SI2-SSE: Multiscale Software for Quantum Simulations in Materials Design, Nano Science and Technology

千万亿次计算平台的出现为通过仿真进行变革性研究带来了前所未有的机遇。然而，未来的突破将取决于高端模拟软件的可用性，这将充分利用这些无与伦比的资源，并为国家利益的关键领域的科学进步提供长期寻求的第三条途径。该奖项将在材料设计，纳米科学和纳米技术的广泛领域提供一套开源的千万亿次量子模拟工具。材料预测和设计是最近创建的材料基因组计划的关键方面，该计划旨在“以一小部分成本以至少两倍的速度部署先进材料。“计算材料设计是该倡议的关键方面，它依赖于计算指导实验。后者的结果将反过来导致迭代反馈循环中的后续计算。纳米科学在纳米尺度上研究材料的性质和过程，有望开发出具有全新性质的材料和系统。然而，纳米尺度的性质很难测量，甚至更难从理论上预测。只有能够完全解释基本尺度上的复杂性和可变性的模拟，才有机会预测和利用出现的宏观特性。这确实需要千万亿次资源和高效的千万亿次软件工具。该奖项将开发基于真实空间多重网格（RMG）软件套件的软件工具，并将其分发给全国用户社区。RMG代码已经扩展到128，000个CPU核心和18，000个GPU节点。该奖项将通过开发具有改进收敛性的新迭代方法，优化现有和新千万亿次计算平台的附加模块，以及创建易于使用的主代码接口来进一步增强RMG。RMG使用研讨会将在XSEDE研讨会和NSF超级计算中心的其他会议上进行。RMG将通过一个门户网站分发，该门户网站还将包含用户论坛和视频教程，在现场用户会议上录制。将维持一个主要千万亿次平台的代表性实例库。RMG将实现前所未有的量子模拟，使功能纳米或生物纳米结构的构建模块的研究成为可能，这些结构通常涉及数千个原子，必须以必要的保真度进行描述。千万亿次量子模拟软件及其用户社区的发展将导致各领域内和各领域之间的思想交流。在这一领域接受培训的学生和博士后将有重大的晋升机会，并对自己产生重大影响。

The emergence of petascale computing platforms brings unprecedented opportunities for transformational research through simulation. However, future breakthroughs will depend on the availability of high-end simulation software, which will fully utilize these unparalleled resources and provide the long-sought third avenue for scientific progress in key areas of national interest. This award will deliver a set of open source petascale quantum simulation tools in the broad areas of materials design, nano science and nanotechnology. Materials prediction and design are key aspects to the recently created Materials Genome initiative, which seeks to "deploy advanced materials at least twice as fast, at a fraction of the cost." Computational materials design is the critical aspect of that initiative, which relies on computation guiding experiments. The outcomes of the latter will in turn lead to follow-up computation in an iterative feedback loop. Nanoscience, which studies properties of materials and processes on fundamental scale of nanometers, promises development of materials and systems with radically new properties. However, the nanoscale properties are hard to measure and even harder to predict theoretically. Only simulations that can fully account for the complexity and variability at that fundamental scale stand a chance of predicting and utilizing the macroscopic properties that emerge. This truly requires petascale resources and efficient petascale software tools. This award will develop software tools build on the real-space multigrid (RMG) software suite and distribute them to the national user community. The RMG code already scales to 128,000 CPU cores and 18,000 GPU nodes. The award will further enhance RMG through development of new iterative methods with improved convergence, optimization of additional modules for existing and new petascale computing platforms, and creation of ease-to-use interfaces to the main codes. Workshops in RMG usage will be conducted at XSEDE workshops and other meetings of NSF supercomputing centers. RMG will be distributed through a web portal, which will also contain user forums and video tutorials, recorded at live user sessions. A library of representative examples for the main petascale platforms will be maintained. RMG will enable quantum simulations of unprecedented size, enabling studies of the building blocks of functional nano or bio-nano structures, which often involve thousands of atoms and must be described with the requisite fidelity. The development of petascale quantum simulation software and its user community will lead to cross-fertilization of ideas both within and across fields. Students and postdocs trained in this area will have significant opportunities for advancement and making substantial impact on their own.