Accepted to ApJ Preprint typeset using LATEX style emulateapj v. 5/2/11 UNCOVERING THE INTRINSIC VARIABILITY OF GAMMA-RAY BURSTS

We develop a robust technique to determine the minimum variability timescale for Gamma-ray Burst light curves, utilizing Haar wavelets. Our approach averages over the data for a given GRB, providing an aggregate measure of signal variation while also retaining sensitivity to narrow pulses within complicated time-series. In contrast to previous studies using wavelets, which simply define the minimum timescale in reference to the measurement noise floor, our approach identifies the signature of temporally-smooth features in the wavelet scaleogram and then additionally identifies a break in the scaleogram on longer timescales as signature of a true, temporally-unsmooth light curve feature or features. We apply our technique to the large sample of Swift GRB Gamma-ray light curves and for the first time – due to the presence of a large number of GRBs with measured redshift – determine the distribution of minimum variability timescales in the source frame. We find a median minimum timescale for long-duration GRBs in the source frame of ∆tmin = 0.5 s, with the shortest timescale found being on the order of 10 ms. This short timescale suggests a compact central engine (3 × 103 km). We discuss further implications for the GRB fireball model and present a tantalizing correlation between minimum timescale and redshift, which may in part be due to cosmological time-dilation. Subject headings: gamma-ray burst: general — methods: statistical, data analysi