INSPIRE Track 1: UDiscoverIt: Integrating Expert Knowledge, Constraint-Based Reasoning and Learning to Accelerate Materials Discovery

该INSPIRE奖的部分资金来自计算机与信息科学与工程系信息与智能系统部的信息集成与信息学项目、材料研究部的固态与材料化学项目以及数学与物理科学部的多学科活动办公室。在过去的二十年中，实验高通量实验（HTE）方法得到了快速发展，这对于(i)发现具有高复杂性的新应用材料和（ii）产生对结构/功能，结构/活性和结构/性能关系的深刻理解非常有价值。特别是高光子通量的x射线技术在材料发现方面具有巨大的变革潜力。研究小组利用康奈尔高能同步加速器（CHESS）和加州理工学院人工光合作用联合中心（JCAP）收集的数据。虽然高通量无机文库合成是相对完善的，高通量结构测定，这是在拟议的研究的核心，是在其起步阶段。x射线衍射非常适合于快速收集无机样品中原子排列的信息，但是这些数据不能立即揭示晶体结构。数据分析、数据挖掘和数据解释方法的发展跟不上实验能力的发展。因此，一周内获得的数据可能需要研究人员数月的传统分析。为了最大限度地发挥复杂多维数据集的影响，数据的自动化和机器智能处理是绝对必要的。这个项目直接解决了这个问题；它研究了计算技术，通过约束引导搜索和优化、统计机器学习和推理技术，结合直接的人工输入，允许处理与材料库的HTE结构确定相关的多参数空间。预期的进展包括新的概率方法和计算发现工具，这些工具集成了软硬约束，从材料的底层物理和化学中捕获复杂的背景知识，并从高通量数据分析和机器学习中获得见解。如果该项目成功地在复杂结构确定方面实现了预期的巨大效率收益，它可能会对材料发现和复杂固体化学和物理产生变革性影响。将复杂材料的发现和优化时间从几个月或几年缩短到几个小时或几天的能力，可能会导致产品开发的范式转变，造福社会，技术进步以及对能源、可持续性、健康和生活质量的商业影响。计划向更大的科学界免费传播数据集和计算工具可能会增强该项目的更广泛影响。该项目促进了康奈尔大学计算机科学家和材料科学家之间的跨学科互动，并为两个学科之间的界面培养新一代研究人员提供了更多的机会。

This INSPIRE award is partially funded by the Information Integration and Informatics Program in the Division of Information and Intelligent Systems in the Directorate for Computer and Information Science and Engineering and the Solid State and Materials Chemistry Program in the Division of Materials Research and the Office of Multidisciplinary Activities in the Directorate for Mathematical and Physical Sciences.The past two decades have seen a rapid development in experimental high-throughput experimentation (HTE) methodologies that would be extremely valuable for (i) the discovery of new applied materials with high complexity and (ii) the generation of deep understanding of structure/function, structure/activity and structure/performance relationships. Especially high photon flux X-ray techniques have enormous transformative potential in materials discovery. The research team leverages the data being collected by the Cornell High Energy Synchrotron Source (CHESS) and at Caltechs Joint Center for Artificial Photosynthesis (JCAP). While high-throughput inorganic library synthesis is relatively well-established, high-throughput structure determination, which is at the heart of the proposed research, is in its infancy. X-ray diffraction is well-suited for rapidly collecting information on the atomic arrangements in an inorganic sample, but the data do not immediately reveal a crystal structure. The development of data analysis, data mining and interpretation methodologies has not kept pace with the development of experimental capability. Consequently, data acquired in a week can take many months of traditional analysis by researchers. Automation and machine-intelligent processing of the data are absolutely necessary to maximise the impact of complex multidimensional datasets. This project addresses this state of affairs head-on; It investigates computational techniques that allow dealing with the multiparameter space associated with HTE structure determination of materials libraries, through constraint guided search adn optimization, statistical machine learning, and inference techniques in combination with direct human input into the process. Anticipated advances include new probabilistic methods and computational discovery tools that integrate soft and hard constraints that capture the complex background knowledge from the underlying physics and chemistry of materials with insights gained from high throughput data analytics and machine learning. If the project succeeds in achieving the anticipated enormous efficiency gains in complex structure determination, it could have have a transformative impact on materials discovery and complex solid state chemistry and physics. The ability to reduce complex materials dicovery and optimization from timeframes of months or years to hours or days could lead to a paradigm shift in the development of products benefiting society, with technological advances as well as commercial impact on energy, sustainability, health and quality of life. The planned free dissemination of data sets and computational tools to the larger scientific community is likely to enhance the broader impacts of the project. The project facilitates increased interdisciplinary interactions between computer scientists and material scientists at Cornell University and offer enhanced opportunities for training of a new generation of researchers at the interface between the two disciplines.