Learning the Capabilities of Quantum Computers

Quantum computers can now run interesting programs, but each processor’s capability—the set of programs that it can run successfully—is limited by hardware errors. These errors can be complicated, making it difficult to accurately predict a processor’s capability. Benchmarks can be used to measure capability directly, but current benchmarks have limited flexibility and scale poorly to many-qubit processors. Furthermore, predicting how other, untested circuits will run from benchmarking data is exceedingly challenging.  In this talk, I will show how to construct scalable, efficiently verifiable benchmarks based on any program by using a technique that we call circuit mirroring. I will show how this technique can be used to measure the fidelity with which a given circuit can be implemented, providing a solution to a long-standing verification problem in quantum computation. I will present the results of experiments in which we used these benchmarks to map out the capabilities of twelve publicly available processors. Finally, I will then present results of on-going research into how to apply machine learning techniques to learn a quantum computer’s capability from benchmarking data—by using neural networks to predict how well any circuit will run on the tested quantum computer.

This talk is based on Nature Physics 18 75-79 (2022) [https://www.nature.com/articles/s41567-021-01409-7], arXiv:2204.07568 (2022) [https://arxiv.org/abs/2204.07568], and arXiv:2112.09853 (2021) [https://arxiv.org/abs/2112.09853].