Laboratoire de Physique Théorique
Toulouse - UMR 5152

It is common experimental practice to improve the signal-to-noise ratio of measurements by averaging measurements of N identically prepared systems. If the systems are independent, the overall sensitivity of the measurement improves as the inverse square root of N, a scaling known as "standard quantum limit" or "shot noise limit" (SQL). One of the most promising spin-offs of quantum information theory for science is the idea of using quantum information processing in order to increase the sensitivity of precision measurements, an approach known as "quantum enhanced measurements" (QEM). Indeed, if the N systems are initially entangled, one may achieve in principle a 1/N scaling of the sensitivity, known as the ``Heisenberg limit’’ (HL). This would represent a major improvement with important consequences for a broad range of fields, from gravitational wave sensing over inertial navigation to medical imaging.

However, in spite of about 30 years of efforts, the SQL has been surpassed only by very few experiments so far, and only for very small values of N. Indeed, the standard protocols of QEM require highly entangled states that are typically very prone to decoherence, and are therefore unlikely to scale up to the large numbers of N required before QEM can compete with classical precision measurements.

In this talk I introduce as an alternative our recently developped method of "coherent averaging", that enables Heisenberg limited sensitivity, without using or ever creating any entanglement. The method is robust under local decoherence, in the sense that local decoherence changes only the prefactor but not the scaling with N. I present a general theoretical framework for this new kind of measurement scheme, and propose a possible application in high precision measurements of the length of an optical cavity.