Sandia National Laboratories and Brown University researchers have developed a new method to accelerate computer simulations, significantly speeding up research across various scientific fields. It was recently published in the journal npj Computational Materials and presents a universal approach to enhancing the performance of virtually any type of simulation — from researching drugs to sending rockets to Mars.
“From a user standpoint, there’s no difference between running your simulation tool or running this accelerated simulation tool. It gives you exactly the same predictions. The difference is how much time it takes to get those results,” said Rémi Dingreville, a Sandia researcher.
Sandia National Laboratories scientist Rémi Dingreville has co-developed a way to speed up virtually any kind of computer simulation, a ubiquitous research tool across science and industry. (Photo by Craig Fritz. Sandia National Laboratory )
The team demonstrated their accelerator by conducting materials science simulations 16 times faster than conventional methods. They also highlighted the potential application of this tool in diverse areas such as climate change research, autonomous vehicle navigation, and hardware acceleration.
“The potential to generalize our approach to different systems could lead to more efficient and sustainable technologies,” said Vivek Oommen of Brown University and the paper’s lead author.
This research was funded by Sandia’s Laboratory Directed Research and Development program and received support from the Center for Integrated Nanotechnologies, a Department of Energy Office of Science user facility, and Brown’s Center for Computation and Visualization.
Broader impacts across science
Dingreville, whose childhood passion for speed extended into his career as a scientist, engineered a previous simulation to run 40,000 times faster. Although the 16-fold speed increase might seem modest, Dingreville and the team believe their latest innovation could have a far-reaching impact because it is not restricted to specific problem types like other accelerators.
Recently, Dingreville highlighted in npj Computation Materials the broad applicability and potential impact of this hybrid approach for accelerating various scientific simulations. Potential applications span materials science, climate change research, autonomous vehicle navigation, and hardware acceleration. The scientists demonstrated a 16-fold speed increase in materials science simulations compared to conventional methods. Ultimately, the the approach is not restricted to specific problem types like other accelerators. This universality allows for its potential application in diverse fields, potentially leading to more efficient and sustainable technologies.
“Physics, chemistry, geochemistry, weather prediction — it really doesn’t matter,” said Dingreville, emphasizing the broad applicability of their approach.
The research team hopes this breakthrough will encourage scientists to rethink how they design and use simulations, potentially leading to more efficient research methodologies across various domains.
“I’m deeply fascinated by the challenges and potentials of integrating traditional numerical methods with artificial intelligence to solve complex problems in materials science,” Oommen added.
Enabling new research opportunities
Beyond saving time and money, the accelerator could make it feasible to study slow-developing phenomena that are typically too time-consuming to model. For example, simulating glacial melting processes often requires significant computational resources and time.
“The current state of the art is that you have to use these direct numerical solvers. Even though they are accurate, they’re slow,” Dingreville explained.
Looking ahead, the researchers are eager to see their methodologies applied to a broader range of challenges in fields such as energy, biotechnology, and environmental science.
“I’d love to see this applied in geoscience,” Dingreville added.
