Fast and Accurate Multilayer Polarizable Embedding Strategies for the Static and Dynamic Modeling of Complex Systems

The computational modeling of molecules embedded in complex (bio)matrices is extremely challenging due to the large dimension of these systems and the complex interactions between the various parts. An effective strategy is resorting to the combination of hybrid quantum/classical models in combination with the use of molecular dynamics techniques, such that the interesting part is described with high accuracy, whereas the rest of the environment is described in a cheap and affordable way. Despite the large diffusion of quantum/classical models, their application to larger and larger systems is still challenging. On one side their implementation often relies on quadratically scaling codes in the number of classical atoms, on the other side the model themselves present intrinsic limitations. In this work, we improved existing quantum/classical implementations both based on molecular mechanics and on polarizable continuum models, such that they can be readily applied to very large systems thanks to a computational cost linear scaling in the classical atoms. Finally, we combined an atomistic model and a polarizable continuum model in a three layer quantum/atomistic/continuum description which takes advantage of the strengths of the two kind of models.