Simple Algorithm
Quantum Computing · Rydberg State · 24 min

Interaction-Free Measurement

The Elitzur–Vaidman bomb tester.

A bomb explodes if a single photon touches its trigger. Can you confirm that a live bomb is present without setting it off? Elitzur and Vaidman showed that interference says yes, sometimes. The quantum Zeno effect says yes, almost always.

What you’ll learn

  • Derive the Elitzur–Vaidman outcome statistics for arbitrary beam-splitter reflectivity and prove the efficiency bound η=T/(1+T)<12\eta = T/(1+T) < \tfrac12.
  • Prove the quantum-Zeno bound Psuccess=cos⁡2N ⁣π2N≥1−π24NP_{\text{success}} = \cos^{2N}\!\frac{\pi}{2N} \ge 1 - \frac{\pi^2}{4N}.
  • Model the bomb as a which-path measurement and simulate the Zeno protocol as a circuit.
  • Quantify how optical loss limits the achievable efficiency, and state precisely what "interaction-free" means.

The Elitzur–Vaidman tester

Tune a Mach–Zehnder interferometer so that D1 is dark (φ=0\varphi = 0 in our convention). Place the bomb so that its trigger would absorb any photon in the upper arm. A dud is transparent and leaves the interferometer intact. A live bomb is a perfect which-path detector: it removes one path, and with it the cancellation at D1.

BombD0D1Explosion
dud (transparent)110000
live, 50:50 splitters14\tfrac14 (inconclusive)14\tfrac14 (certified, unexploded)12\tfrac12
DerivationUnbalanced splitters and the efficiency bound
  1. Let BS₁ send the photon into the bomb arm with probability RR and into the free arm with probability T=1−RT = 1 - R, with amplitudes R\sqrt{R} and T\sqrt{T} up to phases.

  2. T x+R y=0,∣x∣2+∣y∣2=1  ⇒  ∣x∣2=R\sqrt{T}\,x + \sqrt{R}\,y = 0,\qquad \abs{x}^2 + \abs{y}^2 = 1 \;\Rightarrow\; \abs{x}^2 = R

    BS₂ must make D1 dark with no bomb. Here xx and yy are its amplitudes from the free and bomb arms to D1 (a row of a unitary).

  3. P(boom)=R,P(D1)=T ∣x∣2=TR,P(D0)=T2P(\text{boom}) = R, \qquad P(\text{D1}) = T\,\abs{x}^2 = TR, \qquad P(\text{D0}) = T^2

    A live bomb absorbs the bomb-arm amplitude; the free-arm amplitude then splits at BS₂ with no partner to cancel.

  4. η=P(D1)P(D1)+P(boom)=TRTR+R=T1+T\eta = \frac{P(\text{D1})}{P(\text{D1}) + P(\text{boom})} = \frac{TR}{TR + R} = \frac{T}{1+T}

    Repeating every inconclusive (D0) run, η\eta is the fraction of live bombs eventually certified unexploded.

  5. η=13  at T=12,η↗12  as T→1\eta = \tfrac13 \ \text{ at } T = \tfrac12, \qquad \eta \nearrow \tfrac12 \ \text{ as } T\to 1

    No choice of splitters beats one half: the scheme is fundamentally lossy.

Check your understanding

With 50:50 splitters, you repeat the test whenever D0 clicks. What fraction of live bombs is eventually certified without exploding?

  1. 14\tfrac14
  2. 13\tfrac13
  3. 12\tfrac12
  4. 34\tfrac34
Show the answer

B. 13\tfrac13

Each conclusive run ends in D1 (probability 14\tfrac14) or an explosion (probability 12\tfrac12), so the certified fraction is 1/41/4+1/2=13\tfrac{1/4}{1/4 + 1/2} = \tfrac13, matching T/(1+T)T/(1+T) at T=12T=\tfrac12.

Zeno: watching keeps the photon out

Kwiat, Weinfurter, Herzog, Zeilinger and Kasevich (1995) proposed a scheme whose efficiency approaches 11. Let the photon's polarization be the qubit, ∣H⟩=∣0⟩\ket{H} = \ket{0} and ∣V⟩=∣1⟩\ket{V} = \ket{1}. Each of NN cycles rotates the polarization by π/(2N)\pi/(2N); then a polarizing beam splitter routes VV through the bomb arm and HH through a free arm, and recombines them. With no bomb the small rotations accumulate coherently. With a live bomb, every cycle is a measurement of "VV?", and frequent measurement freezes the evolution: the quantum Zeno effect (Misra & Sudarshan, 1977).

