Concepts and conditions for error suppression through randomized compiling

Randomized compiling reduces the effects of errors on quantum computers by tailoring arbitrary Markovian errors into stochastic Pauli noise. During this talk, we prove that randomized compiling also tailors non-Markovian errors into local stochastic Pauli noise and investigate the technique's limitations. We show through analysis and numerical results that randomized compiling alters errors in three distinct helpful ways. First, it prevents the coherent accumulation of errors (including hard to remove crosstalk effects) across gate cycles by destroying intercycle coherent correlations. Second, it converts individual gate cycle errors into Pauli noise. Finally, randomized compiling reduces the variability inherent to noisy devices. We confirm these theoretical predictions with the IBM Quantum platform and describe experimental data that illustrates a drastic performance improvement across public devices. These results cement the importance of randomized compiling in near- and long-term quantum information processing.