Federal Project Grant Award Summary

The University of Utah received a $384,964 Project Grant award from the National Science Foundation (NSF) Division of Information and Intelligent Systems under the Computer and Information Science and Engineering (CISE) program (CFDA 47.070), effective May 1, 2026, through April 30, 2029. This collaborative research initiative delivers advanced compiler infrastructure development focused on enhancing scientific computing capabilities. The primary deliverables include methods to synthesize domain-specific and program-specific static analyses that enable scientists to express and statically enforce complex physical and mathematical constraints within compiled code. The project creates semantic MLIR (Multi-Level Intermediate Representation) dialects tailored for scientific computing domains, particularly for floating-point number and tensor operations, while integrating correct static analyses into the widely-used LLVM compiler infrastructure.

The research consists of three technical thrusts designed to establish a synthesis-based foundation for compiler construction: (1) synthesizing analyses through stochastic search and SMT-based verification methodologies; (2) developing semantic MLIR dialects for scientific computing domains; and (3) advancing project-specific correctness verification techniques. The ultimate product is a compiler development process wherein scientists specify computational requirements and automated tools generate the corresponding implementation, thereby reducing the burden of manual proof and enhancing the research capabilities of the broader scientific computing community. This work directly supports NSF's commitment to advancing research cyberinfrastructure and accelerating discovery across computing and information science.