REGISTER HERE

In this final installment of the 2022 Next-Gen Quantum Investigators Colloquia series, Xun Gao, a Lukin Lab postdoc, suggests criteria to use in focusing attempts to demonstrate practical quantum advantage.

Abstract

This talk is motivated by the question: Why do we put so much effort and investment into quantum computing?  A short answer is that we expect the existence of practical quantum advantages. In this talk, I suggest three criteria for evaluating as well as designing a practical quantum advantage protocol: (1) a solid foundation; (2) applications to practical problems; and (3) near-term implementability.  I apply these criteria to study weaknesses in a  recently-published protocol and then go on to propose a new protocol exhibiting two of the three criteria.

Google, in 2019, published a paper in Nature claiming a finding of quantum supremacy for a protocol that sampled bitstrings from random quantum circuits. We used the first criterion to probe the foundation of Google’s experiment, showing that the Linear Cross-Entropy Benchmark - which Google had used to certify the quantum advantage – has several vulnerabilities.  In fact, we were able to construct a very efficient classical algorithm that produces a comparable performance on this benchmark within two seconds, using just a single GPU.

Next, I will present a new neural-sequence quantum model designed to be used for language translation. It will provide better expressive power than any reasonable conventional neural network. This protocol is based on a quantum contextuality that exhibits criteria (1) and (2).   Finally, I will present ongoing work based on the connection between neural networks and Rydberg atom-arrays.  We expect that these protocols will exhibit properties of all three criteria.