Friday, 5 October 2018

Cheap nonadiabatic dynamics simulations

What is the cheapest way to run nonadiabatic dynamics simulations and get results that are at least better than a random number generator? How about parameterising a linear vibronic coupling Hamiltonian using only a single excited-state computation and running surface hopping dynamics with it. This is what we tried in our new paper "Highly efficient surface hopping dynamics using a linear vibronic coupling model" that just appeared in PCCP. And to our surprise, the results were actually a lot better than a random number generator. We could reproduce the main physics of the dynamics of intersystem crossing in SO2, the presence/absence of ultrafast internal conversion in adenine/2-aminopurine, as well as ultrafast intersystem crossing in 2-thiocytosine. Only for 5-azacytosine we were somewhat off the mark.

Some of the referees were a little bit "not amused" because it almost seems like kind of an unfair trick to run dynamics using such a simple setup. But if it works and if it gives you relevant information about the real world, why should you not do it?

No comments: