Federal Grant Award Summary

Cornell University received a $425,666 Project Grant from the National Science Foundation's Office of Advanced Cyberinfrastructure under the Computer and Information Science and Engineering (CISE) program (CFDA 47.070) to establish sparse linear algebra as a unifying computational abstraction for data-intensive scientific discovery. The award, initiated July 1, 2026, and concluding June 30, 2031, will develop systematic methodologies for translating irregular scientific computations onto sparse linear algebra primitives, enabling portable execution across heterogeneous computing systems including CPUs (central processing units), GPUs (graphics processing units), and emerging architectures. The project addresses a critical gap by lowering barriers to advanced high-performance computing systems, thereby accelerating scientific discovery in genomics and other data-intensive fields where computational bottlenecks currently delay research outcomes.

The initiative comprises three primary research thrusts: automatic translation of domain code into sparse linear algebra form using semantic-signature layers and equality saturation techniques; identification of sparsity structure and algebraic provenance for specialized sparse primitive optimization; and extension of sparse linear algebra methodologies across diverse computing platforms. A central workforce development component includes an undergraduate training program that recruits students nationwide to participate in international supercomputing competitions, creating a talent pipeline for high-performance computing expertise and strengthening U.S. computational competitiveness in data-intensive sciences.