Skip to main content

WVU researchers team up with AI in the search for advancements in quantum technology

Quantum materials such as giant magnets and superconductors may help in discovering new, faster technologies and energy-efficient electrical systems. 
Subhasish Mandal, WVU assistant professor in condensed matter physics, wears a plaid suit jacket, white button down dress shirt, dark plastic framed glasses. He has a dark, trimmed mustache and beard.

To further that search, West Virginia University’s Subhasish Mandal, an assistant professor in the Department of Physics and Astronomy, will be leading a group of students in developing an open-source, large-scale database for exploring artificial intelligence tools to categorize quantum materials.

The application of data science in the search for novel quantum materials has been limited by the capabilities of the existing materials databases that catalog them. Mandal’s work, which harnesses the combined power of data science and quantum theory to overcome these limitations and accelerate the discovery of new technology, is supported by a $297,000 grant from the National Science Foundation.

“Data science is a new paradigm in science, a new paradigm in physics, as well,” he said.

Quantum materials databases hold information related to materials that exhibit exotic quantum mechanical properties. These repositories play a crucial role in the field of materials science, providing researchers with access to information for designing and discovering novel materials with unique quantum characteristics that could be useful for real-world applications such as cyber communication, national security and electricity transportation.

“If we want to use data science, we need to have a lot of data,” Mandal said. To obtain them, researchers can run experiments and synthesize new materials in the lab, though this process is slow and expensive. Alternatively, they can use computations to predict how materials will behave and, in turn, use those predictions to build databases of material properties. This approach can be useful for exploring a wide range of materials quickly.

The standard computational model that scientists have been using for decades is a very popular method called Density Functional Theory, or DFT.

“Unfortunately, the traditional DFT doesn’t do a good job for many quantum materials, as it has significant approximations in it,” Mandal said. DFT’s simple approach may overlook some of the distinctive behaviors of electrons and, as a result, a researcher can miss the expectational and special behavior in quantum materials.

“Hence, the existing DFT databases are not very useful. If the data in the database are not very accurate, then your prediction from data science will be limited.”

By contrast, the database Mandal and his team are creating uses a theory called Dynamic Mean Field Theory that includes detailed quantum interactions.

“It goes far beyond the standard DFT,” he said, explaining that the DFT tools will be replaced by DMFT in this new database and will be used in the project’s second phase for machine learning to predict the properties of quantum materials and fix the systematic errors in the existing large-scale DFT databases.

“This is where we can learn something from a small-size, high-fidelity database and correct the existing large databases. We have big databases sitting around, but we know they are not very accurate, and we will have a database which will not be very large but still very accurate.”

The WVU researchers, including undergraduate and graduate students, will be working in collaboration with the National Institute of Standards and Technology and Rutgers University.

Real-world applications include the development of magnetic or solar cell materials, as well as high-temperature superconductors, which Mandal said have long been a dream for mankind.

“In the last 100 years, we have been using the trial-and-error method, and it’s very slow,” he noted. “Having a big database would enhance and accelerate progress on high-temperature superconductors.” He hopes these improvements will allow scientists to be better able to identify materials that could be useful for quantum technology.

Mandal intends to make the database and AI tools freely available for students and scientists to use and to train the next generation of data science researchers.

“We will be conducting free workshops for them across the campus.” he said. “Students will get multidisciplinary knowledge and career opportunities ranging from quantum theory to data and materials sciences. It has very good overlap between these two, the two big domains in physics at this moment — quantum theory and data science.”


lr/9/ /23

MEDIA CONTACT: Laura Roberts

Research Writer

WVU Research Communications


Call 1-855-WVU-NEWS for the latest West Virginia University news and information from WVUToday.

Follow @WVUToday on Twitter.