QuICS Members and Alumni Maintain Strong Presence at Premier Quantum Conference

Jan 12, 2024

Research by fellows, postdoctoral researchers, graduate students and alumni of the Joint Center for Quantum Information and Computer Science (QuICS) will once again be featured in talks at the annual conference on Quantum Information Processing (QIP), held this year from Jan. 13–19, 2024, in Taipei, Taiwan.

Five papers by current QuICS members were accepted for talks and one will be featured in an invited plenary talk—a mark that continues a strong history of QuICS representation at the premier conference dedicated to quantum computer science. Other talks feature research by QuICS alumni, with many current and former QuICS members also contributing their work to the conference’s poster sessions.

QuICS Hartree Postdoctoral Fellow Dong An, QuICS alumnus Jin-Peng Liu, and a colleague will discuss a new technique they developed to simulate a broad class of differential equations on quantum devices. Their method, which expresses the solution of a differential equation as a linear combination of Hamiltonian simulation (LCHS) problems, provides a simpler, more efficient way to solve these problems than previous approaches. It also relies on less mathematical machinery to implement. The authors say that the approach is well suited to quantum and classical devices working in tandem and that it requires only a small amount of quantum resources—a fact that makes it a good target algorithm for emerging small quantum devices.

In a pair of papers accepted for a single merged talk, QuICS Postdoctoral Scholar James Watson and colleagues focus on how to efficiently learn the properties of a quantum state—including the expectation values of measurements made on the state—and parlay that knowledge into learning the characteristics of quantum phases of matter.

In the first paper, the researchers developed a new technique for learning the properties of a particular class of parameterized quantum states and the expectation values of their observables, finding an exponential improvement in terms of how many copies of the state are needed to reach a given accuracy. Based on these results, they developed algorithms for learning the values of observables not just of a single state but of an entire quantum phase of matter. Here, too, their new approach yielded an exponential improvement over the best prior technique.

In the second paper, Watson and his colleagues extended some of these results to a less restricted family of states. They prove that with a relatively small amount of information about the states and observables in a quantum phase of matter, it’s possible to use classical machine learning techniques to learn expectation values of observables within the entire phase of matter—a result that applies to a richer variety of quantum phases than prior work while still maintaining a similar efficiency.

QuICS Fellow Daniel Gottesman, who is also the Brin Family Professor in Theoretical Computer Science at the University of Maryland, and three colleagues will present their work on deepening the fundamental understanding of approximate quantum error correction (AQEC). By definition, AQEC codes don’t correct errors perfectly. But how far can they wander from perfection before they lose any useful notion of “correcting” errors at all? The authors propose that there is a dividing line that separates good AQEC codes from the rest and demonstrate that the boundary is directly tied to the amount of entanglement present in the AQEC codewords. Intriguingly, when treating AQEC codes as physical models the same dividing line corresponds to changes in physical behavior, such as phase transitions in many-body systems—a fact that reinforces this characterization of AQEC codes, the authors say.

QuICS Fellow Michael Gullans, along with QuICS alumni Bill Fefferman and Kunal Sharma, together with colleagues, will present a result that explores how real-world noise, which resembles the noise found in actual quantum devices, affects the outputs of random quantum circuits. The study shows that random circuits subjected to realistic noise produce outputs that behave differently than noiseless circuits or those with simpler noise. In particular, realistic noise prevents the output (the quantum state that results from the application of a random circuit) from spreading out too much, meaning that it won’t end up resembling a uniform distribution. Instead, the output remains more bunched up, suggesting that it would remain hard to simulate on non-quantum devices. The result has implications for hardness (and easiness) proofs that rely on the tendency of the output of a random circuit to bunch up or spread out and could be an important tool in continuing to understand the effects of noise in real quantum devices.

Finally, a colleague of Gullans and QuICS Hartree Postdoctoral Fellow Dominik Hangleiter will deliver an invited plenary talk about a prototype quantum computer made from arrays of individual atoms that the researchers, together with other collaborators, reported recently in the journal Nature.

—Story by Chris Cesare, Joint Quantum Institute communications group

A list of the papers with a full author list for each is provided below.

• “Linear combination of Hamiltonian simulation for nonunitary dynamics with optimal state preparation cost” by Dong An (QuICS/UMD), Jin-Peng Liu, and Lin Lin
• “Efficient learning of ground & thermal states within phases of matter” by Emilio Onorati, Cambyse Rouzé, Daniel Stilck França, and James D. Watson (QuICS/UMD)
• “Provably Efficient Learning of Phases of Matter via Dissipative Evolutions” by Emilio Onorati, Cambyse Rouzé, Daniel Stilck França, and James D. Watson (QuICS/UMD)
• “Complexity and order in approximate quantum error-correcting codes” by Jinmin Yi, Weicheng Ye, Daniel Gottesman (QuICS/UMD), and Zi-Wen Liu
• “Effect of non-unital noise on random circuit sampling” by Bill Fefferman, Soumik Ghosh, Michael Gullans (QuICS/NIST), Kohdai Kuroiwa, and Kunal Sharma
• “Logical quantum processor based on reconfigurable atom arrays” by Dolev Bluvstein, Simon J. Evered, Alexandra A. Geim, Sophie H. Li, Hengyun Zhou, Tom Manovitz, Sepehr Ebadi, Madelyn Cain, Marcin Kalinowski, Dominik Hangleiter (QuICS/UMD), J. Pablo Bonilla Ataides, Nishad Maskara, Iris Cong, Xun Gao, Pedro Sales Rodriguez, Thomas Karolyshyn, Giulia Semeghini, Michael J. Gullans (QuICS/NIST), Markus Greiner, Vladan Vuletić, and Mikhail D. Lukin