Machine learning solves puzzle of how electrons behave in quantum materials

27 Jun 2019

Prof Séamus Davis. Image: Tomás Tyner/UCC

Research led by an Irish scientist has revealed a scientific discovery ‘hiding in plain sight’.

Research from an international team of physicists and computer scientists published in Nature represents a significant breakthrough in our understanding of how electrons behave at an atomic level. This discovery could aid in the development of superconductors, which are super-efficient electricity conductors that could power a new generation of electronics.

Superconductivity is typically observed in certain materials when cooled below a characteristic critical temperature. High-temperature superconductors, however, are materials in which the electrons pair up and travel without friction and loss of energy at relatively high temperatures. In these materials, electrons interact with such force that their behaviour has proven difficult to define.

Quantum materials such as high-temperature superconductors will be needed for the future development of quantum technologies and, for the past two years, a team of scientists led by Prof Séamus Davis has applied machine learning to quantum physics to try and solve this conundrum.

Based on a suite of 80 artificial neural networks that they had designed and trained to recognise different forms of electronic matter, machine learning helped them to discover a new state of quantum matter called a vestigial nematic state (VNS).

With Davis at the helm, Prof Eun-Ah Kim from Cornell University leading the theoretical physicists and Prof E Khatami from San Jose State University leading the computer scientists, the researchers fed an archive of electronic quantum matter images gathered over about 20 years into these artificial neural networks.

“To my amazement, it actually worked!” said Davis, professor of physics at University College Cork and Oxford University.

“The vestigial nematic state had been predicted by theorists to exist for strongly interacting electrons in a disordered environment, but there was no experimental evidence. It was thrilling to see how the new machine-learning technique discovered it hiding in plain sight.”

Accelerating discovery with machine learning

This discovery may also help to rapidly accelerate the process of making further scientific breakthroughs.

In the data age, vast amounts of information require new techniques for analysis. Even experimental datasets from quantum materials research have grown to vast numbers due to automation. It is practically impossible for human researchers to examine all of the data that has accumulated, driving the need for pioneering physicists to find new techniques to push forward with discoveries.

By combining machine learning with quantum matter visualisation, these scientists demonstrated how extensive electronic image archives can be analysed efficiently, accurately and successfully.

It’s possible this new approach can be applied in other quantum materials research, especially in the area of high-temperature superconductivity and the quest for room-temperature quantum computers.

Elaine Burke is the host of For Tech’s Sake, a co-production from Silicon Republic and The HeadStuff Podcast Network. She was previously the editor of Silicon Republic.