High Performance Computing Group
Developing, adapting, and applying novel approaches for large-scale scientific simulations on HPC systems
The High Performance Computing Group has considerable expertise with numerical algorithms, such as linear system and partial differential equation solvers, molecular dynamics, Monte Carlo simulations, domain decomposition, and tensor algebra. We also are experienced with modern parallel programming models, such as CUDA, SYCL, HIP, OpenMP, Kokkos, and MPI. With machine learning playing an increasing role in different scientific domains, we conduct research in developing, adapting, and applying novel approaches for large-scale scientific simulations on HPC systems. As Moore’s law slows down, quantum computing has emerged as a promising technology to power next-generation scientific discovery. The HPC Group actively is investigating using the current noisy intermediate-scale quantum computer for scientific applications with a focus on mitigating errors introduced by imperfect qubits.
We strive to be the nexus between advanced computing technologies and large-scale scientific applications. Please reach out to us for potential collaboration opportunities.
Group Projects
We are constructing a GANs-based Cycle-Consistent Cross-Domain simulation augmentation framework to improve scientific simulations and better represent reality. We also are seeking ways to encode relevant physics conservation laws and symmetries into the framework while applying effective high-performance computing techniques for scalability.
Developing novel algorithms and techniques for addressing some of the most challenging problems in scaling lattice quantum chromodynamics (QCD) simulations for deployment on leadership-class computing facilities. Underpinning this project is a close collaboration with applied mathematicians from the SciDAC-5 FASTMath/PETSc institute.
This project is developing and implementing QC algorithms applicable for studying strongly coupled theories such as QCD in a regime where classical lattice simulations suffer from the fermion sign problem, including problems associated with astrophysical objects (e.g., neutron stars) and heavy-ion collisions.
For NWChemEx, Brookhaven Lab has been involved in the development of density functional theory capabilities, as well as atomic forces. In addition, the Lab has been working on the science challenge use cases to assess the code’s capabilities in biochemistry and catalysis applications.
This project addresses the computational bottlenecks of the PR-DNS models through two research thrusts: (1) use of large-scale heterogeneous high performance computing (HPC) systems, and (2) the development of novel machine learning-based PDE solvers to speed up PR-DNS simulations.
The High Energy Physics Center for Computational Excellence (HEP-CCE) is a multi-lab activity bringing new computational capabilities to bear in support of High Energy Physics science goals.