EAGER: Scalable PetaScale Computing for Dielectric Response in Biophysical, Optomechanical, and Geophysical Systems

智力价值非均匀介质之间的分子间力影响微观和宏观行为的现象很多。这些例子涵盖了生物、工程和物理科学中的一系列现象，并对潜在的相互作用--范德华力/色散力--进行了简单的描述，但其表述方式使得定量预测的计算成本很高，从而限制了它们的适用范围。此外，在所有应用中，由于Lifshitz(1)引起的色散力相互作用的连续理论在介电性质不均匀的情况下如果不是不妥协的话将变得繁琐，(2)受到潜在介电性质的精确知识的强烈限制，以及(3)在纳米量级的交换光子波长处分解，然而我们理解材料的纳米结构对其性质具有控制影响。最后，在许多应用中，例如介电体之间的共振相互作用或控制悬浮液中的絮凝或生物膜的操纵，需要对时空变化的材料进行实时计算。因此，需要一种PB级的计算方法，通过使用新的迭代积分方程技术(包括第一类和第二类Fredholm型积分方程之间的转换)引入快速迭代过程来实现显著的加速。估计后一种加速比将达到10个数量级左右，而并行多区域分解方法的加速比接近N^3阶，其中N是分解区域的个数。通过探索计算准则和使用正、逆散射方法解决这些问题所需的基本岩心介质数据，这项工作将促进计算科学和依赖于它的科学的发展。更广泛的影响这项工作的更广泛影响的一个子集包括：(A)通过将介电问题分解为子问题的高度并行的区域分解技术，高效地使用拍级计算；(B)建立在许多科学和工程领域广泛感兴趣的重要介电材料的性质；(C)基于我们在多种分层材料和许多环境中的计算的经验，结合纳米技术的各个方面的尖端技术；(D)培训这些尖端领域的学生和博士后，使他们能够在研究中采取许多方向，从而赋予他们市场上广泛的技能；以及(E)与政府和大学实验室以及工业界建立牢固的联系，以便尽可能广泛地传播和使用这些成果。

Intellectual Merit There is a wide range of phenomena wherein intermolecular forces between inhomogeneous media influence both microscopic and macroscopic behavior. The examples span a range of phenomena in the biological, engineering and physical sciences and the underlying interactions?Van der Waals/Dispersion forces?are simply described but formulated in a manner that makes quantitative predictions computationally costly and hence limits their range of applicability. Moreover, in all applications the continuous theory of dispersion force interactions due to Lifshitz (1) becomes cumbersome if not intransigent when the dielectric properties are inhomogeneous, (2) is strongly limited by accurate knowledge of the underlying dielectric properties and (3) breaks down at exchange photon wavelengths of the order of a nanometer and yet we understand that the nanostructure of materials has a controlling influence on their properties. Finally, in many applications, such as a resonant interaction between dielectric bodies or control of flocculation in a suspension or manipulation of biomembranes, real time calculations of spatiotemporal varying materials is required. Thus, a petascale computational approach is required to achieve significant speedups through introduction of fast iteration processes via the use of novel iterated integral equation techniques including conversion between Fredholm-type integral equations of the 1st and 2nd kinds. It is estimated that the latter speedups will be of order 10 or so, while parallel multi-domain decomposition methods give speedups nearly of order N^3 where N is the number of decomposed domains. By exploring a computational rubric and the basic core dielectric data needs to attack these problems using both direct and inverse scattering methods, this work will advance both computational science and the science that relies upon it. Broader Impacts A subset of the broader impacts of this work include: (a) The high efficiency use of petascale computing via highly parallel domain decomposition techniques that split dielectric problems into sub-problems; (b) Establishment of the important dielectric materials properties of wide interest across many areas of science and engineering; (c)Building on coupling sophisticated technologies from various aspects of nanotechnology based on our experience with diverse layered materials and computing in many settings; (d) Training a student and post doc in these cutting-edge fields that will provide them the ability to take many directions in research and hence give them marketable broad-based skills; and (e) Establishment of strong ties with government and university labs as well as industry to enable the widest possible dissemination and use of these results.