Simulations of structure and nonequilibrium dynamics of soft materials

材料的宏观行为常常受到分子水平上的机制的密切控制。计算机模拟具有独特的定位来阐明这种机制，并具有解决真实材料复杂性的能力。我们建议使用分子模拟来研究发生在软材料中的两个重要现象。我们的目标之一是了解无序材料（如非晶聚合物或金属玻璃）强度的原子起源，这些材料在许多日常承重应用中使用。模拟可以在纳米尺度上识别塑性的微观机制，并将其与宏观材料响应联系起来。韧性或断裂能是由一系列非线性过程决定的，包括屈服、空化和剪切带的形成。分子水平的结果将使我们能够构建新的模型，将从原子描述到宏观连续体的多个长度和时间尺度连接起来，并可以定量预测材料的行为。我们还将开发新的计算方法来计算带电大分子的结构和功能。几乎所有的生物分子系统，如蛋白质、膜和DNA，都含有侧基，这些侧基在水中分离成一个带电的大分子和许多小的“反离子”。计算这些分子之间产生的长期静电力目前是数值模拟的主要瓶颈。更困难的是要有效地考虑到分子和周围溶剂之间的大介电对比度，这对许多生物功能有很强的影响，如蛋白质折叠或配体与受体的结合。我们的新算法将能够处理这种局部极化效应，并检查其对典型生物物理情况（如带电表面附近的分子）的影响。该技术最终将为离子通过小分子孔或通道的运输提供更精确的模型，这是一个对活细胞功能非常重要的动态调节过程。

Macroscopic material behavior is often intimately controlled by mechanisms on the molecular level. Computer simulations are uniquely positioned to illuminate such mechanisms and have the capability to address the complexity of real materials.  We propose to use molecular simulations to study two important phenomena that occur in soft materials. One of our goals will be to understand the atomistic origin of the strength of disordered materials such as amorphous polymer or metallic glasses, which are used in many everyday load-bearing applications.  Simulations can identify microscopic mechanisms of plasticity on the nanoscale and relate them to the macroscopic material response. The toughness or fracture energy is determined by a range of nonlinear processes that include yielding, cavitation and the formation of shear bands. Molecular level results will enable us to construct new models that bridge the multiple length and time scales from an atomistic description to the macroscopic continuum and that can quantitatively predict material behavior. We will also develop new computational methods to calculate the structure and function of charged macromolecules. Virtually all biomolecular systems such as proteins,  membranes and DNA contain side groups that separate in water into a charged macromolecule and many small "counterions". Computing the resulting long ranged electrostatic forces between these molecules currently poses a major bottleneck to numerical simulations. It is even more difficult to efficiently take into account the large dielectric contrast between the molecules and the surrounding solvent, which has a strong effect on many biological functions such as protein folding or the binding of ligands onto receptors.  Our new algorithms will be capable of treating such local polarization effects and examine their consequences for typical biophysical situations such as molecules near charged surfaces.  The technique will finally lead to more accurate models for the transport of ions through small molecular pores or channels, which is a dynamical regulatory process of great importance for the function of living cells.