Exceptional points (EPs) in non-Hermitian systems have long been proposed as amplifiers of weak perturbations: near them, the spectral response of the system becomes extraordinarily sensitive to tiny changes. A recent theoretical study tempers that promise. In finite-dimensional Lindblad systems with a unique stationary state and a nonzero relaxation gap, EPs offer no fundamental advantage in how the observation time required for reliable detection scales with the strength of the perturbation.

The result does not deny that EPs amplify the spectral response. What it shows is that this amplification does not translate into an improvement in the exponent governing detection: the minimum observation time scales the same at an EP as in a comparable non-Hermitian reference system, provided both share the same decay rate. The advantage, when it exists, is quantitative and depends on choosing the perturbation well. The authors show that, by selecting perturbations that maximize the quantum Fisher information rate, an EP can reach a predefined detection accuracy in less observation time than that same reference system.

To illustrate this, the researchers use an optomechanical model operating at an exceptional point and analyse the detection of a weak force from the output field of the cavity. The EP-based sensor achieves the same reliability as the non-Hermitian reference in a shorter observation time. The work thus establishes fundamental limits on perturbation detection at exceptional points and, at the same time, offers design criteria for faster non-Hermitian sensors.

The implications extend to very different experimental platforms: optomechanical resonators, superconducting qubits and trapped ions. The conclusion for those designing quantum sensors is twofold: EPs are not a shortcut to a more favourable scaling law, but they can still be useful as an engineering tool if the perturbation to be measured is chosen appropriately.