Registered Data

[02049] A Hybrid AMR Low-Rank Tensor Approach for Solving the Boltzmann Equation

  • Session Time & Room : 5D (Aug.25, 15:30-17:10) @D402
  • Type : Contributed Talk
  • Abstract : Computational complexity grows exponentially in a grid-based approach to modeling the Boltzmann equation as dimensionality increases. Scalable low-rank tensor decomposition techniques have been developed to mitigate such issues, but challenges remain when applying real-world boundary conditions in engineering systems with complex geometries. To address these challenges, we propose a novel hybrid algorithm where quad-tree adaptive mesh refinement (AMR) is applied in real space while a low-rank approximation is used in velocity space.
  • Classification : 76P05, 35Q83, 65M50
  • Author(s) :
    • Samuel Jun Araki (Jacobs Technology Inc. / Air Force Research Laboratory)