Ehrenfest Theorem: When Quantum Averages Obey Newton's Laws

What the Ehrenfest Theorem Actually Says

Quantum mechanics replaces the definite trajectory x(t) of a classical particle with a wave function ψ(x,t), from which we can only extract expectation values — probability-weighted averages such as ⟨x⟩ = ∫ ψ* x ψ dx. The natural question Ehrenfest asked in 1927 was: how do these averages move in time? Do they trace out something recognizable as a classical path?

His answer is a pair of exact equations:

• d⟨x⟩/dt = ⟨p⟩/m — the average position drifts at the average velocity, just like a Newtonian particle.
• d⟨p⟩/dt = -⟨∂V/∂x⟩ — the average momentum changes at a rate equal to the average force, -⟨∂V/∂x⟩.

Combined, they give m·d²⟨x⟩/dt² = -⟨∂V/∂x⟩, which is Newton's second law written for the mean. The theorem is not an approximation — for any state, these identities hold exactly. What is approximate is the leap to classical mechanics, which requires ⟨∂V/∂x⟩ ≈ (∂V/∂x)|₍⟨x⟩₎.

Deriving It from the Schrödinger Equation

The derivation follows from a single master formula for how any expectation value evolves. For an operator  with no explicit time dependence, d⟨Â⟩/dt = (i/ℏ)⟨[Ĥ, Â]⟩, where [Ĥ, Â] = Ĥ − ÂĤ is the commutator and Ĥ = p̂²/2m + V(x̂) is the Hamiltonian. This drops straight out of the time-dependent Schrödinger equation iℏ ∂ψ/∂t = Ĥψ and its complex conjugate, followed by integration by parts.

• For  = x̂: only the kinetic term contributes, since [p̂², x̂] = -2iℏp̂. Working through gives d⟨x⟩/dt = ⟨p̂⟩/m.
• For  = p̂: only V(x̂) contributes, because [V(x̂), p̂] = iℏ ∂V/∂x. This yields d⟨p⟩/dt = -⟨∂V/∂x⟩.

Every symbol: ℏ = 1.055×10⁻³⁴ J·s (reduced Planck constant), m = particle mass, V = potential energy, and the angle brackets denote the quantum average over ψ. No approximation enters — the two commutators do all the work, which is why Ehrenfest called it a calculation done "without approximations."

Worked Example: Where It's Exact and Where It Bends

The classical reading — ⟨force⟩ = force at ⟨position⟩ — is exact whenever V is at most quadratic in x, because then ∂V/∂x is linear, and the average of a linear function equals the function of the average.

• Free particle (V = 0): ⟨x⟩(t) = ⟨x⟩₀ + (⟨p⟩₀/m)t. A perfect straight line, forever.
• Uniform field (V = -Fx, e.g. an electron in E = 10⁶ V/m feels F = 1.6×10⁻¹³ N): ⟨x⟩ follows uniform acceleration a = F/m = 1.8×10¹⁷ m/s².
• Harmonic oscillator (V = ½mω²x²): d²⟨x⟩/dt² = -ω²⟨x⟩ exactly — the centroid oscillates at the classical frequency ω regardless of the state.

Now take a cubic anharmonic term, V = ½mω²x² + λx³. Then ∂V/∂x = mω²x + 3λx², and ⟨x²⟩ = ⟨x⟩² + (Δx)². The extra 3λ(Δx)² term means the average force knows about the wave-packet width. As the packet spreads, this correction grows and the classical trajectory and quantum centroid diverge — the Ehrenfest gap.

How It's Observed and Used

The theorem is verified every time a localized wave packet is tracked. In ultracold-atom experiments, a Bose–Einstein condensate or a single trapped ion is a minimum-uncertainty packet whose centroid, imaged by absorption or fluorescence, follows Newtonian motion in the trap to sub-micron precision — a direct Ehrenfest measurement. Rydberg wave packets, launched by short laser pulses in atoms like potassium, orbit the nucleus classically for a few periods before quantum revivals set in.

• Semiclassical dynamics: computational chemistry propagates nuclei on potential surfaces using Ehrenfest's mean-field trajectories ("Ehrenfest dynamics"), the basis of many nonadiabatic molecular-dynamics codes.
• Scanning tunneling and STM manipulation rely on the mean position of tunneling electrons behaving predictably.
• Accelerator and beam physics treat charged-particle bunches via their centroid, an Ehrenfest-justified move.

The theorem also underlies why macroscopic objects — a baseball with Δx/x ~ 10⁻³⁴ — never show quantum weirdness in their motion: their packets are absurdly narrow relative to any potential's curvature.

How It Differs from the Correspondence Principle and the Classical Limit

The Ehrenfest theorem is often confused with Bohr's correspondence principle (1920), but they are distinct. Bohr's principle is a heuristic: quantum predictions should match classical ones for large quantum numbers n. Ehrenfest's is a hard theorem about expectation values that holds for all states, large or small n.

• It is not a full classical limit. Recovering Newton's law for the centroid still requires the packet to stay narrow (Δx small compared to the length over which the force changes appreciably).
• It differs from the WKB / semiclassical approximation, which expands ψ in powers of ℏ; Ehrenfest makes no ℏ-expansion at all.
• It is weaker than decoherence, which explains why interference between paths vanishes; Ehrenfest says nothing about the loss of coherence, only about mean motion.

A famous subtlety: even when the centroid tracks Newton exactly (SHO), the quantum state is nothing like a classical point — it may be a spread-out superposition. Ehrenfest constrains the average, not the full distribution.

Significance, Caveats, and Open Questions

Ehrenfest's two-page note answered a question that had troubled the founders of quantum theory: does the new mechanics contain the old one? The answer — yes, on average, under stated conditions — made the theory credible and gave physicists a precise criterion for when classical intuition is safe.

• Regime of validity: the classical reading requires ⟨V'(x)⟩ ≈ V'(⟨x⟩). This fails for anharmonic potentials once the packet width (Δx) is comparable to the potential's variation scale — the essence of quantum chaos studies, where the "Ehrenfest time" t_E ~ (1/λ)·ln(S/ℏ) marks when a chaotic system's quantum and classical predictions part ways.
• Subtle cases: the theorem needs surface terms in the integration by parts to vanish; for a particle in a box or with hard walls, careful treatment of boundary contributions is required, and naive application can mislead.
• Open direction: relativistic and field-theoretic generalizations (Ehrenfest relations for Dirac particles and centroids) remain an active topic, as does the precise role of the Ehrenfest time in the emergence of classicality.

Classical Newtonian motion vs. the Ehrenfest theorem for the quantum average

Aspect	Classical mechanics	Ehrenfest theorem (quantum)
Equation of motion	dp/dt = -dV/dx at x	d⟨p⟩/dt = -⟨∂V/∂x⟩ over ψ
Force used	F(x) at the particle position	Average of F over the wave packet
Free particle	x(t) = x₀ + (p/m)t	⟨x⟩(t) = ⟨x⟩₀ + (⟨p⟩/m)t (exact)
Harmonic oscillator	x(t) = A·cos(ωt+φ)	⟨x⟩ obeys ẍ = -ω²⟨x⟩ (exact)
Anharmonic / cubic V	Single deterministic path	⟨F(x)⟩ ≠ F(⟨x⟩); classical law fails
Uncertainty	None (point particle)	Δx·Δp ≥ ℏ/2 always present