the physics of flash and a few issues/tricks of the trade
DESCRIPTION
The physics of Flash and A few issues/tricks of the trade. Alan Calder. June 4, 2006. The FLASH Code. Shortly: Relativistic accretion onto NS. Flame-vortex interactions. Type Ia Supernova. Compressed turbulence. The FLASH code Parallel, adaptive-mesh simulation code - PowerPoint PPT PresentationTRANSCRIPT
An Advanced Simulation and Computing (ASC) Academic Strategic Alliances Program (ASAP) Center
at The University of Chicago
The Center for Astrophysical Thermonuclear Flashes
The physics of Flashand
A few issues/tricks of the trade
Alan Calder
June 4, 2006
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
The FLASH Code
Cellular detonation
Compressed turbulence
Helium burning on neutron stars
Richtmyer-Meshkov instability
Laser-driven shock instabilitiesNova outbursts on white dwarfs Rayleigh-Taylor instability
Flame-vortex interactions
Gravitational collapse/Jeans instability
Wave breaking on white dwarfs
Shortly: Relativistic accretion onto NS
Orzag/Tang MHDvortex
Type Ia Supernova
Intracluster interactions
MagneticRayleigh-Taylor
The FLASH code1. Parallel, adaptive-mesh simulation code2. Designed for compressible reactive flows3. Ideal, Resistive, and Hall MHD (Cartesian coords)4. Has a modern CS-influenced architecture5. Can solve a broad range of (astro)physics problems6. Portable- runs on many massively-parallel systems6. Scales and performs well- Gordon Bell prize7. Is available on the web: http://flash.uchicago.edu8. Flash 3 now (pre-)alpha released!
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Hydrodynamics
PPM hydrodynamics based on the Prometheus code of Fryxell.
Directionally split, direct Eulerian implementation of Colella and Woodward (1984) that allows for non-ideal gasses (Colella and Glaz 1985).
2nd-order Strang split in time.
Solves Euler equations for inviscid compressible hydrodynamics in 1, 2, and 3 dimensions and several geometries (Cartesian, 2-d cylindrical, 1-d spherical)
Other `flavors’ of PPM may be released.
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Hydrodynamics
Contact steepener controlled by parameter use_steepening.
Modified states version for use in simulations of objects in hydrostatic equilibrium. Contribution to pressure in Riemann solver from gravity removed. Parameter ppm_modifiedstates.
Interpolation/monotonization procedure of PPM can introduce errors in abundances of species. There is an implementation of the consistent mass advection method of Plewa and Muller (1999). Parameter use_cma_flattening.
Odd/even instability can occur when shocks are aligned with the grid (Quirk 1997). Fix is to switch to HLLE solver in shocks. Parameter hybrid_riemann. Test problem odd_even.
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Euler Equations w/gravity
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Internal Energy Advection
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Multiple Species
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Relativistic Hydrodynamics
Module based on the Pluto code of A. Mignone.
Extension to PPM.
1-, 2-, and 3-d Cartesian, 2-d cylindrical, 1-d spherical geometries
Ideal gas EOS
Directionally split version of Mignone et al. 2005 implemented in Flash.
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
MHD
Based on a finite-volume cell-centered method proposed Powell et al. 1999.
Ideal, Resistive, Hall MHD in Cartesian coords.
Works with other modules: self-gravity, multi-species, burning, general EOS.
Verified against standard benchmarks: MHD shock tube, Brio-Wu problem, shock-cloud, Orszag-Tang problem.
Details of resistive MHD in Malyshkin, Linde et al. (2005)
Dongwook Lee coming to Flash soon!
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Equations of state
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Equations of state
“Helmholtz” EOS for degenerate plasma (stellar material)
P = Pion + Prad + Pele + Ppos + Pcoul
Pion = ideal = 5/3 gas for ionized nuclei
Prad = blackbody = 1/3 aT4
Pele and Ppos = non-interacting Fermions
Pcoul = correction for Coulomb interactions between ions and the surrounding e- gas
Fryxell et al. (2000), Timmes and Arnett (1999)
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Source Terms
Nuclear reactions
7 nuclide “-chain” + Si burning network
13 nuclide “-chain” + heavy-ion network
19 nuclide “-chain” + heavy-ion + H burning network
Someday a general network?
Non-equilibrium Ionization
Stirring
Heating
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Aside: Mesh Adaptivity and R-T Instability
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Finite Volume Hydrodynamics Method (PPM)
Divide the domain into zones that interact with fluxes
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Fluxes at jumps in mesh refinement
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Riemann Problem: Shock Tube
Initial conditions: a discontinuity in density and pressure
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Riemann Problem: Shock Tube
World diagram for Riemann problem
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Riemann Problem: Shock Tube
PPM has special algorithms for these features
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Verification Test: Sod Shock Tube
Demonstrates expected 1st order convergence of error
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Verification Test: Isentropic Vortex
Demonstrates expected 2nd order convergence of error
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Sod Tube W/ AMR
Demonstrates expected 1st order convergence of error, but…
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Riemann Problem: Convex EOS
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
New Validation Results: Vortex-dominated Flows
“Cylinder” of SF6 hit by Mach 1.2 shockLANL
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Shocked Cylinder Experiment
Snapshots at 50, 190, 330, 470, 610, 750 s
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
New Validation Results: Vortex-dominated Flows
Visualization magic from ANL Futures Lab
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Shocked Cylinder Simulations
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
4 shock problem
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
4 contact problem
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Single-mode 3-d Rayleigh-Taylor
Density (g/cc)4 8 16 32 64 128 (grid points) t = 3.1 sec
256
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Boundary Condition
Construct divide_domain for a particular problem
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Summary/Conclusions
Numerical diffusion is a resolution-dependent effect that can significantly alter results.
Care must be taken when adding physics to hydro (e.g. convex EOS)
AMR is tricky.
Need right balance between computational savings and accuracy of solution.
Refinement criteria are problem-dependent and can affect the results of simulations.
The ASC/Alliances Center for Astrophysical Thermonuclear FlashesThe University of Chicago
Bibliography
T. F. M.
Fryxell et al., ApJS, 131 273 (2000)
Calder et al., in Proc. Supercomputing 2000, sc2000.org/proceedings
Calder et al., ApJS, 143 201 (2002)
Plewa and Muller, A&A, 142, 349 (1999)
Mignone, Plewa, and Bodo, ApJS, 160 199 (2005)
Powell et al. JCP, 154, 284 (1999)
Timmes & Arnett ApJS, 125, 294 (1999)
Malyshkin, Linde, & Kulsrud, Phys. Plasmas, 12 (10), 102902, 2005 astro-ph/0508094