Multiscale modeling computational algorithms

Columns (challenges to achieving discoveries) combinatorial search algorithms, data assimilation, multiscale modeling, structure-function and structure-activity relationships, large-scale computation, how to create interdisciplinary teams as well as support the critical need for disciplinary research... 

At least two recent trends within computational chemistry, depending on the increasing computational strength and on algorithm development, can be identified first, the exploration of the domain between electronic structure calculations and molecular level simulations using methods such as QM/MM, Car-Parrinello, reactive force fields, etc. second, multiscale modelling, where results from complex level calculations are used as input in more macroscopic approaches in a coupled model. ... 

Molecular modeling for nonequilibrium processes, such as heterogeneous catalysis and growth of advanced materials, is a relatively new tool in the chemical process industry. The advent of algorithms and advances in computer power on the one hand, and the development of coarse graining and multiscale... 

One of the issues in dynamical multiscale coupling is the tailoring of the time step to the different subdomains. If the same time step is used in both the atomistic and the continuum regions, computations will be wasted in the continuum model. However, if in the hand-shake region the size of the FEM elements is reduced to coincide with the individual atoms, it is difficult to tailor the time step. Therefore, the authors of the ODD method chose to use a uniform mesh for the continuum domain, so that a much larger time step could be used in the continuum model than in the atomistic one. A description of such a multiple-time-step algorithm is provided in the paper. 

---