Gravitational-wave and Electromagnetic Signals From Neutron Star Collisions Gauss Centre for Supercomputing e.V.


Gravitational-wave and Electromagnetic Signals From Neutron Star Collisions

Principal Investigator:
Sebastiano Bernuzzi, Bernd Brügmann

Friedrich-Schiller-Universität Jena

Local Project ID:

HPC Platform used:
SuperMUC and SuperMUC-NG of LRZ

Date published:


This project supported the activity of the Computational Relativity (CoRe) collaboration [1] whose goal is to model the gravitational waves (GWs) and the electromagnetic (EM) signals from the coalescence of binary neutron star (NS) mergers using firstprinciples, 4D, nonlinear simulations in general relativity. Binary neutron star mergers are unique astrophysical laboratories to investigate the unknown matter equation of state (EOS) at extreme densities, and they are connected to the production of heavy elements in the Universe (via r-process nucleosynthesis in the mass ejecta) and to some of the most energetic EM transients.

Our simulations are crucial to interpret current and future observations of the multimessenger signals from this phenomenon. This project addressed some of challenges posed by the LIGO-Virgo observations of the binary neutron star signals GW170817 and GW190425. We simulated 50 binaries to compute the gravitational waves merger signals and analyzed the strong-field dynamics of the remnant. Several binary properties and new physical processes were explored for the first time: spin, eccentricity, large mass ratios, microphysical EOS, effects of neutrino transport and turbulent viscosity. The simulations’ data product extends the CoRe database of about 30%. The latter currently represents the largest public collection of waveform and mass ejecta available to the gravitationalwave and astrophysics communities.

Results and Methods

The key feature of our simulations is the detailed treatment of general relativity and dense matter using adaptive mesh refinement techniques to span the multiple scales of the problem. Our largest simulations run efficiently up to 2,000-4,000 cores for more than 2,000,0000 explicit timesteps in order to capture the merger dynamics from the orbital phase to the hydrodynamical and viscous timescales of the remnant. Moreover, these simulations have to be repeated at multiple resolutions to obtain error estimates on the final data, and for several parameters of the binary (masses, EOS, etc). The data generated from one simulation are snapshots of order of several TB, which are then reduced in (parallel) postprocessing steps. The simulations performed in this project used 75(+7.5)M core hours and represent one of the latest major simulation effort in the field of numerical relativity and GW astronomy.

The simulations conducted under pn56zo set a milestone for our research on GW modeling. We produced high-quality gravitational waveform at reduced eccentricity and for binaries with spins. These waveforms
are used for the development of templates for LIGO-Virgo analysis. The inclusion of spin effects is a distinctive feature of our simulation codes. We designed the first complete gravitational-wave spectrum model by combining numerical relativity data with semianalytical templates [2], Fig. 1. Our model will be relevant for advanced and third generation detectors observations. For example, the detection of a single GWs from a merger remnant will allow us to deliver a measurement of the NS minimal radius within the kilometer precision, thus providing unique constraints to the extreme density EOS.

Under pn56zo, we performed the first large set of BNS simulations including a general-relativistic scheme for turbulent viscosity induced by magnetic fields and a new neutrino transport scheme. We studied in detail the thermodynamics and geometrical properties of the remnant, and the winds that are emitted on timescales of hundreds of milliseconds after the merger.

Our simulations indicate that the bright, earlytime (optical and UV) kilonova transient AT2017gfo associated to GW170817 could be explained as the signature of weak r-process nucleosynthesis in a fast expanding wind. The latter is generated by spiral density waves propagating from the remnant to the outer disk, and can be identified only with ab-inito simulations [3], Fig.2. We demonstrated that the r-process nucleosynthesis in mass ejecta with speed, temperature and composition as found in the simulations, can robustly account for all the heavy elements from mass number 75 to actinides and is compatible with solar abundances. By exploring five different microphysical EOS and several binary mass ratios, we concluded that this is a very general feature in binary neutron star mergers and does not require fine tuning. Indeed, the merger outcome of binaries with comparable masses and sufficiently stiff EOS (as compatible with GW170817) is a remnant NS (instead of a black hole) with lifetime of several rotational periods during which the wind can develop.

Mergers with rapid black formation, like GW190425, are particularly interesting as they provide us a connection to the maximum NS mass. Using the simulations of pn56zo we developed an inference method
for LIGO-Virgo observations that, for the first time, quantifies the probability of prompt black hole formation
of the merger remnant. Applied to GW170817 and GW190425, the method supported the current astrophysical interpretation with a rigorous Bayesian inference analysis.

Using the resources of pn56zo we simulated, for the first time, mergers of binaries with mass ratio ~ 1.67-1.8 with microphysics. We discovered that the remnant undergoes accretion-induced prompt collapse [4]. In these mergers, the tidal disruption of the companion and its accretion onto the primary star determine prompt black hole formation. The tidal disruption event results in a black hole remnant with a massive and neutronrich disc around. This is very different from the equal-masses mergers, for which there is no  significant disc around the black hole, Fig.3. Challenging a common belief, our simulations showed that these accretion-induced promptcollapse mergers can power bright electromagnetic counterparts. We predicted that the mass ejected during tidal distruption can power a red, bright and temporally extended kilonova emission. The peculiar feature of this emission might be observed in the near future and can help to constrain the binary mass ratio from future multimessenger observations.

Ongoing Research / Outlook

The HPC resources provided by LRZ covered the entire computational resources for the research conducted by the Jena group in 2018-2019 and the largest fraction of computational resources of the CoRe collaboration. We published 7 peerreview papers based on these data solely and are further developing projects using them. The project has obtained an extension in 2021 (50M core hours). We publicly release the final data product of our simulations on the CoRe website [1] or on We also produced a number of visualizations that can be found on the CoRe YouTube channel [5]. Visualization of simulated data have also been used for the official LIGO-Virgo outreach material for announcement of GW190425 (January 2020, see mp4-video, below).

References and Links

[2] Breschi et al Phys. Rev., D100, no. 10, 104029, 2019.
[3] Nedora et al Astrophys. J., 886, no. 2, L30, 2019.
[4] Bernuzzi et al, MNRAS, 497, Issue 2, 2020.

Research Team

Sebastiano Bernuzzi (PI)1,Bernd Brügmann (PI)1, Matteo Breschi1, Vivek Chaurasia1, Tim Dietrich2, Vsevolod Nedora1, Nestor Ortiz1, Albino Perego3, David Radice4

1Friedrich-Schiller-Universität Jena
2Nikhef, Amsterdam
3Istituto Nazionale Fisica Nucleare, Milano Bicocca
4Princeton, USA

Scientific Contact

Prof. Dr. Sebastiano Bernuzzi
Friedrich Schiller University
Faculty of Physics and Astronomy
Institute for Theoretical Physics
Abbeanum, Fröbelstieg 1, D-07743 Jena (Germany)

NOTE: This report was first published in the book "High Performance Computing in Science and Engineering – Garching/Munich 2020 (2021)" (ISBN 978-3-9816675-4-7)

Local project ID: pn56zo

May 2021


Tags: LRZ Friedrich Schiller Universität Jena Astrophysics Large-Scale Project