Russian scientists have found contradictions in the most popular method of theoretical chemistry – TASS

MOSCOW. 5 Jan. /TASS/. Scientists from the Institute of Organoelement compounds (INEOS) Russian Academy of Sciences together with an American colleague published in the journal Science evidence that the most common method of theoretical chemistry – theory of functional of the electronic density – were going in the wrong direction the last 10 years.

As told TASS lead author Ivan Bushmarinov, the method is widely used since the late 1970-ies and has long gone beyond the works of the theorists, “roughly speaking, that quantum chemistry “on the knee”. “The theory is that electron density – the distribution of electrons in the system – defines the system properties, and the energy of the system written as a functional of this density is a minimum in the ground state. The problem is that to accurately record such functionality is possible only in very rare cases (for example, for a gas of free electrons),” says Bushmarinov.

All the functionals used for the calculation of real systems, are a result of certain approximations – some of them are based on theory, and some of them are obtained solely by numerical methods, parameter fitting. “In 2004 there was a turning point: the fascination of fitting parameters of model systems has started to cause (popular new) functionals has worked for them and only for them, they showed an artificially low error in energy and error in the electron density began to grow… (That is) in the last 10 years, the methods of the density functional theory was moving in the wrong direction,” emphasizes Bushmarinov.

Calculations “Lomonosov”

to prove this hypothesis, the first author, a graduate student, LRSI INEOS Mikhail Medvedev tested on the Russian supercomputer “Lomonosov” almost all density functionals, and there are more than 100. From the first idea until the completion of the work took about six months.

“In this sense, work cannot be considered classically theoretical, today such a large-scale quantum-chemical calculations can be considered an experiment. It was two months of supercomputer time. Thus, we were able to come to judgment about a few functional or systems, and the trends of development of the density functional theory in General,” comments Bushmarinov.

“the results Obtained were in contradiction with the principle known to any programmer: you can’t use the algorithm to obtain the correct result from the wrong data,” said Medvedev, whose words are given in the press release. The third member of the Russian research team was Konstantin Lysenko from laboratory x-ray diffraction analysis INEOS.

Tip of the wizard

given the results of the calculations on the “Lomonosov” and formulated the hypothesis of Russian scientists asked for help from one of the most prominent world experts in this field – Professor John Purdue from temple University (USA, Philadelphia).

He pointed to a mistake originally made by the authors: they used the anions (negatively charged ions) as a test system, and this cannot be done. After this fix, everything fell into place.

Perdue, one of the most cited physicists in the world, has become an official co-author of the study, which was successfully published in Science in late 2016. “Such a serious journal would not send this work to the reviewers, if the contributors were professors from Purdue. It is not Russophobia, it’s just really unexpected that three of crystallography challenged the existing theoretical chemistry system. On the other hand, if we hadn’t invited an American Professor to join our work, we simply would have nothing to send to the journal: that he helped resolve several significant errors at the stage selection model,” explains Bushmarinov.