You are quite right and I think this point is under-appreciated. Even a very
modest amount of pile-up can skew a fit and you can forget any attempt to
learn about the high-energy behaviour of the spectrum.
The method that Pat alluded to in his message was developed by John Davis at
MIT and is released in their ISIS package. I have included an implementation
of John's algorithm in the new version of XSPEC (11.1) which was made public
this week. I would be interested in having people test this. This algorithm
does not provide a complete solution in that a simple parametrization is
adopted to handle grade migration (ie when two or more photons pile-up they
may be assigned a worse grade than the individual events would have been).
The only person I know who has done a complete solution to pile-up is George
Chartas at PSU who analyzed the Chandra focus data using XSPEC, MARX, and
their ACIS simulator, all running in a big loop.
This archive was generated by hypermail 2b29 : Sat May 25 2013 - 01:00:04 EDT