Linear State Space Modeling of Gamma-Ray Burst Lightcurves

Linear State Space Modeling determines the hidden autoregressive (AR) process in a noisy time series; for an AR process the time series' current value is the sum of current stochastic ``noise'' and a linear combination of previous values. We present preliminary results from modeling a sample of 4 channel BATSE LAD lightcurves. We determine the order of the AR process necessary to model the bursts. The comparison of decay constants for different energy bands shows that structure decays more rapidly at high energy. The resulting models can be interpreted physically; for example, they may reveal the response of the burst emission region to the injection of energy.