On the limit between short and long GRBs

Abstract

Two classes of GRBs have been identified thus far without doubt and are prescribed to different physical scenarios-NS-NS or NS-BH mergers, and collapse of massive stars, for short and long GRBs, respectively. The existence of two distinct populations was inferred through a bimodal distribution of the observed durations T 90 , and the commonly applied 2 s limit between short and long GRBs was obtained by fitting a parabola between the two peaks in binned data from BATSE 1B. Herein, by means of a maximum likelihood (ML) method a mixture of two Gaussians is fitted to the datasets from BATSE, Swift, BeppoSAX, and Fermi in search for a local minimum that might serve as a new, more proper, limit for the two GRB classes. It is found that Swift and BeppoSAX distributions are unimodal, hence no local minimum is present, Fermi is consistent with the conventional limit, whereas BATSE gives the limit significantly longer (equal to 3.38 ± 0.27 s) than 2 s. These new values change the fractions of short and long GRBs in the samples examined, and imply that the observed T 90 durations are detector dependent, hence no universal limiting value may be applied to all satellites due to their different instrument specifications. Because of this, and due to the strong overlap of the two-Gaussian components, the straightforward association of short GRBs to mergers and long ones to collapsars is ambiguous.

Classification of gamma-ray burst durations using robust model-comparison techniques

“…However, the statistical significance of all the three tests is marginal, thus implying that both hypothesis cannot be easily distinguished. Our results also agree with the analysis of Tarnopolski (2015), who also compared χ 2 probabilities by carrying out a binned analysis of the duration distribution. He concluded that although the χ 2 probability for a 3-component fit is more than a 2-component one, the p-values range from 14-77% for different values of the binning.…”

“…In summary, we can say that the preferred model at first look is the k = 2 model for the intrinsic GRB case but its significance is low with all the model comparison tests used. When a similar analysis of the intrinsic T90 distribution of 347 Swift GRBs was done by Tarnopolski (2016b), he found that AIC points to three components (albeit with very weak evidence against two), and BIC yielded a very strong support for two components. The GMMbased analysis also showed evidence for two components with ∆BIC of about 6 (Zhang et al 2016b).…”

“…Evidence for a third lognormal component was also found in Swift/BAT data (Horváth et al 2008;Huja et al 2009;Horváth et al 2010) using duration and also from two-dimensional clustering using both duration and hardness (Veres et al 2010 Horváth & Tóth (2016), who pointed that three lognormal distributions provide a better fit to the data than two with 99.9999% confidence level. Tarnopolski (2016a) finds that for the similar Swift GRB dataset consisting of 947 GRBs, three groups are favored in the observer frame, whereas two in the rest frame. The possible physical origin of the third intermediate class of Swift GRBs is attributed to X-ray flashes (Veres et al 2010).…”

“…AIC and BIC have also been previously used for GRB classification by a number of authors (Mukherjee et al 1998;Tarnopolski 2016a,b;Yang et al 2016;Zhang et al 2016b). Frequentist model comparison after binning the data has been used by Zitouni et al (2015); Tarnopolski (2015). More information about AIC and BIC and its application to a variety of astrophysical problems can be found in Liddle (2004Liddle ( , 2007; Tan & Biswas (2012); Shi et al (2012);Shafer (2015).…”

Gamma-Ray Burst Prompt Correlations

“…On the other hand, this elusive intermediate class might be a statistical feature that can be explained by modeling the duration distribution with skewed distributions, instead of the commonly applied standard Gaussians [57][58][59][60][61]. Additionally, GRB classification was shown to be detector-dependent [1,62,63]. Moreover, a subclass classification of LGRBs was proposed [64], and Norris and Bonnell [65] discovered the existence of an intermediate class or SGRBs with extended emission, that show mixed properties between SGRBs and LGRBs.…”