Quantum Zeno Effect

The setup — Rabi oscillation interrupted

Take a two-level quantum system: a ground state |g⟩ and excited state |e⟩, coupled by a constant Hamiltonian H that drives Rabi oscillations between them with angular frequency Ω. Prepare the atom in |g⟩ at t = 0. Without measurement, its population oscillates: P_g(t) = cos²(Ωt/2), reaching zero (fully excited) at t = π/Ω. Standard quantum mechanics.

Now interrupt: at intervals of Δt = t/N during the evolution, perform a projective measurement asking "is the atom in |g⟩?" Each time the answer is "yes" (with probability close to one), and the wavefunction is reprojected onto |g⟩. The atom restarts evolution from scratch every Δt.

For very small Δt, the short-time survival probability is

P_survive(Δt) = |⟨g|U(Δt)|g⟩|² = cos²(ΩΔt/2) ≈ 1 − (ΩΔt/2)²

The leading correction is quadratic in Δt, not linear — this is the key. Linear loss would mean exponential decay even under measurement; quadratic loss makes the loss per step vanish as Δt → 0.

After N measurements over total time t = NΔt:

P_survive(t) = [P_survive(Δt)]^N ≈ [1 − (Ωt/2N)²]^N → 1 as N → ∞

Take the logarithm and expand: ln P_N ≈ −N(Ωt/2N)² = −(Ωt)²/(4N) → 0. The atom is frozen in its initial state.

Worked example — Itano 1990 in numbers

NIST's 1990 experiment used a ⁹Be⁺ ion driven by a 320 MHz microwave field for T = 256 ms. The Rabi frequency was tuned to give a complete inversion at the end of T (ΩT = π).

• Without measurement: probability of having transitioned at t = T is sin²(ΩT/2) = sin²(π/2) = 1.
• With N = 1 measurement at T/2: transition probability ≈ 2 × sin²(π/4)(1 − sin²(π/4)) ≈ 0.5.
• With N = 4: ≈ 0.36.
• With N = 16: ≈ 0.10.
• With N = 64: ≈ 0.025.

The measured points fell exactly on the Zeno curve P ≈ sin²(π/(2N)) ≈ (π/2N)² for large N — proving the dependence is not exponential (which would give a constant rate) but quadratic-then-squared, the Zeno signature.

Zeno time τ_Z

For any initial state |ψ⟩ and Hamiltonian H, the natural timescale for short-time evolution is

τ_Z = ℏ / √(⟨H²⟩ − ⟨H⟩²)  (Zeno time)

This is the half-width of the energy distribution of |ψ⟩ in units of frequency. For measurements much faster than τ_Z the Zeno regime holds. For measurements much slower than τ_Z you recover ordinary exponential or oscillatory decay. The crossover between Zeno suppression and Anti-Zeno acceleration happens when Δt is comparable to the bath correlation time, not τ_Z directly.

Δt vs τ_Z	Regime	Survival after t
Δt ≪ τ_Z	Zeno (frozen)	P → 1 as Δt → 0
Δt ~ τ_Z	Transition	Sub-exponential, model-dependent
Δt ≪ τ_bath	Anti-Zeno possible	P decays faster than unitary
Δt ≫ τ_bath	Standard decay	P ≈ exp(−Γt)
No measurement	Unitary	cos²(Ωt/2) (Rabi)
Continuous monitoring	Strong-coupling Zeno	Dynamically decoupled subspace

JavaScript — simulating Rabi vs Zeno

// Two-level system, Rabi oscillation, measured every Δt
function rabiSurvival(t, Omega) {
  // P(|g⟩ at time t | start in |g⟩) = cos²(Ω t / 2)
  return Math.cos(Omega * t / 2) ** 2;
}

function zenoSurvivalAfterN(totalT, Omega, N) {
  const dt = totalT / N;
  const perStep = rabiSurvival(dt, Omega);  // each step survives with this prob
  return Math.pow(perStep, N);
}

