Efficient Construction of Excited-State Hessian Matrices with Machine Learning Accelerated Multilayer Energy-Based Fragment Method.

摘要: Recently, we have developed a multilayer energy-based fragment (MLEBF) method to describe excited states of large systems in which photochemically active and inert regions are separately treated with multiconfigurational single-reference electronic structure their mutual polarization effects naturally described within the many-body expansion framework. This MLEBF has been demonstrated provide highly accurate energies gradients. In this work, further derived excited-state Hessian matrices efficiently constructed. Moreover, combination recently proposed embedded atom neural network (EANN) model machine learning (ML) accelerated (i.e., ML-MLEBF) region is entirely replaced trained ML models. ML-MLEBF found improve computational efficiency particular for systems. Furthermore, both methods parallel exhibit low-scaling cost multiple CPUs. The present developments could motivate combining various techniques fragment-based explore Hessian-matrix-based properties