all possible paths between two points, not just the classical one. In the limit ℏ → 0, the phase of each path's amplitude oscillates wildly, and the contributions cancel almost everywhere — except near the path where the phase is stationary, which is precisely the path of stationary action: the classical trajectory. Classical mechanics is not a separate theory imposed by fiat; it is the saddle-point approximation to quantum mechanics. This perspective explains why quantum effects matter when different paths have phases of order ℏ or less (microscopic systems) and why they vanish for macroscopic objects, where the action along any conceivable path vastly exceeds ℏ.