Flavio Baccari - Università di Padova

 Classical simulation methods to benchmark quantum computing

Classical simulation remains an essential tool for understanding and validating quantum computers, even as they approach regimes believed to be beyond classical reach. In this talk, I will discuss two applications of large-scale classical simulation to quantum computing. In the first part, I will present results from circuit-level simulation of Shor's factoring algorithm based on tensor network methods. I will show how they allow us to assess the algorithm's real-world performance—including its success probability and resilience to hardware errors — for problem sizes and error rates that might soon be accessible to quantum devices.In the second part, I will turn to the problem of benchmarking quantum devices as they enter the quantum advantage regime, where no classical simulation of the output is accessible. I will introduce a benchmarking scheme that combines a target quantum computation with structured variants of it, whose average yields classically computable correlation functions. Crucially, these variants preserve the exact circuit architecture and depth without simplifying the gates into a classically simulable set. This average-computation approach detects noise beyond the reach of Clifford-based benchmarking and requires only a limited number of circuit realizations, offering a practical route to validating computations even where full classical simulation is impossible.