Then get errors on photon index, gaussian energy, width, norm. for 10, 20, 50, 100 ks exposures.
2-. Grab point source arf, with optical blocking filter.
It would be interesting to try iterating through the different rmfs to see what impact those different designs have but start with the current best estimate one.
Models to start with would be
powerlaw + diskline, ignoring 0.0-2.0 and 10.-** (i.e., ignore below 2 keV since the disk line won't have any impact below something like 4 keV and you want to go below where the diskline is contributing strongly to the spectrum to make sure the powerlaw is being fit properly). Then fit this with the input diskline model, trying to unfreeze various parameters to see if they are being constrained. That is a qualitative statement but a quantitative version would be that the parameter errors don't hit boundaries that are built into the model. xspec lets you set a "soft" and a "hard" range for each parameter. I don't recall exactly what soft means but it impacts the fitting procedure in a way that the fit is steered away from the soft boundary. The hard boundary means that the fit parameter is not allowed to go past that point. So for photon index, the default hard boundary is -3 to 10 and the soft boundary is -2 to 9. You can set this when you do "newpar", e.g.,
newpar 1 2.0 0.01 -4 -3 10 15
changes the boundaries to -4 to 15 and -3 to 10.
The default ranges for parameters for xspec models are usually physically motivated and/or based on experience... it is very rare to see a photon index < -2 or > 9. Typical values are 0-3. So if the fit goes to -2 or -3 there is a good chance that something is wrong or that the power-law is really being used to fit something that isn't really a power-law.
4-You can also do the above with absorption (i.e., due to gas between the satellite and the source, again mostly due to gas in the Milky Way and the host galaxy of the source):
model phabs * (powerlaw + diskline)
5-Also try
phabs * apec
6-model phabs * (powerlaw + diskline)
-model phabs * apec # with different values of Nh,kT, and abundance
fakeit none
-Then fits first with the apec model itself and see how constraints on Nh, kT, A come out and then fit with phabs * brem and then add in lines one at time
model phabs * brem
fit
# get errors
7. Take a simple power-law only model, run fakeit a larger number times, get the error on photon index, then plot the histogram of photon index best-fit values. From that histogram see where the 90% of the value are.
