Old data, new forensics: The first second of SN 1987A neutrino emission

The next Milky Way supernova will be an epochal event in multimessenger astronomy, critical to tests of supernovae, neutrinos, and new physics. Realizing this potential depends on having realistic simulations of core collapse. We investigate the neutrino predictions of modern models (1-, 2-, and 3-D) over the first $\simeq 1\mathrm{s}$, making the first detailed comparisons of these models to each other and to the SN 1987A neutrino data. Even with different methods and inputs, the models generally agree with each other. However, even considering the low neutrino counts, the models generally disagree with data. What can cause this? We show that neither neutrino oscillations nor different progenitor masses appear to be a sufficient solution. We outline urgently needed work.

Figure 1

Neutrino (${\overline{\nu }}_e$; others shown in Appendix pp3) luminosity and RMS energy profiles from supernova simulations [18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28].

Figure 2

Predicted counts and average energies of supernova models (colors as in Fig. 1), compared to each other and to SN 1987A data. Our main calculations are in the inset.

Figure 3

Similar to Fig. 2, with the theory predictions from the alcar 3-D model for $20M_{\odot }$ [27] and different oscillation scenarios. The simulation cutoff time is 0.68 s. The gray symbols show the p-values from other models (see Appendixes pp7 and pp8).

Figure 4

Similar to Fig. 2, with the theory predictions corresponding to different progenitors from Fornax 2-D models [36], with no oscillations included.

Figure 5

Neutrino luminosities (top) and RMS energies (bottom) of ${\nu }_e$, ${\overline{\nu }}_e$, and ${\nu }_x$ from numerical simulations. The neutronization burst luminosity goes up to $\simeq 5\times {10}^{53}\mathrm{erg}/\mathrm{s}$. The ${\nu }_x$ luminosity is shown for only one of the four flavors. The two PNS cooling models are from Refs. [40, 92].

Figure 6

$\sqrt{⟨E_{\nu }^2⟩}$ as a function of $⟨E_{\nu }⟩$ obtained from the numerical simulations where both are provided. The black line is the relation we use in this work.

Figure 7

Left: the black dots represent the energy uncertainty for each event detected in Kam-II, whereas the continuous line is a fit to this data; we use the fit to get the width of the Gaussian energy resolution function defined in the text. Right: the efficiency of Kam-II as a function of the true positron total energy. Note that our calculation perfectly overlaps the line provided by Kam-II.

Figure 8

Same as Fig. 7, but for IMB, which only provides efficiencies at a few energy points.

Figure 9

A demonstration of p-value calculation. Left: the cumulative energy distribution of the IDSA 2-D model and 1987A data and the corresponding test statistic. Right: the test statistic distribution.

Figure 10

Cumulative counts (top panel) and cumulative spectra (bottom panel) of models compared to SN 1987A data. The green shading shows the density of all models considered in this work, including the alternative scenarios with neutrino oscillation and different progenitor masses.

Figure 11

Comparison of supernova models: the 1-, 2-, 3-D models used in our work (purple lines; see Fig. 1) and the 1-D models used in Ref. [32] (orange lines, corresponding to various PNS masses). From bottom to top at a time of 1.0 s, these are 1.36, 1.44, 1.62, 1.77, and $1.93M_{\odot }$; we show the $1.93M_{\odot }$ case with a thick line. These correspond to progenitor masses of 9.0, 18.8, 18.6, 27, and $20M_{\odot }$ (note that the order is nonmonotonic); we show the $20M_{\odot }$ case with a thick line. Note that in their fits they emphasize models with substantially lower neutrino luminosities (and somewhat lower average energies) than found in the multi-D models we consider, either because their models employ progenitors that give intrinsically lower luminosities or because they are artificially cut off accretion near 0.6 s by their explosion prescription.

Figure 12

Cumulative numbers of events for different progenitors simulated with FLASH 1-D [82].

Figure 13

The p-values obtained through an energy spectrum comparison with Kam-II data for different progenitors masses, simulated with FLASH [82], vertex [22], and Fornax [28, 36, 78, 79, 80, 81], with a cutoff time of 0.5 s (left) and 1.3 s (right).

Figure 14

Different oscillation scenarios for 3-D simulations.

Figure 15

Different oscillation scenarios for 2-D simulations.

Figure 16

Different oscillation scenarios for 1-D simulations.

Figure 17

Figure 16 continued.