Home > Research > Supernovae and Merger Publications > SciDAC4

SciDAC4 at Princeton: Towards Exascale Astrophysics of Mergers and Supernovae (TEAMS)-


Supernovae and Merger Publications
Presentations and Talks

Fornax Code Description

Fornax Code Paper

Fornax Code Tests

SciDAC4 TEAMS Initiative Summary:

We are the products of nearly 14 billion years of cosmic chemical evolution, stretching back to the fusion of newly formed neutrons and protons into helium and other light elements in the Big Bang. In the ninety six years since Eddington suggested that the billion-year brightness of stars could be powered by the transmutation of the elements, our understanding of cosmic chemical evolution has matured greatly. However, questions still remain, chief among these the source of the heaviest elements. As a result, the Committee on the Physics of the Universe convened by the National Research Council in 2000 included the question "How Were the Elements from Iron to Uranium Made?" in their list of Eleven Science Questions for the New Century. Burbidge, Burbidge, Fowler & Hoyle (1957; Rev. Mod. Phys., 29 547) and Cameron (1957; PASP, 69 201) divided the observed abundance of the isotopes heavier than iron between three distinct processes. On the basis of the correlation between nuclear properties and cosmic abundances, we know that roughly half of these isotopes are the result of a slow neutron capture process, termed the s-process. Based on well-measured nuclear data from stable nuclei and mature models of stellar evolution, we have considerable confidence that the s-process occurs over thousands of years in the hydrogen and helium-rich nuclear burning shells that sit atop the inert carbon-oxygen core in an Asymptotic Giant Branch star. This confidence is buoyed by observations of red-giant stars with large over abundances of s-process elements like barium and radioactive technetium. Most of the isotopes that the s-process can not explain result from a rapid neutron capture process, the r-process, while a scattering of rarer proton-rich isotopes are ascribed to a p-process, originally thought to be driven by proton capture.

In the cases of the r-process (and the p-process), we do not have well-measured nuclear data for the highly radioactive isotopes involved, mature models of the potential astrophysical sites, or conclusive direct observations. In part to address the nuclear uncertainties, the 2007 and 2015 Long Range Plans of the Nuclear Science Advisory Committee recommended continued support for current radioactive ion beam facilities and the construction of the Facility for Rare Isotope Beams. However, the nuclear data these accelerators, and their peers being operated and constructed around the world, will provide is alone not enough to correct our knowledge concerning the formation of these elements. We propose an effort to mature the modeling of many of the suggested astrophysical sites of the r-process, which will also reveal much about the nucleosynthesis of elements from oxygen to germanium. This effort is not independent of the experimental efforts, as calculations of the r-process and p-process rely on the nuclear data. However, the details of the nuclear data that are needed vary between the different potential sites. Winnowing the list of possible r-process or p-process sites reduces the nuclear data needs, allowing the simulations we propose herein to guide the experimental efforts in nuclear astrophysics.

The observed pattern of r-process species requires a combination of high-entropy and neutron-rich material, acting on a timescale of seconds, with the exact balance varying between the different proposed sites. Arguments based on the chemical evolution history of our galaxy call for as many as three sites to explain the development of the r-process abundance pattern over time; for instance, a weak (low atomic mass) r-process, possibly associated with ordinary core-collapse supernovae, a main r-process, sometimes associated with peculiar supernovae driven by magnetohydrodynamic jets and a fission-recycling (high atomic mass) r-process, thought to be associated with neutron star mergers. Simulations of each of these scenarios rely on a similar toolkit: magnetohydrodynamics, thermonuclear kinetics, the equation of state of nuclear matter, and radiation transport of neutrinos.

Under this SciDAC proposal, we will improve existing astrophysical application codes to take full advantage of the coming generation of exascale computers and develop new modules with greater physical fidelity that is only achievable on these machines. In parallel, we will improve the microscopic physics in the simulations, implementing improved nuclear equations of states and neutrino opacities and, where appropriate, approximating the impact of neutrino oscillations. The simulations we will run using these new/improved codes will begin with the final stages of stellar evolution and push outward the state-of-the-art in the modeling of many of the proposed sites of the r-process. Among the products of our simulations will be the finest estimates to date of the photonic, neutrino, gravitational wave and nucleosynthesis signatures of these cosmic explosions. We will also seek to quantify the uncertainty in these predictions. In particular, we will examine both the nuclear physics and astrophysics uncertainties in the production of r-process isotopes and other nucleosynthesis. The development of codes under this proposal will leverage developments supported by the Exascale Models of Stellar Explosions team of the DOE Exascale Computing Project. We also expect to leverage development funded at LANL by the Advanced Technology Development and Mitigation (ATDM) subprogram of the DoE Advanced Simulation and Computing (ASC) Program.

The neutrino pulse is the best diagnostic of the physics of collapse, explosion, and neutron star/black hole birth. The neutrino spectra, flavor mix, and time evolution are stamped with the various hydrodynamic phases through which the core and shock wave progress. The detailed characterization of the neutrino burst would provide ground truth for any numerical model and would be a central means for code and theory validation. Using recently generated two-dimensional (2D) and one-dimensional (1D) multi-group radiation/hydrodynamic models employing the code Fornax (Burrows et al. 2017; Radice et al. 2017, Skinner et al. 2016; Wallace et al. 2016; Vartanyan et al. 2017; Skinner et al. 2018), we focus on the general overarching features of the neutrino light curve in Super-K, DUNE, JUNO, and ICE CUBE during approximately the first second after bounce as a function of progenitor mass and between exploding and non-exploding models. The first second is the most dynamical phase. We also explore the potential identification of features in the accreted density structure, such as the silicon/oxygen interface, and estimate the effects of neutrino oscillations on the detected signals (Mirizzi et al.2016; Scholberg 2017). We use the 10, 16, 17, 19, and 21 solar-mass progenitor models of Sukhbold et al. (2016), as evolved in Radice et al. (2017), Burrows et al. (2017), and Vartanyan et al. (2017), for our representative progenitor model suite. The 2D variants of these models explode (Vartanyan et al. 2017; Burrows et al. 2017; Radice et al. 2017), while the spherically-symmetric 1D models do not. We use this difference to gauge the difference between exploding and non-exploding models and find that after explosion the neutrino event rate is diagnostically very different between them. Also, we can identify in many of the non-exploding models a clear signature of the accretion of the silicon/oxygen interface. Moreover, we see other characteristic timescales that can readily be discerned, as well as clear differences in all detectors between event-rate models in the various detectors for non-oscillation, normal-hierarchy, and inverted-hierarchy realizations.

TEAMS Proposal

Updated on: June 19, 2018