A a new concept appeared in the literature of quantum physics in the 1980s, under the name of “quantum nondemolition [QND] measurements.”1 The idea quickly became very popular, and it is now used in many contexts, with a wide agreement about its meaning and usefulness.2–4
However, Christopher Monroe argued in a letter to PHYSICS TODAY (January 2011, page 8) that QND measurements are a useless concept because they are just standard quantum measurements, as defined, for example, by John von Neumann at the dawn of quantum mechanics. Here I argue that this view is too schematic, to say the least.
Measuring the polarization of a photon—a representative quantum measurement—typically destroys the photon. The simple but crucial question addressed by QND measurements is the following: Is it possible to perform a quantum measurement in such a way that the system will continue to exist even though its state may be altered by the measurement that has been performed? The answer is yes, but one has then to use an indirect measurement rather than a direct one, and that is exactly where QND measurements come into play1,2—in quite a useful way.
For such a measurement to work, the trick is to use an auxiliary quantum system, usually called the “meter,” and to devise the system–meter interaction in such a way that entanglement is created between the two. Then a direct measurement performed on the meter will result in a projective measurement on the system. Therefore, the system will evolve into a post-measurement state as expected, but it will not be demolished; hence the name quantum nondemolition.
The crucial role of the meter was already pointed out by von Neumann in the 1930s, but QND gives condi-tions for the measurement chain—entanglement, direct meter measurement, projecting the system onto the final state—to work properly.1 For instance, the interaction Hamiltonian between the system and the meter should commute with the system observable to be measured. In addition, various QND criteria have been introduced to characterize the quality of such measurements.2 Those criteria are useful tools to quantify the success of a real QND measurement; they are actually quite flexible and may be adapted to any given experimental situation, with continuous2 or discrete3,4 observables.
Summarizing, the basic goal of the concept of QND measurement is to define, implement, and quantify a chain of measuring events. It may be said that a QND measurement is the “experimental grease” counterpart of the pure theoretical concept of an ideal projective von Neumann measurement.
Performing a QND measurement is not always required; sometimes a direct readout is quite enough. But when a QND measurement is required—for instance, as an intermediate step in a quantum calculation—understanding and controlling the chain of events becomes quite crucial. That is why QND measurements, beyond being useful, become building blocks of more complex operations as soon as individual quantum objects are manipulated and measured.
I thank Alain Aspect for helpful comments.
