Speaker
Description
Quantum sensing protocols based on superconducting qubits are emerging as promising tools for applications ranging from fundamental physics to the search for axion dark matter with haloscopes. Achieving high-sensitivity photon detection with low dark-count rates is crucial for resolving single-photon wavepackets and weak coherent fields. A promising platform employs networks of superconducting qubits dispersively coupled to a microwave resonator. However, accurately modeling and optimizing these systems under realistic noise conditions requires numerical methods beyond tractable analytical approaches.
We present QSOpt (Quantum Sensing Optimization), a simulation and optimization framework designed for networks of superconducting qubits coupled to a resonator. The key innovation is the application of gradient-based optimization techniques to photon detection protocols in regimes where noise and system complexity preclude analytical solutions. QSOpt integrates the QuTiP quantum simulation library with a JAX backend, enabling efficient automatic differentiation and scalable optimization.
The framework supports joint optimization of hardware parameters, such as dispersive shifts and cavity couplings, and quantum circuits for state preparation. The latter is implemented within a quantum machine learning paradigm, allowing exploration of both separable and entangled multi-qubit states. This approach enhances detection performance under realistic noise and enables the discovery of previously unexplored multi-qubit sensing protocols.