This first-of-its-kind workshop, sponsored by the US National Science Foundation (award CMMI-2315913) and the Air Force Office of Scientific Research, is organized by Elif Ertekin (U. Illinois), Giulia Galli (U. Chicago), and Ali Sayir (AFOSR). This workshop will be held at the Basic Research Innovation Collaboration Center (BRICC) on June 8-9, 2023 in Arlington, VA.
Workshop Overview: With substantial developments and focused efforts targeting the creation of a Materials Innovation Infrastructure over the last ten years, computation, modeling, and simulation have become integral to the materials discovery, innovation, and deployment cycle. As a result, quantifying how sources of error in computation propagate through material property prediction, synthesis and processing, and the manufacturing techniques to incorporate a new material into a product are keys to realizing the goals of the Materials Genome Initiative [1], the National Strategy for Advanced Manufacturing [2], and the NSF Big Idea “Harnessing the Data Revolution” [3]. The purpose of this workshop is to bring together researchers in first-principles modeling, machine learning/informatics approaches, multiscale modeling, and modeling/artificial intelligence for process design/manufacturing. We will discuss cross-cutting pathways, challenges, and opportunities for systematic quantification of errors and uncertainty. We aim to identify best practices and create momentum towards the integration of error and uncertainty analysis into standard computational workflows, so that rigorous error determination and reporting become a standard for the community of modelers.
Workshop Format: To promote a vigorous discussion of the need, challenges, and opportunities for quantifying errors and uncertainty in areas of modeling that are key to materials discovery, manufacture, and deployment, this two-day workshop will feature ~20 minute presentations from invited speakers and leave ample time for focused group discussion around a set of concrete questions. Discussions will center on (i) error control for electronic structure, first principles, and high-throughput approaches, (ii) error control in the application of machine learning/artificial intelligence approaches, (iii) error control for materials synthesis, processing, and manufacture, (iv) best practices for modelers when working with experimentalists, and (v) incorporating error analysis into workflows and materials data repositories. Cross-cutting themes pertinent to all these topics include (i) quantification of error, uncertainty propagation, epistemic and aleatoric uncertainty, (ii) verification, validation, and benchmarking, (iii) model vs numerical errors, and model selection.
Guidelines/Information: The primary goal of the workshop is to create a set of guidelines and recommendations to collectively move the modeling community towards best practices in error analysis and uncertainty quantification. Invited speakers are tasked with presenting their perspective in the form of a high level overview on the need, challenges, and opportunities for error and uncertainty analysis. Discussion leaders are asked to work together with the conference organizers to create a concrete, focused set of questions that will help converge the audience on best practices and approaches. The outcome of the workshop will be a community journal paper to be shared with the community on best practices in error control and uncertainty analysis across the materials innovation pipeline.
Attendance/Support: Participation is limited to 120 attendees, a portion of which is reserved for student participants. We have funding available to support the participation of students (undergraduate or graduate students currently enrolled at institutions in the United States or Canada) to attend the workshop, especially students coming from backgrounds traditionally underrepresented in STEM disciplines - please contact via e-mail Elif Ertekin (ertekin [at] illinois [dot] edu) for inquiries.
[1] Materials Genome Initiative Strategic Plan, National Science and Technology Council, 2021; National Academies Report, 2022.
[2] National Strategy for Advanced Manufacturing, 2022.
[3] NSF’s 10 Big Ideas, 2016.