Last update July, 2013.  

The focus of my recent research is on the modeling of materials and biological systems. Currently, I am working on the following projects:

  1. Coarse-graining bio-molecular models. Quantifying the dissipative forces.

  2. Nano-scale heat conduction.

  3. Data-driven parameter estimation and uncertainty quantification; Prediction of equilibrium and non-equilibrium properties.

  4. Material interface problems, from quantum-mechanical and atomistic descriptions to interface properties.

  5. Numerical methods for multiscale stochastic models.

  6. Coupling atomistic and continuum models for crystalline solids; Absorbing boundary condition for molecular dynamics models.

  7. Mori-Zwanzig formalism for the reduction of complex dynamical systems.

  8. Mathematical analysis of multiscale methods;

  9. Atomistic-based finite element and boundary element methods for material interface problems;

  10. Understanding crack propagation problem via bifurcation theory;

  11. Non-equilibrium molecular dynamics models.
0. Coarse-graining biomolecular systems.

With collaborators (Liu and Ma), I am interested in the characterizing and quantifying the dissipative forces in the process of coarse-graining a bio-molecular systems. The main goal is to bring back the frictional forces so that dynamic properties can be predicted with good accuracy. In particular, our effort has been placed on:

  • The derivation of the generalized Langevin equation (GLE) from a Langevin dynamics model;
  • Efficient approximations of the memory kernel and random noise terms;
  • Preserving the second fluctuation-dissipation theorem;
  • The connections to subspace Galerkin projections;
  • Data driven parameterizations.

1. Heat conduction in nanoscale materials.

The distinct heat conduction properties have made nano-scale materials extremely successful in application. The standard heat equation, a textbook example of a partial differential equation, and a traditionally successful heat conduction model, fails to describe the conduction process. In spite of tremendous interest and application, little theory exists to provide a more suitable mathematical model. On the other hand, these problems also pose a much greater challenge for direct non-equilibrium molecular dynamics (NEMD) simulations at the atomic scale. For instance, while it is often relatively straightforward to incorporate quantities such as displacement, velocity, temperature and pressure into MD simulations as constraints, temperature gradient is very difficult to impose. The temperature gradient that can be imposed is usually is too large to model realistic systems.

We derive a coarse-grained model for heat conduction in nano-scale mechanical systems. Starting with an all-atom description, this approach yields a reduced model, in the form of conservation laws of momentum and energy. The model closure is accomplished by introducing a quasi-local thermodynamic equilibrium, followed by a linear response approximation. Of particular interest is the constitutive relation for the hear flux, which is expressed non-locally in terms of the spatial and temporal variation of the temperature.

  • Heat conduction in nanoscale materials: A statistical-mechanics derivation of the local heat flux, Physical Review E, 90 032112 (2014) .
  • With Xiaojie Wu, On Consistent Definitions of Momentum and Energy Fluxes for Molecular Dynamics Models with Multi-body Interatomic Potentials, Modelling and Simulation in Materials Science and Engineering, Accepted.

3. Multiscale models for material interfaces.

The mechanical properties of materials are largely determined by the underlying interface structures. We develop computational models based on density-functional theory or/and molecular dynamics models so that the microstructures can be properly taken into account.

4. Multiscale stochastic dynamics.

Stochastic models, especially the ones arising from bio-molecular systems, exhibit multiple time scales. We are interested in developing integrators so that one can reach longer time scales with compromising the numerical accuracy.

5. Coupled Atomistic and Continuum Models for Solids

The main purpose of this project is to develop a coupled atomsitic and continuum models, which is not limited by the temporal and length scales of the atomistic model, or the inaccuracy of the continuum models.

The main component of the multiscale method is a coupling condition at the interface of the two models. The starting point is the observation that both the continuum and atomistic models can be recast into the form of conservation laws. As a result, the coupling condition can be accomplished at the level of fluxes. Such a formulation also allows us to use existing numerical methods for solving conservation laws to better capture elastic waves with sharp front.

A subproblem is the absorbing boundary condition for molecular dynamics model. The goal is to derive and implement the appropriate boundary condition so that the phonons generated by local defects can propagation through the boundary, instead of be reflected back into the system to generate numerical error.

The method has been applied to brittle crack problems to study the impact of strong shocks.

6. Mori-Zwanzig formalism for coarse-graining molecular models of materials and biological system.

In this project, we consider the following problem: Given a subset of atomic degrees of freedom of full MD, called coarse-grain (CG) variables, can we derive a set of equations that only involve the CG variables from the full MD model? Such question is of considerable theoretical and practical interest. This project aims to reduce the dimension of the problem and derive coarse scale models. The models will have a lot of applications.

The reduction of the number of degrees of freedom can be accomplished under the general framework of Mori-Zwanzig. In particular, with the help of a projection operator, the Mori-Zwanzig formalism offers a compact and elegant scheme to derive the coarse-grained models. Compared to the coupling methods, the current approach is more frst-principle based because the continuum description is not introduced at the beginning.

With proper approximations, the procedure yields a set of generalized Langevin equations (GLE) (Li, 2010). Recently, we have made the connection to the conventional Galerkin projection methods. As it turns out, the GLEs can be derived by using a Galerkin projection to an extended subspace.

We have developed an algorithm for evaluating the memory functions in the GLEs. More importantly, the random noise in the GLE can be approximated within the framework, and the fluctuation-dissipation theorem (Kubo 66) is automatically satisfied. Our applications include dislocation dynamics, crack propagations and protein dynamics.

7. Analysis of multiscale models.

The goal of this work is to understand the modeling associated with a coarse-grained model. We have considered quasi-continuum type of methods for systems near and beyond bifurcation points. Our current work focuses on dynamic coupling methods.

8. Atomistic-based Boundary Element Method.

We are developing a new reduced computational model, called atomistic-based boundary element model (ABEM), derived from a full atomistic model for a crystalline solid system. The procedure is based on a domain decomposition method, where atoms near crystal defects are separated from the surrounding region, and a reduction method, which similar to the boundary integral method for continuum models, removes the atoms in the surrounding region. The reduction procedure relies on the lattice Green's function and a summation-by-parts procedure, and it gives rise to a system of equations only involving the atoms at the remote boundary and the interfaces with local defects. As a result, the model is drastically simplied.

This method will be aplied to various types of material interfaces.

8. Mechanics and Physics of Crack Propagation.

We study the transition of the stability for the problem of crack propagation based on nonlinear lattice models. In particular, we study the bifurcation behavior as the load parameters are varied. We make connections to the traditional fracture mechanics theory. More importantly, we study the effect of the strain rate in fracture initiation and propagation.

9. The Mathematical Foundation of Molecular Dynamics Simulations.

Non-Periodic boundary conditions;
Atomic expressions of continuum quantities, e.g. stress and heat fluxes, from non-equilibrium molecular dynamics models;
Asymptotic approximations of ensemble averages;