Jean-Louis Roch

Papers

Total Citations

H-Index

About

Jean-Louis Roch is a distinguished researcher in high-performance computing, parallel algorithms, and numerical linear algebra. His work focuses on developing efficient parallel methods for exact computations, particularly in the domain of matrix triangularizations—a foundational operation in scientific computing. Roch’s most-cited paper, "On parallel block algorithms for exact triangularizations" (2002), with 10 citations, introduces novel block-based strategies that optimize parallelism while preserving numerical exactness, addressing critical challenges in scalability and precision for large-scale systems. This contribution has influenced subsequent research in fault-tolerant and communication-avoiding algorithms, bridging theoretical rigor with practical implementation. Roch’s broader impact lies in advancing the understanding of how to decompose complex linear algebra tasks into parallel workflows that minimize overhead and maximize throughput, a key enabler for modern supercomputing and data-intensive applications. His work is particularly valued for its clarity in exposing trade-offs between parallelism, memory access, and numerical stability, making it a reference point for students and engineers designing high-performance solvers. Through his research, Roch has helped shape the landscape of parallel exact computation, leaving a lasting mark on both algorithmic theory and real-world computational science.