SphericalNR: A new framework for post-merger simulations of binary neutron stars

Vassilios Mewes

National Center for Computational Sciences and Physics Division, Oak Ridge National Laboratory

The new era of gravitational wave multi-messenger astrophysics began with the recent detection of the binary neutron star merger GW170817. Our theoretical understanding of these systems relies on high fidelity numerical relativity simulations including general relativistic magnetohydrodynamics, realistic equations of state for matter up to nuclear densities, and neutrino radiation hydrodynamics. The approximate symmetries of the post-merger stage of the evolution, namely hypermassive neutron stars and black hole torus systems, make spherical coordinates better suited than Cartesian coordinates for the numerical modelling of these systems. This seminar will present SphericalNR, a new framework within the publicly available Einstein Toolkit to numerically solve the Einstein field equations of general relativity coupled to the equations of general relativistic magnetohydrodynamics in spherical coordinates without symmetry assumptions. A description of a reference metric approach together with algorithmic details enabling the use of spherical coordinates in the originally Cartesian code base of the Einstein Toolkit will be presented, followed by a description of ongoing algorithmic and code development work regarding a double FFT filter with the aim to alleviate the extremely severe timestep restrictions when solving hyperbolic PDEs in spherical coordinates with high angular resolutions. The outlook will touch upon future development for SphericalNR, focusing on extending the multi-physics capabilities of the framework, as well as challenges for increasing the parallel efficiency of the code with a view on the upcoming exascale era of HPC.