It is six orders of magnitude faster than the quantum chemical method. An adiabatic artificial neural network method based on adiabatic state can accelerate the simulation of dual nitrogen benzene der

Light induced chemical processes are ubiquitous in nature , And has a wide range of technical applications . for example , Photoisomerization can make drugs with light switchable scaffolds activated by light . In principle, , Have the required photophysical properties （ Such as high isomerization quantum yield ） Optical switch of , It can be identified by virtual screening of reaction simulation .

But in practice , These simulations are rarely used for screening , Because they require hundreds of trajectories and expensive quantum chemical methods to explain the diabatic excited state effect .

ad locum , Researchers from Harvard University and Massachusetts Institute of Technology , An adiabatic artificial neural network based on adiabatic state is developed （DANN）, It is used to accelerate the simulation of dual azobenzene derivatives and such molecules . The network is six orders of magnitude faster than the quantum chemical method used for training .DANN Azobenzene molecules that can be transferred outside the training set , Predict the quantum yield of the unknown species related to the experiment .

Researchers use this model to virtually screen 3100 A hypothetical molecule , And identify 「 A new species 」. Confirm the model prediction using high-precision diabatic dynamics . The results pave the way for rapid and accurate virtual screening of photoactive compounds .

The research 「 Excited state non-adiabatic dynamics of large photoswitchable molecules using a chemically transferable machine learning potential」 entitled , On 2022 year 6 month 15 Published on 《 Nature Communications》.

Thesis link ： https://www.nature.com/articles/s41467-022-30999-w