Our primary scientific and computational focus is on tera- to exa-scale simulation of supernovae of both classes in the Universe.
Core collapse supernovae are the death throes of massive stars, more than 8-10 times the mass of our sun. They are a dominant source of elements in the Universe, without which life would not be possible. Our group is focused on ascertaining the core collapse supernova mechanism - i.e., how the explosions of these stars are initiated. Core collapse supernovae are three-dimensional, multi-physics events. Three-dimensional general relativistic radiation magnetohydrodynamics simulations must be performed to ascertain definitively the supernova mechanism and to predict all of the associated supernova observables. Core collapse supernovae are driven by neutrinos (radiation) and perhaps magnetic fields. Thus, our group has developed discretizations, solution algorithms, and codes for the solution of the multi-dimensional neutrino (radiation) transport equations and the three-dimensional magnetohydrodynamics and Poisson equations, the latter for the star's self gravity. This is ultimately a petascale to exascale computational problem.
Members of our group are also engaged in research on Type Ia supernovae - in particular, the mechanism whereby white dwarf stars (the endpoint of stellar evolution for stars less than 8 times the mass of the Sun) end their lives in stellar explosions as well. Understanding these stellar explosions is particularly important in light of the fact they provide the means to probe the evolution of the Universe as a whole. Indeed, observations of Type Ia supernovae and conclusions based on them have led to the startling fact our universe is expanding at an accelerated rate, which has significant implications for its future and ultimate fate. Type Ia supernovae are driven by thermonuclear runaway. The challenge here is the modeling of a turbulent flame, centimeters thick, in a white dwarf star the size of the Earth. Thus, three-dimensional simulations of chemically reactive flows are required, with realistic sub-grid models.
Our group has been engaged in the development of (1) physics-based preconditioners for the solution of the large, sparse linear systems of equations underpinning the numerical solution of the neutrino Boltzmann kinetic equations, in collaboration with the CSMD Applied Mathematics Group (Ed D'Azevedo), (2) strategies for successful tera- to peta-scale data management (including parallel I/O), (3) visualization of multi-dimensional scalar, vector, and tensor supernova data, in close collaboration with the NCCS Visualization Task (Ross Toedte, Sean Ahern), and (4) strategies for adaptive mesh refinement for memory-intensive applications involving radiation transport.
A code for fully general relativistic neutrino (radiation) hydrodynamics for spherically symmetric stellar collapse and core collapse supernova simulation. Agile-BOLTZTRAN couples the general relativistic neutrino Boltzmann kinetic equations to the general relativistic hydrodynamics equations and exact Einstein equations for gravity. It also implements a detailed nuclear equation of state for the nucleons, nuclei, electrons, positrons, and photons in the stellar core (Lattimer-Swesty), as well as a realistic set of neutrino interactions with these stellar core constituents.
Agile-BOLTZTRAN is our platform for high-quality, detailed one-dimensional simulations of stellar core-collapse and supernovae. Agile is a spherically symmetric, adaptive-mesh hydrodynamics code with a complete treatment of general relativity. Agile is coupled to BOLTZTRAN, a spherically symmetric, multi-flavor, multi-group, discrete ordinates Boltzmann neutrino transport code with detailed neutrino-matter interactions. Agile-BOLTZTRAN remains a valuable tool for problems where it's strengths (full general relativity and Boltzmann neutrino transport) outway it's primary weakness (the assumption of spherical symmetry). We make significant use of Agile-BOLTZTRAN as a verification standard for mCHIMERA and improvements to BOLTZTRAN also benefit the eventual bCHIMERA development. Source - http://astro.phys.utk.edu/activities:boltztran
A code that couples multigroup flux-limited diffusion neutrino transport (a sophisticated approximation of Boltzmann transport) along radial rays (the ray-by-ray-plus approximation) to three-dimensional hydrodynamics, a nuclear burning network, Newtonian self gravity with a spherical general relativistic correction, an industry standard nuclear equation of state (Lattimer-Swesty, Shen, Wilson), and with state of the art neutrino interactions. Two-dimensional multi-physics simulations of core collapse supernovae have been performed with CHIMERA (see News and Highlights) and three-dimensional simulations are underway.
Unleashing CHIMERA: Multidimensional Supernova Simulations
Oct 02, 2008
Core-collapse supernovas, stars whose iron cores exceed the Chandrasekhar mass and implode under their own weight, litter the universe with most of the elements in the periodic table—all of the gold in California is the result of their demise. They are also prodigious sources of neutrinos, gravitational waves, and photons across the electromagnetic spectrum and lead to the formation of neutron stars and black holes. Despite their importance in the universe, however, the mechanism responsible for the explosion of these supernovas is not fully understood. Studying these phenomena will provide researchers with information outside the bounds of traditional, visual observation. A team led by Oak Ridge National Laboratory's Bronson Messer will apply Kraken's petascale computing power to more closely study these stellar events and determine if in fact the neutrino-driven mechanism is a viable explanation for the explosion. Messer and his team will use CHIMERA, a massively parallel multiphysics code especially designed to simulate core collapse supernovas and the only three-dimensional code that accounts for the differing energies of neutrinos, necessary in deciphering one of astrophysics' greatest remaining mysteries. Kraken's raw computing power and its ability to run jobs at scale will provide the perfect setting for the achievement of this goal. http://www.nics.tennessee.edu/category/astrophysics/unleashing-chimera-multidimensional-supernova-simulations
A code that couples six-dimensional (three space, three momentum space) special relativistic Boltzmann neutrino transport to three-dimensional special relativistic magnetohydrodynamics, Newtonian self gravity, a nuclear equation of state, and state of the art neutrino opacities. Two-dimensional, multi-physics simulations of core collapse supernovae will be performed later this year. Three-dimensional simulations with GenASiS will require petascale to exascale computing resources.
GenASiS: AMR code for True 2D/3D Neutrino Transport
To study the mechanism of core-collapse supernova with true 2D/3D neutrino transport, we are developing a new adaptive mesh refinement (AMR) code that couples six-dimensional (three space, three momentum space) special relativistic Boltzmann neutrino transport to three-dimensional special relativistic magnetohydrodynamics, Newtonian self gravity, a nuclear equation of state, and state of the art neutrino opacities.
This code - called GenASiS, for General Astrophysical Simulation System - is designed for modularity and extensibility of the physics. Presently in use or under development are capabilities for Newtonian self-gravity, Newtonian and special relativistic magnetohydrodynamics (with 'realistic' equation of state), and special relativistic energy- and angle-dependent neutrino transport - including full treatment of the energy and angle dependence of scattering and pair interactions. Early version of GenASiS' neutrino transport solvers have been subjected to almost every transport test problem we can ﬁnd in the literature, including some we fashioned.
Written in Fortran 95 and designed with object-oriented paradigm, GenASiS is still currently under heavy development. http://astro.phys.utk.edu/activities:genasis
Info to follow.