There have been many advances in technology over the past decade with and much of that is down to artificial intelligence (AI). From autonomous cars to smart search engines, AI is everywhere, and now researchers are using it to accelerate infrared spectra too.
Infrared spectroscopy is a fantastic way to learn all about the world of molecules. Infrared spectra are the individual chemical fingerprints that contain all the data on the composition and properties of the substance or material, and they’re often very complex. Precise predictions of infrared spectra can be enabled through the use quantum chemical reactions but are still no easy task. Because of the high computational effort that’s associated with them, infrared spectra can only really be relied upon for small chemical systems.
Philipp Marquetand from the Faculty of Chemistry at the University of Vienna and a group of international researchers have discovered how to accelerate this simulation process through the use of AI. To do this, they used artificial neural networks in which to model the human brain. These networks are what learn the complex quantum mechanical relationships needed to model the infrared spectra using a minimum number of examples.
Doing it this way the researchers can carry out simulations in a matter of minutes opposed to the thousands of years it would take even a supercomputer to do the same job without compromising on reliability. “We can now finally simulate chemical problems that could not be overcome with the simulation techniques used up to now,” says the first author of the study, Michael Gastegger. Moving forward the researchers are confident this new method of predicting spectra will be used to analyze experimental infrared spectra in the near future.
More News to Read