with Dr. Shane McQuarrie
I'm working on additions to the OpInf package and associated research. I helped create a tutorial on preprocessing techniques like Lift and Learn. Now we're exploring what differentiable programming opens up for operator inference — can an OpInf-derived ROM do parameter recovery? Can trajectory-based post-tuning of operators and bases actually learn the underlying dynamics?
python · jax · model reduction · inverse problems · differential equations
with Dr. AJ Rasmusson
I'm designing an adaptive protocol to estimate the relevant parameters of an ion trap without the Lamb-Dicke approximation, so calibration still works in warm or fast regimes. I built a digital twin of a single ion trap in JAX for real time control. A gradient-based pulse optimizer with sequential Monte Carlo finds the pulse sequence with maximum expected information gain.
quantum information · trapped ions · bayesian inference · digital twins · optimal control