Surrogate light curve models for kilonovae with comprehensive wind ejecta outflows and parameter estimation for AT2017gfo

Open Access

Surrogate light curve models for kilonovae with comprehensive wind ejecta outflows and parameter estimation for AT2017gfo

Atul Kedia, Marko Ristic, Richard O'Shaughnessy, Anjali B. Yelikar, Ryan T. Wollaeger, Oleg Korobkin, Eve A. Chase, Christopher L. Fryer, and Christopher J. Fontes

Phys. Rev. Research 5, 013168 – Published 13 March 2023

Abstract

The electromagnetic emission resulting from neutron star mergers have been shown to encode properties of the ejected material in their light curves. The ejecta properties inferred from the kilonova emission has been in tension with those calculated based on the gravitational wave signal and numerical relativity models. Motivated by this tension, we construct a broad set of surrogate light curve models derived for kilonova ejecta. The four-parameter family of two-dimensional anisotropic simulations and its associated surrogate explore different assumptions about the wind outflow morphology and outflow composition, keeping the dynamical ejecta component consistent. We present the capabilities of these surrogate models in interpolating kilonova light curves across various ejecta parameters and perform parameter estimation for AT2017gfo both without any assumptions on the outflow and under the assumption that the outflow must be representative of solar $r$ -process abundance patterns. Our parameter estimation for AT2017gfo shows these surrogate models help alleviate the ejecta property discrepancy while also illustrating the impact of systematic modeling uncertainties on these properties, urging further investigation.

Received 8 November 2022
Accepted 20 January 2023

DOI:https://doi.org/10.1103/PhysRevResearch.5.013168

Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.

Published by the American Physical Society

Physics Subject Headings (PhySH)

Explosive burning Gravitational waves Nuclear physics of explosive environments Nucleosynthesis in explosive environments r process

Neutron stars & pulsars

Astrophysical electromagnetic fields

Astrophysical & cosmological simulations

Gravitation, Cosmology & AstrophysicsParticles & Fields

Authors & Affiliations

Atul Kedia ¹, Marko Ristic ¹, Richard O'Shaughnessy ¹, Anjali B. Yelikar ¹, Ryan T. Wollaeger ^2,3, Oleg Korobkin ^2,4, Eve A. Chase ^2,5, Christopher L. Fryer ^2,3,6,7,8, and Christopher J. Fontes ^2,9

¹Center for Computational Relativity and Gravitation, Rochester Institute of Technology, Rochester, New York 14623, USA
²Center for Theoretical Astrophysics, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA
³Computer, Computational, and Statistical Sciences Division, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA
⁴Theoretical Division, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA
⁵Intelligence and Space Research Division, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA
⁶University of Arizona, Tucson, Arizona 85721, USA
⁷Department of Physics and Astronomy, University of New Mexico, Albuquerque, New Mexico 87131, USA
⁸George Washington University, Washington, DC 20052, USA
⁹Computational Physics Division, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA

Article Text

Click to Expand

References

Click to Expand

Issue

Vol. 5, Iss. 1 — March - May 2023

Subject Areas

Reuse & Permissions

Author publication services for translation and copyediting assistance advertisement

