CASC’s Research Software Engineering (RSE) Group was founded in 2022 as a bridge between LLNL’s domain scientists and software teams. We apply software engineering expertise and a deep understanding of scientific processes to deliver high-quality software that advances cutting-edge research. In collaborating with research projects across LLNL, our work spans a wide range of HPC topics and software engineering best practices such as packaging, build systems, and continuous integration (CI) testing. We also lead and contribute to a number of open-source software projects, including Abmarl for multi-agent reinforcement learning, Caliper and Hatchet for performance analysis, MFEM for finite element methods and adaptive mesh refinement, Spack for package management, Umpire for heterogeneous memory management, and the multi-institutional ParFlow code for watershed flow modeling.
David Beckingsale: performance analysis, code optimization, mini-applications and parallel programming models
David Boehme: performance analysis tools, performance optimization, parallel and distributed architectures, parallel programming paradigms
Stephanie Brink: performance analysis tools, performance optimization, power-aware HPC, low-level power management and control
Yohann Dudouit: high-order methods, finite element methods, discontinuous Galerkin methods, GPU programming, high performance computing, hp-adaptivity
Ephraim Rusu: multi-agent reinforcement learning, agent-based simulations, software architecture and design
Abhik Sarkar: multi-processor architecture, system software performance analysis and optimization, binary translation and processor virtualization
Steve Smith: parallel computer network simulation, co-simulation of communication and power grid networks, HPC application development, code optimization