DerivationSurvival probability
  1. one cycle: RY(π/N),cos⁡π2N∣H⟩+sin⁡π2N∣V⟩\text{one cycle: } R_Y(\pi/N), \qquad \cos\tfrac{\pi}{2N}\ket{H} + \sin\tfrac{\pi}{2N}\ket{V}

    A polarization rotation by δ\delta is RY(2δ)R_Y(2\delta) on the Bloch sphere; here δ=π/(2N)\delta = \pi/(2N).

  2. dud: RY(π/N)N∣H⟩=RY(π)∣H⟩=∣V⟩\text{dud: } R_Y(\pi/N)^N\ket{H} = R_Y(\pi)\ket{H} = \ket{V}

    No measurement: the rotations add to a quarter turn of polarization, and the photon exits VV.

  3. live: Psuccess=(cos⁡2π2N)N,photon exits ∣H⟩\text{live: } P_{\text{success}} = \left(\cos^2\tfrac{\pi}{2N}\right)^{N}, \qquad \text{photon exits } \ket{H}

    Each cycle survives with probability cos⁡2(π/2N)\cos^2(\pi/2N) and projects the photon back onto ∣H⟩\ket{H}.

  4. cos⁡2Nπ2N=(1−sin⁡2π2N)N≥1−Nsin⁡2π2N≥1−π24N\cos^{2N}\tfrac{\pi}{2N} = \left(1 - \sin^2\tfrac{\pi}{2N}\right)^N \ge 1 - N\sin^2\tfrac{\pi}{2N} \ge 1 - \frac{\pi^2}{4N}

    Bernoulli's inequality (1−u)N≥1−Nu(1-u)^N \ge 1 - Nu, then sin⁡u≤u\sin u \le u.

Predict before you look

With N=10N = 10 cycles, what is the probability (in percent) that a live bomb is certified without exploding?

Show the answer

78 %

cos⁡20(π/20)=cos⁡20(9°)≈0.98820≈0.78\cos^{20}(\pi/20) = \cos^{20}(9°) \approx 0.988^{20} \approx 0.78. The bound 1−π2/40≈0.751 - \pi^2/40 \approx 0.75 is close but slightly pessimistic.

Loss, interpretation and uses

DerivationLoss sets an optimal number of cycles
  1. P(N)≈(1−ℓ)Ncos⁡2Nπ2N≈exp⁡ ⁣(−Nℓ−π24N)P(N) \approx (1-\ell)^N\cos^{2N}\tfrac{\pi}{2N} \approx \exp\!\left(-N\ell - \frac{\pi^2}{4N}\right)

    With a probability ℓ\ell of losing the photon per cycle, more cycles cost more loss.

  2. ddN(Nℓ+π24N)=0  ⇒  N∗=π2ℓ\frac{d}{dN}\left(N\ell + \frac{\pi^2}{4N}\right) = 0 \;\Rightarrow\; N^* = \frac{\pi}{2\sqrt{\ell}}

    Balance loss against Zeno leakage.

  3. P(N∗)≈e−πℓ,ℓ=1%  ⇒  N∗≈16, P≈0.73P(N^*) \approx e^{-\pi\sqrt{\ell}}, \qquad \ell = 1\% \;\Rightarrow\; N^*\approx 16,\ P \approx 0.73

    The achievable efficiency degrades only as ℓ\sqrt{\ell}, but it never reaches 11 with lossy optics.

What is "interaction-free" about it? In every certifying run the bomb absorbs nothing: its classical state is unchanged and no energy is exchanged. The information comes from the absence of an amplitude that would otherwise have cancelled. It is a counterfactual inference: had the bomb been absent, the photon would have gone elsewhere. The bomb's presence still changes the photon's dynamics, as a measurement apparatus on one path. For a "quantum bomb" in a superposition of present and absent, the protocol entangles the bomb with the photon, which is the starting point of counterfactual computation (Hosten et al., 2006).

Check your understanding

Which statement about the Zeno bomb tester is accurate?

  1. The photon tunnels through the bomb without being absorbed
  2. In certifying runs no photon is absorbed, but the bomb's presence alters the photon's evolution by repeatedly projecting it onto ∣H⟩\ket{H}
  3. The protocol works for any object, even one that does not absorb or deflect photons
  4. The bomb is triggered, but only weakly
Show the answer

B. In certifying runs no photon is absorbed, but the bomb's presence alters the photon's evolution by repeatedly projecting it onto ∣H⟩\ket{H}

The bomb acts as a which-path measurement on the VV component. Conditioned on survival, each such measurement projects the photon back onto ∣H⟩\ket{H}, so it never accumulates the rotation that a dud allows. No absorption occurs in those runs.

