A Physics-based, AI-accelerated Codesign Approach for the Optimization of Superconducting Quantum Circuits

Radiation impacts are a leading cause of information degredation in superconducting quantum devices due to their frequency in occurrence and instigation of widespread correlated errors. Recent developments have built Monte-Carlo models using Geant4 and G4CMP to track the generation of e/h pairs, phonons, and superconducting quasiparticles. Building upon this work, we have developed additional models to track the quantum errors and correction failures in superconducting quantum processors (QPUs) during radiation impacts. This framework of models can be used to develop new architecture and quantum error correction (QEC) code designs that better resist correlated errors from radiation impacts. To best exploit this model framework, a machine learning environment has been designed to optimize QPU performance. This approach is holistic, taking into consideration both QPU and QEC designs and their interplay. In this presentation, we will discuss both the model and our machine learning results.