Classification of Swift's gamma-ray bursts

TL;DR

This study reanalyzes Swift gamma-ray burst durations, confirming the existence of three distinct groups with statistical significance, suggesting the need for models to explain these different burst types.

Contribution

It provides the first statistical confirmation of three gamma-ray burst classes in Swift data, differing from previous BATSE results, due to instrument sensitivity differences.

Findings

Three log-normal groups fit the data significantly better than two.

The relative frequencies are 7% short, 35% intermediate, 58% long.

Different instrument sensitivities explain the variation in group proportions.

Abstract

Two classes of gamma-ray bursts have been identified in the BATSE catalogs characterized by durations shorter and longer than about 2 seconds. There are, however, some indications for the existence of a third class. Swift satellite detectors have different spectral sensitivity than pre-Swift ones for gamma-ray bursts. Therefore, it is worth to reanalyze the durations and their distribution. We analyze, the maximum likelihood estimation, the bursts duration distribution, published in The First BAT Catalog, whether it contains two, three or more groups. The three log-normal fit is significantly (99.54% probability) better than the two for the duration distribution. Monte-Carlo simulations also confirm this probability (99.2%). Similarly, in previous results we found that the fourth component is not needed. The relative frequencies of the distribution of the groups are 7% short 35% intermediate and 58% long. Although the relative frequencies of the groups are different than in the BATSE GRB sample, the difference in the instrument spectral sensitivities can explain this bias on a natural way. This means theoretical models may be needed to explain three different type of gamma-ray bursts.