Realizing adaptive algorithms on FPGA for spin-qubit readout (Part 2)

Open-source controllers realized on radio-frequency (RF) enabled FPGAs allow implementation of custom control and readout blocks, typically inhibited on proprietary instruments. This enables integration of machine learning (ML) based discrimination of qubit states on-board close to the readout signal generation and acquisition. This work utilizes a modified QICK controller [1] with custom readout block designed around an ML accelerator to classify gate-defined qubits in Ge/SiGe quantum dot devices. The expected reduction in latency due to faster discrimination and information transmission on board could in future allow implementation of mid-circuit measurements and faster feedback protocols for error correction.

[1] Stefanazzi, L. et al. The QICK (Quantum Instrumentation Control Kit): Readout and control for qubits and detectors. (2021)