The detector response maps energies to events and their pulse invariant amplitude (charges). rmfgen calculates a pile-up-corrected detector response matrix through tracing simulated charges that are added to the observed data through the pn event reconstruction process.
Charge(s) for given energies are simulated by randomizing over the detector response, over the point spread function and over the pattern distribution. These simulated charge(s) are added to the frames which contain the observed charges. rmfgen then applies the event reconstruction algorithm to each frame. Comparing the results of the event reconstruction for the frames with and without added simulated charge(s) allows mapping the energies against events and their pulse invariant amplitude. Such simulations are performed many times for all input energies and for all observed frames. To generate the detector response matrix a sub-set of the traced charges is used such that the distribution of number of events in the frames with simulated charges equals the distribution of number of events in the observed frames, i.e. the count rate corresponding to the frames used to generate the detector response equals the observed count rate.
