Statistical study of observed and intrinsic durations among BATSE and Swift/BAT GRBs

Abstract

Studies of \textit{BATSE} bursts \citep{kouveliotou:93} have resulted in the widespread adoption of a two-group categorization: long bursts (those with durations $\geq 2$ seconds) and short bursts (those with durations $\leq 2$ seconds). This categorization, one must recall, used the observed $T_{90}$ time durations for bursts (during which 90\% of a burst's fluence is measured). In this work, we have explored two ideas: 1) a statistical search for a possible third, intermediate category of bursts (between the "short" and the "long" ones) among 2041 \textit{BATSE} GRBs and 757 \textit{Swift/BAT} ones, 2) a study of bursts' intrinsic durations, where durations in the bursts' reference frames (instead of the observed durations) are considered, for this, 248 \textit{Swift/BAT} bursts that have redshift measurements were statistically analyzed for the same categorization goal. We first use a Monte Carlo method to determine the proper binning of each GRB, considering that bursts come with different uncertainties on their durations. Then, using the method of minimization of chi-square $\chi^2$, we search for the best fit of the normalized frequency distributions $\frac{1}{N_0}\frac{dN}{d\ln{T}}$ of durations, this allows us to compare fits with two groups ("short" and "long") with fits with three groups ("short", "long", and "intermediate"). Our results indicate that the distributions of observed durations are better fitted by three groups than two groups for \textit{Swift/BAT} data, interestingly, the "intermediate" group appears rather clearly for both observed and intrinsic durations. For BATSE data, the statistical test does not prefer three groups over two. We discuss the results, their possible underlying causes, and reasonable interpretations.