Dead time is a common problem in the timing analysis of bright sources. For some dead time cases, analytical expressions exist (e.g. for the cospectrum), but often, these leave systematic biases in the data that can significantly alter the inference of certain parameters (e.g. the fractional rms amplitude) important to the physical interpretation of PSDs.
On the other hand, dead time is relatively simple to simulate, and it is thus easy and inexpensive to simulate dead time-affected periodograms of stochastic signals. Here, we introduce a method for accurate parameter inference using Approximate Bayesian Computation.
- Daniela Huppenkothen
- Matteo Bachetti
The NuSTAR observation used in this project is available at HEASARC. Pre-recorded simulations to reproduce the experiments on simulated data can be downloaded from Zenodo.