Are gamma-ray bursts the same at high redshift and low redshift?

The majority of Swift gamma-ray bursts (GRBs) observed at z > 6 have prompt durations of T90 < 30s, which, at first sight, is surprising given that cosmological time-dilation means this corresponds to < 5s in their rest frames. We have tested whether these high-redshift GRBs are consistent with being drawn from the same population as those observed at low-redshift by comparing them to an artificially red-shifted sample of 114 z < 4 bursts. This is accomplished using two methods to produce realistic high-z simulations of light curves based on the observed characteristics of the low-z sample. In Method 1 we use the Swift/BAT data directly, taking the photons detected in the harder bands to predict what would be seen in the softest energy band if the burst were seen at higher-z. In Method 2 we fit the light curves with a model, and use that to extrapolate the expected behaviour over the whole BAT energy range at any redshift. Based on the results of Method 2, a K-S test of their durations finds a ~1% probability that the high-z GRB sample is drawn from the same population as the bright low-z sample. Although apparently marginally significant, we must bear in mind that this test was partially a posteriori, since the rest-frame short durations of several high-z bursts motivated the study in the first instance.
Accepted for publication in MNRAS, 17 pages, 14 figures, 3 tables