Dr. Ramesh Balakrishnan is a Computational Scientist (a.k.a Fluid Mechanic) in the Computational Science (CPS) and Leadership Computing Facility (LCF) Divisions at Argonne National Laboratory. He develops scalable flow solvers, using a variety of spatial discretization techniques, for simulations on heterogeneous computing platforms, such as Aurora. In keeping with the hacker ethic, he does share his hacked codes (and codelets) with his collaborators, who include researchers from industry and academia who pursue computational research in flow simulations on ALCF platforms. Currently, he is developing open source codes for large scale turbulence simulations of canonical and real flows, where the aim is to use the data from highly resolved canonical flow simulations to inform the sub-grid model development for “real” flow simulations on coarser meshes.
His R&D exploits are funded by DOE/ASCR, DOE/ECP, and the DOE/EERE WETO office. By way of formal background, he has a Ph.D. in Aerospace Engineering (Specialization in Computational Fluid Dynamics) and has been pursuing research-like work in turbulence simulations for around two decades. All of his “science” simulations have been on the IBM Blue Gene/Q (Mira) (now retired), and are now on the Cray XC40 (Theta) at ALCF. He also uses the IBM CPU+GPU platform (Summit) at OLCF for studying the performance of scientific codes and has used the Cray XC40 (Cori) for some scaling/performance studies as well. In addition, he is known to have a broad understanding of PDE solvers, numerical linear algebra (and libraries), programming models (MPI, Kokkos, OpenMP), and some familiarity with machine learning.
Research Interests:
- Turbulence
- Computational Fluid Dynamics
- Numerical Linear Algebra
- Uncertainty Quantification
- Machine Learning
- Programming Models