How do we verify and validate a quantum computer?

As superconducting qubit devices increase in coherence and size, the idea of quantum computing becomes more tangible. However, this also raises new issues as standard methods for characterizing small scale devices are insufficient. Randomized benchmarking and tomographic techniques are essential tools for verifying one- and two-qubit gates, but have limited practicality in large systems. Techniques that ignore neighboring qubits are blind to potential crosstalk terms and miss important errors that could be present in multi-qubit quantum circuits and, crucially, could impact error correction codes. This talk will discuss a variety of methods for verifying and validating multi-qubit devices, with a particular emphasis on crosstalk errors. State-of-the-art devices have single-qubit error rates lower than 10^-3 and two-qubit error rates less than 10^-2, but these error rates do not account for total errors found in algorithmic applications. Such errors can be studied using extensions to randomized benchmarking, small logical codes, and targeted error amplification sequences. Each of these techniques is designed to either identify an error, characterize that error, or to verify and validate operations after developing new calibration procedures. These steps combined make up our approach to improving quantum controls on superconducting qubit devices as we continually detect new errors and improve fidelity.