Automated, physics-guided, multi-parameter design optimization for superconducting quantum devices

The design of nonlinear superconducting quantum circuits often relies on time-consuming iterative electromagnetic simulations requiring manual intervention. These interventions entail, for example, adjusting design variables such as resonator lengths or Josephson junction energies to meet target parameters such as mode frequencies, decay rates, and coupling strengths. Here, we present a method to efficiently automate the optimization of superconducting circuits, which significantly reduces the need for manual intervention. The method's efficiency arises from user-defined, physics-informed, nonlinear models that guide parameter updates toward the desired targets. Additionally, we provide a full implementation of our optimization method as an open-source Python package, QDesignOptimizer. The package automates the design workflow by combining high-accuracy electromagnetic simulations in Ansys HFSS and Energy Participation Ratio (pyEPR) analysis integrated with the design tool Qiskit-Metal. Our implementation supports modular and flexible subsystem-level analysis and is easily extensible to optimize for additional parameters. The method is not specific to superconducting circuits; as such, it can be applied to a range of nonlinear optimization problems across science and technology.