Key takeaways

  • A live bomb is a which-path detector: it removes the cancellation at the dark port, so a dark-port click certifies it without absorption.
  • Elitzur–Vaidman with reflectivity R=1−TR = 1 - T: P(boom)=RP(\text{boom}) = R, P(D1)=TRP(\text{D1}) = TR, efficiency T/(1+T)<12T/(1+T) < \tfrac12.
  • The quantum Zeno scheme certifies with probability cos⁡2N(π/2N)≥1−π2/(4N)\cos^{2N}(\pi/2N) \ge 1 - \pi^2/(4N), approaching 11.
  • Loss ℓ\ell per cycle sets an optimum N∗≈π/(2ℓ)N^* \approx \pi/(2\sqrt\ell) and efficiency ≈e−πℓ\approx e^{-\pi\sqrt\ell}.

Check yourself

In the interactive lesson these are scored with your confidence, and what you miss comes back for review.

Question 1

A live bomb sits in one arm of a balanced 50:50 interferometer with a dark D1. For one photon, what are the probabilities of explosion, D0 and D1?

  1. 12\tfrac12, 14\tfrac14, 14\tfrac14
  2. 12\tfrac12, 12\tfrac12, 00
  3. 14\tfrac14, 12\tfrac12, 14\tfrac14
  4. 13\tfrac13, 13\tfrac13, 13\tfrac13
Show the answer

A. 12\tfrac12, 14\tfrac14, 14\tfrac14

Half the amplitude enters the bomb arm and is absorbed. The other half meets BS₂ alone and splits evenly: 14\tfrac14 to each detector. D1, dark with no bomb, now clicks.

Question 2

Using unbalanced splitters with transmission T→1T\to1 toward the free arm, the Elitzur–Vaidman certification efficiency approaches:

  1. 11
  2. 12\tfrac12
  3. 13\tfrac13
  4. 00
Show the answer

B. 12\tfrac12

η=T/(1+T)→12\eta = T/(1+T) \to \tfrac12. Explosions (RR) and certifications (TRTR) both become rare, but their ratio tends to 1:11:1.

Question 3

Roughly how many Zeno cycles are needed to push the explosion probability below 1%?

  1. about 10
  2. about 25
  3. about 250
  4. about 2500
Show the answer

C. about 250

P(boom)≈π2/(4N)≤0.01P(\text{boom}) \approx \pi^2/(4N) \le 0.01 gives N≥π2/0.04≈247N \ge \pi^2/0.04 \approx 247.

Question 4

In the Zeno scheme, why does a dud bomb send the photon out vertically polarized?

  1. The dud absorbs the horizontal component
  2. The polarizing beam splitter flips the polarization
  3. The outcome is random; it is vertical half the time
  4. Without any measurement, the NN small rotations add coherently to RY(π)R_Y(\pi), taking ∣H⟩\ket{H} to ∣V⟩\ket{V}
Show the answer

D. Without any measurement, the NN small rotations add coherently to RY(π)R_Y(\pi), taking ∣H⟩\ket{H} to ∣V⟩\ket{V}

With no which-path record, the PBS split and recombination act as the identity, so the NN rotations by π/(2N)\pi/(2N) compose to a full π/2\pi/2 polarization rotation.

Further reading

  • A. C. Elitzur & L. Vaidman, "Quantum mechanical interaction-free measurements", Found. Phys. 23, 987 (1993).
  • P. Kwiat, H. Weinfurter, T. Herzog, A. Zeilinger & M. A. Kasevich, "Interaction-free measurement", Phys. Rev. Lett. 74, 4763 (1995).
  • P. G. Kwiat, A. G. White, J. R. Mitchell, O. Nairz, G. Weihs, H. Weinfurter & A. Zeilinger, "High-efficiency quantum interrogation measurements via the quantum Zeno effect", Phys. Rev. Lett. 83, 4725 (1999).
  • B. Misra & E. C. G. Sudarshan, "The Zeno's paradox in quantum theory", J. Math. Phys. 18, 756 (1977).
  • O. Hosten, M. T. Rakher, J. T. Barreiro, N. A. Peters & P. G. Kwiat, "Counterfactual quantum computation through quantum interrogation", Nature 439, 949 (2006).