Classical machine learning techniques make useful predictions about quantum materials — ScienceDaily


Quantum computing has garnered a lot of attention, and with good reason. The futuristic computers are designed to mimic what happens in nature on a microscopic scale, meaning they are able to better understand the quantum realm and accelerate the discovery of new materials, including pharmaceuticals, eco-friendly chemicals and more . However, experts say viable quantum computers are still a decade or more away. What should the researchers do in the meantime?

A new Caltech-led study in the journal Science describes how machine learning tools running on classical computers can be used to make predictions about quantum systems, helping researchers solve some of the most challenging problems in physics and chemistry. While this notion has already been demonstrated experimentally, the new report is the first to show mathematically that the method works.

“Quantum computers are ideal for many types of physics and materials science problems,” says lead author Hsin-Yuan (Robert) Huang, a graduate student working with John Preskill, the Richard P. Feynman Professor of Theoretical Physics and the Allen VC Davis and Lenabelle Davis Executive Chair of the Institute for Quantum Science and Technology (IQIM). “But we haven’t got that far yet and to our surprise we found out that classic machine learning methods can now be used. Ultimately, this paper is about showing what humans can learn about the physical world.”

READ:  Joint Appointments to Bring New Expertise in Machine Learning, Algorithm Design, and Network Design to the Epstein Department - USC Viterbi

At the microscopic level, the physical world is becoming an incredibly complex place governed by the laws of quantum physics. In this region, particles can exist in a superposition of states or in two states at the same time. And a superposition of states can lead to entanglement, a phenomenon in which particles are connected or correlated without actually being in contact with each other. These strange states and compounds, widespread in natural and man-made materials, are very difficult to describe mathematically.

“Predicting the low-energy state of a material is very difficult,” says Huang. “There’s a huge number of atoms, and they’re superimposed and entangled. You can’t write an equation to describe everything.”

The new study is the first mathematical demonstration that classical machine learning can be used to bridge the gap between us and the quantum world. Machine learning is a type of computer application that mimics the human brain to learn from data.

READ:  Quantum Leap Cast: Where You've Seen The Stars Of The NBC Series Before

“We are classical beings living in a quantum world,” says Preskill. “Our brains and computers are classical, and that limits our ability to interact with and understand quantum reality.”

While previous studies have shown that machine learning applications are capable of solving some quantum problems, these methods typically work in ways that make it difficult for researchers to learn how the machines arrived at their solutions.

“Usually with machine learning, you don’t know how the machine solved the problem. It’s a black box,” says Huang. “But now we’ve essentially figured out what’s happening inside the box through our numerical simulations.” Huang and his colleagues, in collaboration with the AWS Center for Quantum Computing at Caltech, ran extensive numerical simulations that confirmed their theoretical results.

The new study will help scientists better understand and classify complex and exotic phases of quantum matter.

“The concern was that people who create new quantum states in the laboratory might not be able to understand them,” explains Preskill. “But now we can get reasonable classical data to explain what’s going on. The classic machines not only give us an answer like an oracle, but lead us to a deeper understanding.”

READ:  Danish initiative aims to develop first quantum computer for life sciences

Co-author Victor V. Albert, a National Institute of Standards and Technology (NIST) physicist and former DuBridge Prize postdoctoral fellow at Caltech, agrees. “The part that excites me the most about this work is that we are now closer to a tool that will help you understand the underlying phase of a quantum state without having to know much about that state in advance.”

Ultimately, future quantum-based machine learning tools will naturally outperform classical methods, the scientists say. In a related study appearing in June 10, 2022 ScienceHuang, Preskill and their collaborators report using Google’s Sycamore processor, a rudimentary quantum computer, to demonstrate that quantum machine learning is superior to classical approaches.

“We are still at the very beginning in this area,” says Huang. “But we know that quantum machine learning will ultimately be the most efficient.”



Source link