Skip to main content

Neural nets to simulate molecular motion cast

New work from Los Alamos National Laboratory, the University of North Carolina at Chapel Hill, and the University of Florida is showing that artificial neural nets can be trained to encode quantum mechanical laws to describe the motions of molecules, supercharging simulations potentially across a broad range of fields.
"This means we can now model materials and molecular dynamics billions of times faster compared to conventional quantum methods, while retaining the same level of accuracy," said Justin Smith, Los Alamos physicist and Metropolis Fellow in the laboratory's Theoretical Division. Understanding how molecules move is critical to tapping their potential value for drug development, protein simulations and reactive chemistry, for example, and both quantum mechanics and experimental (empirical) methods feed into the simulations.
The new technique, called the ANI-1ccx potential, promises to advance the capabilities of researchers in many fields and improve the accuracy of machine learning-based potentials in future studies of metal alloys and detonation physics.
Quantum mechanical (QM) algorithms, used on classical computers, can accurately describe the mechanical motions of a compound in its operational environment. But QM scales very poorly with varying molecular sizes, severely limiting the scope of possible simulations. Even a slight increase in molecular size within a simulation can dramatically increase the computational burden. So practitioners often resort to using empirical information, which describes the motion of atoms in terms of classical physics and Newton's Laws, enabling simulations that scale to billions of atoms or millions of chemical compounds.
Traditionally, empirical potentials have had to strike a tradeoff between accuracy and transferability. When the many parameters of the potential are finely tuned for one compound, the accuracy decreases on other compounds.
Instead, the Los Alamos team, with the University of North Carolina at Chapel Hill and University of Florida, has developed a machine learning approach called transfer learning that lets them build empirical potentials by learning from data collected about millions of other compounds. The new approach with the machine learning empirical potential can be applied to new molecules in milliseconds, enabling research into a far greater number of compounds over much longer timescales.
Story Source:
Materials provided by DOE/Los Alamos National Laboratory. 
Note: Content may be edited.

Comments

Post a Comment

Popular posts from this blog

Home births as safe as hospital births: International study suggests

A large international study led by McMaster University shows that low risk pregnant women who intend to give birth at home have no increased chance of the baby's perinatal or neonatal death compared to other low risk women who intend to give birth in a hospital. The results have been published by  The Lancet 's  EClinicalMedicine  journal. "More women in well-resourced countries are choosing birth at home, but concerns have persisted about their safety," said Eileen Hutton, professor emeritus of obstetrics and gynecology at McMaster, founding director of the McMaster Midwifery Research Centre and first author of the paper. "This research clearly demonstrates the risk is no different when the birth is intended to be at home or in hospital." The study examined the safety of place of birth by reporting on the risk of death at the time of birth or within the first four weeks, and found no clinically important or statistically different risk between home...
Post a Comment
Read more

Dark matter may be older than the Big Bang

Dark matter, which researchers believe make up about 80% of the universe's mass, is one of the most elusive mysteries in modern physics. What exactly it is and how it came to be is a mystery, but a new Johns Hopkins University study now suggests that dark matter may have existed before the Big Bang. The study, published August 7 in  Physical Review Letters , presents a new idea of how dark matter was born and how to identify it with astronomical observations. "The study revealed a new connection between particle physics and astronomy. If dark matter consists of new particles that were born before the Big Bang, they affect the way galaxies are distributed in the sky in a unique way. This connection may be used to reveal their identity and make conclusions about the times before the Big Bang too," says Tommi Tenkanen, a postdoctoral fellow in Physics and Astronomy at the Johns Hopkins University and the study's author. While not much is known about its origins,...
Post a Comment
Read more

Scientists challenge notion of binary sexuality with naming of new plant species

A collaborative team of scientists from the US and Australia has named a new plant species from the remote Outback. Bucknell University biology postdoctoral fellow Angela McDonnell and professor Chris Martine led the description of the plant that had confounded field biologists for decades because of the unusual fluidity of its flower form. The discovery, published in the open access journal  PhytoKeys , offers a powerful example of the diversity of sexual forms found among plants. The new species of bush tomato discovered in remote Australia provides a compelling example of the fact that sexuality among Earth's living creatures is far more diverse -- and interesting -- than many people likely realize. Bucknell University postdoctoral fellow Angela McDonnell and biology professor Chris Martine led the study following an expedition last year to relocate populations of the new plant, which were first noted by Australian botanists during the 1970s. Herbarium specimens from th...
Post a Comment
Read more