Researches

TODO: Add showcase in posts

I’m interested in Numerical PDE methods, mostly focusing on aspects related to spectral element method, with the aide of high-performance computing. My current research is centered on Nek5000/NekRS¹, working with ECP/CEED² project.

• High-Order Methods for Numerical Partial Differential Equations
  • Incompressible Navier-Stokes (NS), Reynolds-Averaged Navier-Stokes (RANS)
  • Poisson-Nernst-Planck (PNP)
  • Steady state solver with Jacobian-free Newton Krylov (JFNK) method.
  • Preconditioner techniques for Poisson and advection-diffusion system(s).
    Domain decomposition, multigrid methods, low-order preconditioner, and more.

• High-Order Mesh Generation
  • Voronoi cell based all-hex meshing algorithms for dense-packed spheres
  • Mesh smoothing and optimization

  

• High-Performance Computing
  • Large-scale simulations on top 10 supercomputers including CPUs and GPUs
  • Exascale workflow and performance studies
  • In-situ visualization

• NekCEM/Nek5000/NekRS user, developer, and forum contributor

---

1. Nek5000 is an efficient scalable open source computational fluid dynamics (CFD) solver using spectral element method (SEM) and NekRS is its GPU variant. They both target large scale high fidelity turbulence simulations on supercomputers. ↩

2. Center for Efficient Exascale Discretizations [Website] ↩