Automated Model Selection with First-Order Gauge-Invariant Parameters

Gate set tomography is one of the most powerful tools available for the characterization of real-world quantum computers. However, this power comes with great cost, in part because the space of all possible errors that a quantum processor may experience is incredibly vast. In practice we observe that in real experiments only a small fraction of possible errors are relevant for any given device, making the rest unnecessary for the description of that quantum computer’s operations. The inclusion of such unneeded errors in our models makes the task of performing and interpreting tomography results harder. Automated model selection is an algorithm that solves this problem by finding a model with the least number of parameters that is still able to effectively describe the data collected from a device. An obstacle for this algorithm, is that due to a property called gauge freedom, many different models give the same physical predictions and thus are equally effective in representing empirical data. As a consequence, traversing reduced-model space to identify which parameters are necessary to describe a device is not trivial. In this project, we implement the automated model selection algorithm using first-order gauge-invariant (FOGI) parameters, which eliminates the gauge freedom problem and simplifies the landscape.