Timely problems ranging from the design of fuel cells to macromolecular dynamics, to the spread of epidemics and climate modelling, are characterized by a vast disparity of scales, ranging from the microscopic to the macroscopic. While microscopic simulation methods (Molecular Dynamics (MD) and Monte Carlo (MC) algorithms) can describe aspects of such complex systems, they are limited to short scales relative to device sizes and morphologies observed in experiments. In addition to this disparity in scales within the same model, additional challenges arise from "disparity in modelling": in phenomena with fluid/surface interactions (e.g. deposition processes, tropical and open ocean convection, etc.) it is necessary to couple microscopic, possibly stochastic models describing an active at small scales boundary layer to continuum PDE, modelling an adjacent bulk fluid phase. Thus, features of the microscopic model will enter as an under resolved sub-grid effect in the coupling with the coarse computational grid of the macroscopic PDE.
The proposed projects focus on key aspects of the aforementioned issues, roughly divided in two categories: (I) Mathematical strategies for the coarse-graining of microscopic MD/MC models; loss of information and adaptivity in coarse-graining (II) Sub-grid modelling, analysis and simulation for hybrid deterministic/stochastic systems describing phenomena with fluid/surface interactions. An array of novel multi-scale mathematics and simulations is proposed to tackle these problems, drawing ideas from statistical mechanics, non-linear PDE, information theory and finite elements. A key feature of our approach already illustrated in some of our recent work, is to microscopically derive a hierarchy of coarse-grained stochastic models permitting dramatic speed-ups in simulations since they involve a reduced set of observables but still incorporate microscopic interactions and noise.
Call for proposal
See other projects for this call