Nov 20 2019
The electronic characteristics of molecules and the molecular wave functions can be predicted using artificial intelligence (AI).
This novel AI technique was created by a research team from the University of Warwick, the University of Luxembourg, and the Technical University of Berlin. The method could be used to expedite the design of novel materials or drug molecules.
Machine learning algorithms and AI are often used for predicting customers’ purchasing behavior and to identify their handwriting or faces. In scientific studies, AI is turning out to be an important tool for scientific innovation.
In the field of chemistry, AI has become an essential tool in estimating the experimental results or replications of quantum systems. In order to realize this, AI should be able to methodically include the basic laws of physics.
An interdisciplinary research team, which included computer scientists, physicists, and chemists headed by the University of Warwick, the University of Luxembourg, and the Technical University of Berlin, has now created a deep machine learning algorithm. This algorithm is capable of predicting the molecules’ quantum states, or the so-called wave functions, which govern all the characteristics of the molecules.
This task was achieved by the AI that learned to solve important equations of quantum mechanics as described in the study titled “Unifying machine learning and quantum chemistry with a deep neural network for molecular wavefunctions,” which has been published in Nature Communications.
To solve these fundamental equations of quantum mechanics in the traditional method, enormous amounts of high-performance computing resources (months of computing time) are required. It is this requirement that usually limits the computational design of the latest purpose-built molecules meant for industrial and medical applications.
Now, the recently developed AI algorithm can provide precise predictions on a mobile phone or laptop within a matter of seconds.
This has been a joint three-year effort, which required computer science know-how to develop an artificial intelligence algorithm flexible enough to capture the shape and behaviour of wave functions, but also chemistry and physics know-how to process and represent quantum chemical data in a form that is manageable for the algorithm.
Dr Reinhard Maurer, Department of Chemistry, University of Warwick
During an interdisciplinary three-month fellowship program conducted at IPAM (UCLA), the researchers collectively explored the topic of machine learning in quantum physics.
This interdisciplinary work is an important progress as it shows that, AI methods can efficiently perform the most difficult aspects of quantum molecular simulations. Within the next few years, AI methods will establish themselves as essential part of the discovery process in computational chemistry and molecular physics.
Dr Klaus Robert-Muller, Professor, Institute of Software Engineering and Theoretical Computer Science, Technical University of Berlin
“This work enables a new level of compound design where both electronic and structural properties of a molecule can be tuned simultaneously to achieve desired application criteria,” concluded Professor Dr Alexandre Tkatchenko from the Department of Physics and Materials Research at the University of Luxembourg.