Images

Figure 1
Illustrating off-sample interpolations for two morphologies. Top panel: The bottom panel from Fig. 4 of Ref. [32] showing interpolation of various LCs versus time for the TPwind2 morphology adopted in that paper. Different colors denote different filter bands, described in the legend. The dashed lines show full simulation output for each band. The colored points show our interpolated magnitude predictions at the evaluation times. The simulated parameters and viewing angle for this configuration are ( $m_{d} / M_{⊙}$ , $v_{d} / c$ , $m_{w} / M_{⊙}$ , $v_{w} / c$ ) = (0.097, 0.198, 0.084, 0.298) at $0^{\circ}$ [32]. Bottom panel: As above, but for the TSwind2 model morphology. The shaded region surrounding each solid curve shows our estimated GP fitting uncertainty at each time. The differences between these two panels illustrate the impact of outflow morphology on our results.
Reuse & Permissions
Figure 2
Wavelength interpolation using our surrogate model for an off-sample wavelength bands CFHT (1451 nm, i.e., between the $J$ - and $H$ -bands) and F182M (1839 nm, between $H$ - and $K$ -bands), generated for the TPwind1 model with parameters ( $m_{d} / M_{⊙}$ , $v_{d} / c$ , $m_{w} / M_{⊙}$ , $v_{w} / c$ ) = (0.014, 0.183, 0.085, 0.053) viewed along the symmetry axis $θ = 0$ . The dashed curves (red for CFHT and green for F182M) along with their respective bands indicate the predicted LC from our surrogate model and solid curves indicate the LC generated from the code SuperNu via the mechanism explained in Sec. 2a.
Reuse & Permissions
Figure 3
Left panel: Illustration of systematics in modeling ejecta from binary mergers, compared to a fiducial inference of those ejecta parameters. Black curves are deduced from inferences on GW170817 mass estimates combined with different models for the ejecta; for example the black solid line is derived from Eqs. (1, 2, 3). To isolate the effect of ejecta modeling systematics, all predictions are shown under the optimistic assumption of a perfectly known nuclear equation of state, here APR4 [70], without adding additional uncertainties to account for estimated fitting errors in the relations between ejecta and binary parameters. While the ejecta predictions for GW170817 are relatively consistent with one another, they are in tension with our interpretation of the GW170817-associated kilonova's ejecta assuming a TPwind2 outflow form. Curves correspond to 90% credible regions. Right panel: Parameter inference for LC based on EM paper prior (uniform in velocities, log-uniform in masses) [32] (blue-solid) and $r$ -process prior [69] (green-dashed), but allowing for relative systematic uncertainty $exp (\pm 0.3)$ in both ejecta masses. The inner and outer curves on the two-dimensional plots correspond to 68% and 95% credible regions. The ejecta estimate illustrated from GW includes no systematics, either from uncertainty in the EOS or from intrinsic uncertainty in the estimate itself; compare to Fig. 6.
Reuse & Permissions
Figure 4
Angular dependence for the three surrogate models for each morphology and composition combination, i.e., TPwind1, TSwind1, and TSwind2. This figure compares the $g$ -, $y$ -, and $K$ -band luminosity (top to bottom) at select times as a function of viewing angle. Different colors indicate different extraction times. Points show simulation AB magnitude results vs angle $θ$ , while the solid curves and shaded regions show our prediction and its expected (statistical interpolation) uncertainty. All simulations and surrogate LCs use the parameters ( $m_{d} / M_{⊙}$ , $v_{d} / c$ , $m_{w} / M_{⊙}$ , $v_{w} / c$ ) = (0.097, 0.198, 0.084, 0.298); because this configuration has $v_{w} > v_{d}$ (i.e., wind outside the dynamical ejecta), the wind ejecta's emission and angular dependence could dominate trends versus angle. Left column: Angular dependence for TPwind1 light bands. This model shows the largest viewing angle dependence. Center column: Angular dependence for TSwind1. A change in angular dependence is highly apparent at late times particularly in the $g$ band. Right column: Angular dependence for TSwind2. Differences from the first two models are apparent. This configuration shows the weakest angular dependence in its emission and the brightest emission.
Reuse & Permissions
Figure 5
Posterior predicted LCs for AT2017gfo, compared with AT2017gfo observations. Points and error bars denote the observations, with each color denoting a different filter band. Following [32], the solid curves and shaded intervals show the median and 90% credible expected LC, deduced by fitting our models to these observations. The left panel of figures denotes our unconstrained models; the right panels require each outflow to be consistent with Solar $r$ -process abundance (i.e., we also adopt a prior on $m_{w} / m_{d}$ such that the ejected material; see Sec. 2e). First row: TPwind1; second row: TPwind2 [32]; third row: TSwind1; fourth row: TSwind2.
Reuse & Permissions
Figure 6
Morphology and composition dependent parameter inference. This figure shows parameter inference for ejecta properties ( $m_{d} / M_{⊙}$ , $v_{d} / c$ , $m_{w} / M_{⊙}$ , $v_{w} / c$ ) derived from GW170817 and AT2017gfo. Colored solid and dotted curves show purely electromagnetic inferences derived using different families of two-component LC surrogate models and a prior on source inclination. Different solid colors indicate different choices of outflow and composition with a uniform logarithmic prior on ejecta masses, while colored dotted curves show inferences with the corresponding morphology/composition pairs while also requiring consistency with the Solar $r$ -process abundance pattern. The black solid and dotted curves by contrast show three estimates inferred solely from GW measurements, combined with forward models for the outflow. The legend indicates the morphology pair (e.g., TP = Torus and Peanut), while the integer indicates the wind composition (1 refers to the wind1 sequence and so on). The inferred wind (and thus total) mass depends substantially on the assumed wind composition and morphology. The two-dimensional contours correspond to 90% credible intervals.
Reuse & Permissions

Physical Review Research