Scholarlink

In simulations of relativistic gas flows, complex structures often arise in localized regions much smaller than the computational domain. However, processes occurring outside these regions can significantly influence—and even govern—their formation and dynamics. Adaptive mesh refinement (AMR) provides an efficient framework for domain discretization when resolving such multiscale phenomena. We present a novel Patch-Block-Structured AMR technique tailored for special relativistic hydrodynamic flows, coupled with a redesigned numerical method optimized for Coarray Fortran parallelization. Our implementation achieves near-linear scalability, with a 30 \(\times \) speedup on 32 cores using 32 Coarray Fortran images. As a demonstration, we apply this parallelized AMR approach to simulate the interaction of a relativistic wind with interstellar clouds.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Adaptive Mesh Refinement Technique for High-Performance Hydrodynamic Modeling of Relativistic Flows

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Igor Kulikov,
  • Igor Chernykh,
  • Vardan Elbakyan,
  • Dmitry Karavaev,
  • Ivan Ulyanichev,
  • Anna Sapetina,
  • Vladimir Prigarin,
  • Oleg Zavyalov,
  • Sergey Lomakin,
  • Elena Akimova,
  • Vladimir Misilov

摘要

In simulations of relativistic gas flows, complex structures often arise in localized regions much smaller than the computational domain. However, processes occurring outside these regions can significantly influence—and even govern—their formation and dynamics. Adaptive mesh refinement (AMR) provides an efficient framework for domain discretization when resolving such multiscale phenomena. We present a novel Patch-Block-Structured AMR technique tailored for special relativistic hydrodynamic flows, coupled with a redesigned numerical method optimized for Coarray Fortran parallelization. Our implementation achieves near-linear scalability, with a 30 \(\times \) speedup on 32 cores using 32 Coarray Fortran images. As a demonstration, we apply this parallelized AMR approach to simulate the interaction of a relativistic wind with interstellar clouds.