Physicists are using artificial intelligence (AI) to solve challenging and massive quantum physics problems that previously required 100,000 equations, without loss of accuracy, in a manageable effort of just four equations. reduced to
Specifically, scientists managed to capture the movement of electronic Traverse square grids in simulations that previously required hundreds of thousands of individual equations.
results published in physical review letter, could change the way scientists analyze systems with multiple interacting electrons. Moreover, if the technique can be adapted to other situations, it could help create materials with desirable properties, such as superconductivity and utility in clean energy generation, according to a Simons Foundation press release.
This computational feat could also help solve one of the most difficult problems in quantum physics: “many electrons” attempts to describe a system containing a large number of interacting electrons.
As such, the model also serves as a testing ground before applying new approaches to more sophisticated quantum systems.
Using AI, CCQ researchers and colleagues compressed a difficult quantum problem requiring 100,000 equations into a bite-sized task of just four equations without sacrificing accuracy. read more: https://t.co/hI48r7DLr3
— Flatiron CCQ (@FlatironCCQ) September 26, 2022
“We start with this gigantic object of all these coupled differential equations. Then we use machine learning to transform it into something small enough to be counted on your fingers,” said the study’s lead author. said. Domenico Disante Visiting Scholar at the Center for Computational Quantum Physics (CCQ) at the Flatiron Laboratory, New York, and Assistant Professor at the University of Bologna (Italy).
Electrons entangled at the quantum level
The problem to be addressed refers to the behavior of electrons as they move through them. grid network. When two electrons occupy the same lattice site, they interact. This arrangement, known as the Hubbard model, is an idealization of several important classes of materials that allow scientists to understand how the behavior of electrons leads to the phases of interest. , according to the statement.
This is, for example, a theoretical model of the state, Superconductivity electrons flow through the material without resistance.
The problem is that the electrons in question are intermixed at the quantum level and above all cannot be treated individually. A few entangled electrons end up with hundreds of thousands of interrelated equations, each describing an individual pair, increasing the computational complexity exponentially. However, the AI system’s neural network applied to it succeeded in solving this gigantic system of equations.
AI can detect hidden patterns
Training the algorithm was computationally intensive and took weeks. However, the systems that exist today are at a level where they can be adapted to other complex problems without starting from scratch.
“It’s basically a machine that can detect hidden patterns.” Di Sante explained. “When I saw the results, I said, ‘Wow, this exceeded my expectations.’ We really captured the physics involved.”
The biggest unknown now is whether the new method will work in more complex quantum systems, such as matter, where electrons interact over long distances. In addition, Di Santé sees interesting potential for applying artificial intelligence to other fields that deal with renormalization clusters, such as cosmology and neurology.