Frameworks: An Interoperable Software Ecosystem for Many-Body Electronic Structure Calculations

先进材料的发展是能源生产、储存和分配、无线通信和量子技术等具有国家战略意义的领域取得进展的关键驱动力。例如，太阳能电池板、固态电池、触摸屏、平板显示器和量子计算机原型的运行在很大程度上依赖于精确到单个原子尺度的先进材料的特性和功能。进一步提高这些材料的性能，以及设计具有新功能的全新材料，需要详细了解宏观性能，如载电、吸收和发射光的能力，以及储存化学能的能力，是如何从每种化合物的元素组成和原子结构中产生的。在这种背景下，在超级计算机上通过求解量子力学基本方程来模拟材料行为已成为实验研究不可或缺的补充。今天，有大量的高性能计算软件来研究和预测处于最低能量状态或基态的材料的性质。这些工具主要基于密度泛函理论，这是一种极其成功的概念范式，允许我们找到整个材料的量子力学薛定谔方程的近似而准确的解。虽然这些方法对于预测热力学相图等结构和能量性质是必不可少的，但它们不适合描述更高级的功能性质，如光-物质相互作用，电场和磁场下的电荷输运，以及超导等宏观量子现象。目前的项目通过开发一个全面的软件生态系统来填补这一空白，以计算和预测材料的功能性质，而不是目前使用密度泛函理论所能做到的。这笔赠款支持的网络基础设施将使先进功能材料能够在原子尺度上进行合理设计，并将为能源、计算和量子技术的下一代材料的发展奠定基础。研究计划将与教育活动紧密结合，以促进不同社区的科学研究。为此，每年将为用户和开发人员组织网络研讨会、学校和黑客松。该项目的目标是创建一个可互操作的软件生态系统，以使用多体场论方法在原子尺度上对材料进行建模和设计。多体电子结构方法定义了精确度、可靠性和预测能力的黄金标准，但这些方法在学术界和工业中的广泛采用受到基本理论和算法的复杂性以及缺乏广泛的互操作性和共享数据标准的阻碍。该项目扩展并结合了EPW、BerkeleyGW和SternheimerGW三个软件包的互补优势，形成了一个以用户为中心、集装箱化的模拟实验室，具有共享的数据格式和主要密度函数理论代码的内置兼容层。这一网络基础设施促进了对先进材料中电子和晶格自由度之间相互作用的理解，并扩大了可以预测精度计算的性质范围，包括：有限温度准粒子能带结构；光吸收和发射光谱；激子、极化子及其耦合；超导；载流子输运；以及驱动量子系统。此外，这个网络基础设施将通过分发经过管理的、可重复使用的和可互操作的开放源代码，并通过提供一个平台来开发和测试多体电子结构方法的新算法和软件，从而加速未来的软件开发。拟议工作的核心是培养一支多元化、包容性和具有全球竞争力的STEM劳动力队伍，涵盖数据驱动的材料研究和网络基础设施开发。该奖项由高级网络基础设施办公室颁发，由NSF数学和物理科学局内的材料研究部联合支持。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The development of advanced materials is a key driver of progress in areas of national strategic importance, such as energy generation, storage, and distribution, wireless communications, and quantum technologies. For example the operation of solar panels, solid-state batteries, touch screens, flat-panel displays, and quantum computer prototypes critically relies on the properties and functionalities of advanced materials down to the scale of individual atoms. Further improving the performance of these materials, as well as designing brand new materials with novel functionalities, requires a detailed understanding of how macroscopic properties, such as the ability to carry electricity, to absorb and emit light, and to store chemical energy, emerge from the elemental composition and the atomic structure of each compound. In this context, simulating materials behavior by solving the fundamental equations of quantum mechanics on supercomputers has become an indispensable complement to experimental research. Today there exists an abundance of high-performance computing software to investigate and predict the properties of materials in their lowest energy state, or ground state. These tools are primarily based on density functional theory, an incredibly successful conceptual paradigm that allows us to find approximate yet accurate solutions of the Schrödinger equation of quantum mechanics for entire materials. While these methods are essential for predicting structural and energetic properties such as thermodynamic phase diagrams, they are not suitable to describe more advanced functional properties such as light-matter interactions, charge transport under electric and magnetic fields, and macroscopic quantum phenomena such as superconductivity. The current project fills this gap by developing a comprehensive software ecosystem to compute and predict functional properties of materials beyond what is currently possible with density functional theory. The cyberinfrastructure supported by this grant will enable the rational design of advanced functional materials at the atomic scale, and will underpin the development of next-generation materials for energy, computing, and quantum technologies. The research program will be tightly integrated with educational activities to promote scientific research in diverse communities. To this end, webinars, schools and hackathons for users and developers will be organized annually.The aim of this project is to create an interoperable software ecosystem to model and design materials at the atomic scale using many-body field-theoretic approaches. Many-body electronic structure methods define a gold standard for accuracy, reliability, and predictive power, but the widespread adoption of these methods in academia and in industry is hindered by the complexity of the underlying theories and algorithms, as well as the lack of broad interoperability and shared data standards. The project expands and combines the complementary strengths of three software packages, EPW, BerkeleyGW, and SternheimerGW, into a user-centric, containerized simulation laboratory with shared data formats and built-in compatibility layers for major density-functional theory codes. This cyberinfrastructure advances understanding of the interplay between electronic and lattice degrees of freedom in advanced materials, and expands the range of properties that can be calculated with predictive accuracy, including: finite-temperature quasiparticle band structures; light absorption and emission spectra; excitons, polarons, and their couplings; superconductivity; carrier transport; and driven quantum systems. Furthermore, this cyberinfrastructure will accelerate future software development by distributing curated, reusable, and interoperable open-source code, and by providing a platform to develop and test new algorithms and software for many-body electronic structure methods. Central to the proposed effort is the training of a diverse, inclusive, and globally competitive STEM workforce cutting across data-driven materials research and cyberinfrastructure development.This award by the Office of Advanced Cyberinfrastructure is jointly supported by the Division of Materials Research within the NSF Directorate for Mathematical and Physical Sciences.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.