MHD studies towards simulating White Dwarf Mergers using Castro Mar´ ıa G. Barrios Sazo Stony Brook University
Castro Castro is an AMR code that solves the compressible hydrodynamic equations for astrophysical flows ◮ conservation of mass ∂ρ ∂ t = −∇ · ( ρ u u u ) + S ext ,ρ ◮ conservation of momentum ∂ ( ρ u u u ) = −∇ · ( ρ u u uu u u ) − ∇ p + ρ g g g + S ext ,ρ u u u ∂ t ◮ conservation of total energy ∂ ( ρ E ) = −∇ · ( ρ u uE + pu u u ) + ρ H nuc + ρ u u u u · g g g + S ext ,ρ E ∂ t User can supply EOS and reaction networks Self gravity Radiation, gray and multigroup FLD Supports 1-d, 2-d and 3-d
Castro Zingale et al. 2017 × 10 4 10 7 Adaptive Mesh 4 10 6 Refinement, built on 10 5 AMReX 2 10 4 Simultaneous refinement � y (km) in space and time cm 3 10 3 � g 0 ρ Hybrid parallelism, 10 2 MPI+OpenMP 2 10 1 Regression tests 10 0 4 Ongoing development for 10 − 1 4 2 0 2 4 GPUs × 10 4 x (km) Open Source WD merger showing 2-level grid structure github.com/AMReX-Astro/Castro
Scaling OLCF Titan Castro 3-d wdmerger (hydro + Poisson gravity) 10 2 NERSC Cori avg. time / step number of threads 10000 20000 30000 40000 50000 60000 10 2 10 1 Castro 3-d wdmerger (hydro + Poisson gravity) avg. time / step 256 3 no AMR 512 3 base + one 4x level 10 1 10 0 10 2 10 3 10 4 number of cores StarLord GPU performance 120 100 10 0 256 3 gnu 512 3 gnu no AMR 256 3 intel 512 3 intel base + one 4x level zones/ s updated 80 10 1 10 2 number of nodes 60 40 Zingale et al. 2017 20 0 0 20 40 60 80 100 120 number of GPUs OLCF Summitdev
MHD – Algorithm Assumes perfectly conducting fluid Only electric fields induced by motion of magnetized fluid Miniati et al. 2011 Numerical schemes must ensure ∇ B B = 0 ∇ ∇ · B Magnetic field defined at face center Electric field at cell edges updates it
MHD port to Castro Integrating MHD solver from Nyx (cosmology code) Support general EOS Add species Coupled with self-gravity conservative form Verification Shock tube test problem solved using 256 and 800 grid cells at t = 0.1 s.
White Dwarf mergers Continue with WD merger calculations Katz et. al. 2016 Study feasibility of thermonuclear detonation Explore different initial models ◮ He layer ◮ CO + He mergers Investigate algorithmic Presence of magnetic fields improvements and effect on fluid ◮ Well-balanced scheme instabilities could drive a with rotation hotspot into a detonation ◮ Self-consistent initial conditions Include MHD in WD Merger simulation
Timeline for next year Other work Sept. 2018 – Verification and optimization of MHD solver Contributing on GPU efforts Oct. 2018 – Explore WD merger initial models Other planned research events Jan. 2018 – MHD solver with gravity tested BNL GPU Hackathon, Sept. Apr. 2019 – WD merger 2018 investigation of self-consistent initial SIAM CSE conditions Minisymposium Summer. 2019 – WD merger presentation, Feb. simulations with MHD and 2019 additional physics
Recommend
More recommend
