Accelerated Bayesian experiment design for high-speed quantum sensing

In a swept-frequency magnetic resonance measurement, most of the acquisition time is often spent on system settings with little bearing on the results. Optimal Bayesian experiment design methods use Bayesian statistics to adaptively predict which system settings yield useful data; our previous implementation of these methods has provided a 40-fold reduction in the number of data points needed to produce accurate magnetic field measurements with a nitrogen--vacancy (NV) diamond sensor. [Phys. Rev. Appl. 14, 054036 (2020)] While the acquisition speed in this example was limited by the emission rate of the NV centers, the computation time of the adaptive algorithm will ultimately become the limiting factor for stronger signals. In this presentation, a simplified adaptive Bayesian algorithm for high speed quantum sensing with NV centers is described, yielding a 2.5 fold speed improvement in settings computation over previous versions. Fixed-point arithmetic and parallel execution of the Bayesian inference, optimization, and resampling subroutines make this algorithm suitable for implementation in a field-programmable gate array (FPGA) platform, which we project will result in a 1000 fold reduction in processing time compared to the software version.