"Virtual dimensions in the docking method based on tensor train decompositions"
Zheltkov D.A., Tyrtyshnikov E.E.

A modification of the docking method based on tensor train decompositions using the idea of virtual dimensions adding is proposed to find the position of the energy global minimum for the ligand-protein system. The proposed method is compared with the TTDock (docking based on tensor train decompositions) program, which uses only physical dimensions. According to the testing results, the modified method is 5–10 times faster with the same level of reliability.

Keywords: tensor train decomposition, virtual dimensions, cross interpolation method, global optimization, docking, computer drug design

Zheltkov D.A., e-mail: dmitry.zheltkov@gmail.com – Moscow State University, Faculty of Computational Mathematics and Cybernetics; Leninskiye Gory 1-52, Moscow, 119991, Russia;
Tyrtyshnikov E.E., e-mail: eugene.tyrtyshnikov@gmail.com – Institute of Numerical Mathematics, Russian Academy of Sciences; ulitsa Gubkina 8, Moscow, 119333, Russia