Classic machine learning techniques make useful predictions about quantum materials – ScienceDaily

There has been a lot of hype around quantum computers, and for good reason. Future computers are designed to mimic what happens in nature at microscopic scales, meaning they have the potential to better understand the quantum world and speed up the discovery of new materials, including environmentally friendly drugs, chemicals and more. However, experts say viable quantum computers are still a decade or more away. What should researchers do in the meantime?

A new study led by Caltech is in the journal Sciences It describes how machine learning tools, running on classical computers, can be used to make predictions about quantum systems, thus helping researchers solve some of the toughest problems in physics and chemistry. While this idea has been demonstrated experimentally before, the new report is the first to mathematically prove that the method works.

says lead author Hsin-Yuan (Robert) Huang, a graduate student working with John Preskill, Richard P. Feynman Professor of Theoretical Physics and Allen VC Davis and Lenabelle Davis chair of leadership at the Institute of Quantum Science and Technology (IQIM). “But we’re not there yet and we were surprised to learn that classic machine learning methods can be used today. Ultimately, this paper looks at what humans can learn about the physical world.”

On microscopic levels, the physical world becomes an incredibly complex place governed by the laws of quantum physics. In this field, particles can exist in a superposition of states, or in two states simultaneously. And the superposition of states can lead to entanglement, a phenomenon in which particles are linked, or interconnected, without coming into contact with each other. These strange states and connections common in natural and synthetic materials are difficult to describe mathematically.

“Predicting the low-energy state of matter is very difficult,” Huang says. “There are huge numbers of atoms, and they are superimposed and entangled. You cannot write an equation to describe them all.”

The new study is the first mathematical proof 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 simulates the human brain to learn from data.

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

While previous studies have shown that applications of machine learning have the potential to solve some quantum problems, these approaches typically work in ways that make it difficult for researchers to know how the machines arrived at their solutions.

“Usually, when it comes to machine learning, you don’t know how the machine managed to solve the problem,” Huang says. “It’s a black box.” “But now we’ve basically figured out what’s going on in the box through numerical simulations.” Huang and colleagues conducted extensive numerical simulations in collaboration with the AWS Center for Quantum Computing at Caltech, which corroborated their theoretical results.

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

“The concern was that people who create new quantum states in the lab might not be able to understand them,” Preskill explains. “But now we can have plausible classical data to explain what’s going on. Classical machines not only give us an answer like an argument, they direct us toward a deeper understanding.”

Victor F. agrees. Albert, a physicist at the National Institute of Standards and Technology (NIST) and a former postdoctoral researcher at Caltech, shares this view. “The most exciting part for me about this work is that we are now closer to a tool that helps you understand the basic stage of a quantum state without requiring you to know much about that state beforehand.”

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

“We’re still at the beginning of this field,” Huang says. “But we know that quantum machine learning will ultimately be the most efficient.”