// Itano-style parameters: Ω·T = π
const T = 256e-3;        // seconds
const Omega = Math.PI / T;

console.log('Unmeasured P(transition at T):', (1 - rabiSurvival(T, Omega)).toFixed(3));   // 1
console.log('N=1  P(survive in |g⟩):', zenoSurvivalAfterN(T, Omega, 1).toFixed(4));         // 0
console.log('N=2  P(survive):',         zenoSurvivalAfterN(T, Omega, 2).toFixed(4));         // 0.5
console.log('N=4  P(survive):',         zenoSurvivalAfterN(T, Omega, 4).toFixed(4));         // 0.640
console.log('N=8  P(survive):',         zenoSurvivalAfterN(T, Omega, 8).toFixed(4));         // 0.795
console.log('N=16 P(survive):',         zenoSurvivalAfterN(T, Omega, 16).toFixed(4));        // 0.892
console.log('N=64 P(survive):',         zenoSurvivalAfterN(T, Omega, 64).toFixed(4));        // 0.971
console.log('N=256 P(survive):',        zenoSurvivalAfterN(T, Omega, 256).toFixed(4));       // 0.993

// Asymptotic Zeno scaling: P ≈ 1 − (ΩT/(2√N))² for large N
function zenoAsymptotic(N) {
  return 1 - (Omega * T)**2 / (4 * N);
}
console.log('Asymp N=64:',  zenoAsymptotic(64).toFixed(4));    // 0.961
console.log('Asymp N=256:', zenoAsymptotic(256).toFixed(4));   // 0.990

// Anti-Zeno regime: if measurement timing matches a bath resonance,
// repeated projection can accelerate decay. Schematic with a Lorentzian
// bath spectrum centered at frequency ω₀, width γ:
function antiZenoRate(measInterval_dt, omega0, gamma) {
  // Zeno measurement rate
  const r = 1 / measInterval_dt;
  // For Lorentzian bath: effective rate ≈ Lorentzian(r; ω₀, γ)
  const L = (1 / Math.PI) * (gamma / ((r - omega0)**2 + gamma**2));
  return L;
}

console.log('Effective decay rate, slow measurement (Anti-Zeno):',
  antiZenoRate(1e-3, 1e3, 100).toExponential(3));
console.log('Effective decay rate, fast measurement (Zeno regime):',
  antiZenoRate(1e-6, 1e3, 100).toExponential(3));

Where the Zeno effect matters

• Quantum error correction. Stabilizer measurements project the encoded state back into the code subspace before errors accumulate — a Zeno-style protection used in surface codes and modern fault-tolerant architectures.
• Decoherence-free subspaces. Strong driving or symmetry can dynamically decouple a chosen subspace from the environment — the dynamical Zeno effect at the heart of Lloyd-Viola-Knill bang-bang decoupling and dynamical-decoupling sequences.
• Atomic clocks and interferometers. Engineering the interrogation duty cycle controls dephasing rates and lets metrologists trade off systematic vs. statistical sensitivity using Zeno scaling.
• Cold-atom physics. Bose-Einstein condensates split by frequent laser pulses survive longer; lossy lattice sites have suppressed tunneling by the dissipative Zeno effect.
• Foundational tests. The Zeno-Anti-Zeno crossover probes the structure of the system-environment correlation function — a window into non-Markovian dynamics.
• Quantum biology debates. Some proposals invoke Zeno protection for radical-pair coherence in avian magnetoreception or singlet-triplet dynamics in photosynthesis; experimental support is mixed but the mechanism is theoretically clean.

Common mistakes

• "Watching slows time." Time is unchanged. What changes is the probability distribution over states: each measurement collapses the wavefunction.
• Assuming linear short-time decay. Quantum survival is quadratic in Δt for short times; assuming linear (Markovian) decay misses the entire effect.
• Calling it the "watched pot" theorem. The aphorism is suggestive but classical pots boil from energy input. Zeno freezes amplitude flow in Hilbert space.
• Confusing it with continuous-time observation. Continuous "weak" monitoring gives a stochastic master equation; the Zeno limit is N projective measurements with N → ∞.
• "It violates the no-cloning theorem." No — projection is not cloning. Repeated projection onto |ψ⟩ doesn't create new copies; it merely keeps the same state.
• Believing it works for any system. Anti-Zeno can occur when Δt is comparable to bath correlation time. Choosing the wrong Δt accelerates decay.