Machine learning paces up Atomistic Simulations of Ice and Water

Machine learning paces up Atomistic Simulations of Ice and Water

A collaborative study between researchers at the University of Vienna and École Polytechnique Fédérale de Lausanne at the University of Göttingen has revealed answers for questions pertaining to water. The questions are related to water being densest at 4 degree Celsius, reason for ice to float, reason for heavy water having a different melting point than normal water, and so on. The findings for these questions have been published in the American journal Proceedings of the National Academy of Sciences. Researchers have obtained findings for water phenomenon by combining quantum mechanics and data-driven machine learning techniques.

Common knowledge says, electrons and nuclei are the building blocks of most matter. The behavior of most matter can be described with respect to their wave function, based on the laws of quantum mechanics. Schrodinger equation allows to make models and projections of any material, which includes water. However, the model does not hold true every time.

Quantum Mechanical Solutions unaffordable for small Systems

With increase in number of electrons and nuclei, the complexity becomes intractable even with the fastest supercomputers. This is so even after a century of success in optimizing of such calculations. As a matter of fact, quantum mechanical calculations are uneconomical for systems that have over a few hundred atoms, or for a time period over a nanosecond.

To address these harsh limitations, researchers leveraged an artificial neural network to comprehend atomic interactions from quantum mechanics. Theoretically, an artificial neural network can be represented as several layers of interconnected nodes. The network mimics the structure of neurons in the human brain.

Leave a Reply

Your email address will not be published. Required fields are marked *