Quantum State Discrimination Enhanced by FPGA-Based AI Engine Technology
Identifying the state of a quantum bit (qubit), known as quantum state discrimination, is a crucial operation in quantum computing. However, it has been the most error-prone and time-consuming operation on superconducting quantum processors. Due to stringent timing constraints and algorithmic complexity, most qubit state discrimination methods are executed offline. In this work, we present an enhanced real-time quantum state discrimination system leveraging FPGA-based AI Engine technology. A multi-layer neural network has been developed and implemented on the AMD Xilinx VCK190 FPGA platform, enabling accurate in-situ state discrimination and supporting mid-circuit measurement experiments for multiple qubits. Our approach leverages recent advancements in architecture research and design, utilizing specialized AI/ML accelerators to optimize quantum experiments and reduce the use of FPGA resources.