Data Reduction for Science
This grant seeks innovative algorithms & workflows for scientific data reduction, applicable across multiple disciplines, focusing on accuracy, efficiency, & uncertainty quantification.
This grant is no longer accepting proposals
Office of Science has archived this opportunity.
Funder: Office of Science - U.S. Department of Energy
Due Dates: May 7, 2024 (Application Deadline) | June 6, 2024 (Archive Date)
Funding Amounts: Total program funding approximately $15,000,000; individual awards range from $600,000 to $3,000,000
Summary: Supports innovative applied mathematics and computer science research to develop rigorous, trustworthy, and broadly applicable scientific data reduction algorithms and workflows.
Key Information: Preference for proposals addressing multiple science applications and one or more priority research directions including accuracy, efficiency, uncertainty quantification, streaming, and new architectures.
Description
The Department of Energy’s Office of Science invites proposals for research focused on advancing scientific data reduction techniques. Modern scientific experiments, observations, and simulations generate data at volumes that exceed current capabilities for storage, analysis, streaming, and archiving in raw form. This funding opportunity aims to support the development of innovative algorithms, techniques, and workflows that reduce data volume while preserving scientifically relevant information with mathematical rigor.
Key challenges include ensuring accuracy, efficiency, and trustworthiness of data reduction methods, enabling progressive data streaming, quantifying uncertainty in preserved features, and adapting methods to emerging computing architectures. The research should be broadly applicable across multiple scientific domains relevant to the DOE mission, such as light sources, accelerators, radio astronomy, cosmology, fusion, climate science, materials science, combustion, power grid, and genomics.
The FOA encourages approaches that go beyond incremental improvements and instead propose significant innovations or unconventional strategies. Cross-cutting themes like artificial intelligence and trust in data reduction are emphasized. Proposals focusing on a single science application may be discouraged unless they demonstrate broader applicability.
This initiative builds on outcomes from a 2021 DOE workshop that identified four priority research directions (PRDs):
- Effective, trusted algorithms and tools for accuracy and efficiency.
- Progressive reduction algorithms for prioritized, efficient data streaming.
- Algorithms preserving features and quantities of interest with quantified uncertainty.
- Mapping data reduction techniques to new computing architectures and use cases.
Research may target high-performance computing environments, scientific edge computing, or any setting where scientific data is collected or processed.
