Every day there seems to be a new example of machine learning (ML) or deep learning (DL) speeding, simplifying, or improving what was previously an intractable or at least computationally-expensive simulation model. Recently, researchers from Argonne National Laboratory Center for Nanoscale Materials (CNM) used ML developed a new, computationally inexpensive water model that more accurately represents the thermodynamic properties of water, including how water changes to ice at the molecular scale.

Demystifying how water behaves – despite its deceptively simple structure – has long been a challenge. Today there are roughly 50 different models according to ANL researcher Subramanian Sankaranarayanan, the corresponding author on a recent Nature paper reporting the work. There is also an article on the study posted on the ANL site.

Trying to create quantum mechanical or atomistic models to capture water’s behavior had flummoxed researchers because they are so computationally intensive and still fail to reproduce many temperature-dependent properties of water. the researchers, the choice to use entire water molecules as the fundamental unit in the model allowed them to perform the simulation at low computational cost.

To achieve the high accuracy of the coarse-grained model, the researchers trained the model using information drawn from nearly a billion atomic-scale configurations involving temperature-dependent properties that are well known. ​“Essentially, we said to our model, ​‘look, this is what the properties are,’ and asked it to give us parameters that were able to reproduce them,” said Henry Chan, Argonne postdoctoral researcher and the lead author of the study.

Training the model involved what Chan called a ​“hierarchical approach,” in which each candidate model was put through a series of tests or evaluations, starting with basic essential properties before working its way up to more complex ones. ​“You can think of it like trying to teach a child a skill,” Chan said. ​“You start with something fundamental and work your way up once you see progress.”

The team used the Mira supercomputer at the Argonne Leadership Computing Facility, to perform simulations of up to 8 million water molecules to study the growth and formation of interfaces in polycrystalline ice (link to an ANL video of the simulation).

Here their paper’s abstract:

“An accurate and computationally efficient molecular level description of mesoscopic behavior of ice-water systems remains a major challenge. Here, we introduce a set of machine-learned coarse-grained (CG) models (ML-BOP, ML-BOPdih, and ML-mW) that accurately describe the structure and thermodynamic anomalies of both water and ice at mesoscopic scales, all at two orders of magnitude cheaper computational cost than existing atomistic models. In a significant departure from conventional force-field fitting, we use a multilevel evolutionary strategy that trains CG models against not just energetics from first-principles and experiments but also temperature-dependent properties inferred from on-the-fly molecular dynamics (~ 10’s of milliseconds of overall trajectories). Our ML BOP models predict both the correct experimental melting point of ice and the temperature of maximum density of liquid water that remained elusive to-date. Our ML workflow navigates efficiently through the high-dimensional parameter space to even improve upon existing high-quality CG models (e.g. mW model).”

“The beauty is that this molecular model has no right to be as accurate as the atomistic models, but still ends up being so,” says Mathew Cherukara, CNM assistant scientist and co-first author.

Link to Argonne National Laboratory article: https://www.anl.gov/article/through-machine-learning-new-model-holds-water

Link to Nature paper (Machine learning coarse grained models for water): https://www.nature.com/articles/s41467-018-08222-6

Feature Image Caption & Source:
Molecular dynamics simulations based on machine learning show how grains of ice form and coalesce in supercooled water, which results in ice with imperfections. These simulations help scientists learn about the movement of the boundary between ice grains (yellow/green/cyan) and the stacking disorder that occurs when hexagonal (orange) and cubic (blue) pieces of ice freeze together. This information is important in applications such as climate modeling and cryogenics. Researchers performed these simulations on Mira at the Argonne Leadership Computing Facility and Carbon at the Center for Nanoscale Materials; ALCF and CNM are both DOE Office of Science User Facilities. (Image by Argonne National Laboratory.)