Physical Model Gate Set Tomography

Gate set tomography (GST) has proven to be enormously successful for building predictive models of quantum information processor dynamics. But the process matrix models that are estimated by GST generally are described by a large number of free parameters that can be difficult to interpret. Connecting these process matrices to experimentally accessible parameters (such as laser intensity errors or magnetic field strength fluctuations) is an important step in improving devices, but is often done only in an ad hoc manner. In this talk, I'll discuss an extension of the GST framework that enables direct fitting of models for quantum devices that are expressed directly in terms of physically relevant quantities. These models often require expensive forward simulation, and so can be slow to compute and difficult to incorporate with iterative optimization routines. We overcome this with a caching and interpolation approach based on error generators. Our method enables resource-efficient GST experiments that can directly and accurately estimate experimental parameters.