Researchers have discovered that the original way to speed up chemical processes is to reboot
A new study by Tel Aviv University reveals that "the legality" that is true for the world of computers is probably also true for the world of chemistry. The researchers discovered that in order to speed up samples of chemical simulations, all that needs to be done is to stop them and restart them.
overcome the simulation time limit
The research was conducted under the leadership of doctoral student Ofir Bloomer, in collaboration with Prof. Shlomi Reuvani and Dr. Barak Hirschberg from the school of chemistry at the Faculty of Exact Sciences by Raymond and Burley Sackler. The study was published in the prestigious journal Nature Communications.
The researchers explain that molecular dynamics simulations are like a virtual microscope. They follow the movement in time of each of the atoms in chemical, physical and biological systems such as proteins, liquids and crystals. They provide insights into a wide variety of processes, and are used in several technological applications, including the development of new drugs. However, the simulations are limited to times shorter than a millionth of a second, and therefore are unable to describe processes that occur more slowly, such as protein folding and crystal formation. This limitation is known as the time scaling problem, and is one of the biggest challenges in the field.
"In the new study, we showed that the limitation can be overcome by means of random initialization of the simulations (stochastic resetting)," explains doctoral student Ofir Blumer. "At first glance, this seems counterintuitive - how is it possible that the simulations will end faster if you start them over? But when you examine the issue in depth, it turns out that the answer lies in the fact that if we repeat the simulation experiment many times, the time it takes to finish will vary greatly. Sometimes it will end quickly, and sometimes it will get stuck In intermediate states it takes a long time. The initialization of the simulations prevents them from getting stuck in these intermediate states, and shortens the average time to finish the process."
In the study, the researchers combined the random initialization with metadynamics, a popular method for simulations of slow processes. The combination enabled greater acceleration than either method alone. Furthermore, metadynamics needs a lot of prior knowledge about the process to successfully speed up the simulations. The combination with random initialization greatly reduces this dependency, and saves chemists a lot of effort to run them. Finally, the researchers showed that the combination allows for a more accurate prediction of the rate of the slow processes. The combined method was successfully used to accelerate the sampling of protein folding in water, and in the future will allow to accelerate simulations of even larger systems.
More of the topic in Hayadan: