Harnessing multimode nonlinearity for quantum state measurement using quantum reservoir computing

Nonlinearity is often considered parasitic in circuit QED measurement chains, with substantial efforts made to operate quantum devices such as amplifiers in linear regimes. In this work we show how the quantum nonlinearity of quantum devices - referred to as quantum reservoirs - can in fact be harnessed as a resource in measurement chains for resource-efficient processing of quantum signals. We apply our general framework to the task of classifying two-mode squeezed states that differ only in the quadrature of joint squeezing, a task relevant to recent interferometric approaches to dispersive qubit readout [2]. We show that using simple quantum reservoirs consisting of coupled Kerr oscillators in the measurement chain obviates the need for nonlinear post-processing or state-conditioned joint readout, and instead enables quantum state classification using a simple linear discriminator on in-principle a single measured quadrature. We also precisely quantify the role of entanglement of the quantum reservoir outputs, which can reduce the noise in the optimal measured quadrature to below its vacuum value, leading to improved classification performance. Our work paves the way for the consideration of nonlinear multimode quantum devices as general cryogenic processors for quantum systems.

[1] S. Khan et al., arXiv: 2110.13849

[2] Liu et al., arXiv: 2007.15460