XSPEC session thread

EPIC

1. Start XSPEC
   
   xspec

2. Load spectrum data:
   
   data 1 PNspectrum.pi

3. Load the redistribution matrix
   
   response 1 PN.rmf

4. Load the ancillary file
   
   arf 1 PN.arf

5. Load the background spectrum
   
   backgrnd 1 PNspectrum_background.pi

6. Create the output graphic plotting window
   
   cpd /xs

7. Define how data has to be plotted, e.g. in energy bins (wavelength would be a valid alternative):
   
   setplot energy

8. Relax getting a look into your data
   
   plot data
   

   
   

   
   Fig.1: XSPEC data plot

9. Ignore bad channels in data:
   
   ignore bad

10. Ignore energy intervals where the pn matrices are currently not calibrated (e.g.: beyond the energy band 0.2-15.0 keV)
    
    ignore **-0.2 15.-**

11. Load a model to be fitted (in this case a simple power law model with photoelectric absorption)
    
    model wabs*powerlaw
    
    (you will have to hit return several times for every parameter of the model. The following screen will appear:

     mo = wa((po)).
      Model:  wabs[1]( powerlaw[2] )
    Input parameter value, delta, min, bot, top, and max values for ...
    Current:           1     0.001         0         0     1E+05     1E+06
    wabs:nH>
    Current:           1      0.01        -3        -2         9        10
    powerlaw:PhoIndex>
    Current:           1      0.01         0         0     1E+24     1E+24
    powerlaw:norm>
      ---------------------------------------------------------------------------
      ---------------------------------------------------------------------------
      Model:  wabs[1]( powerlaw[2] )
      Model Fit Model Component  Parameter  Unit     Value
      par   par comp
        1    1    1   wabs       nH       10^22      1.000     +/-      0.
        2    2    2   powerlaw   PhoIndex            1.000     +/-      0.
        3    3    2   powerlaw   norm                1.000     +/-      0.
      ---------------------------------------------------------------------------
      ---------------------------------------------------------------------------
     Chi-Squared =     2.7158111E+09 using  1675 PHA bins.
     Reduced chi-squared =      1624289.     for   1672 degrees of freedom
     Null hypothesis probability =    0.
    

12. Fit the parameters
    
    fit 100 1e-1

    The former number indicates the maximum number of iterations before the minimization routine stops. The latter number indicate the minimum difference in chi-squared between two minimization runs, before the routine stops

13. set the rebinning of your choice (for plotting purposes only!)
    
    setplot rebin 3 4096

    The former number indicates the maximum number of sigma to be accumulated in a rebinned channel. The latter, the maximum number of channels to bu summed.

14. plot data with model adjustment and e.g. the residuals  (alternatives, ratio data/model, absolute chi-squared, etc)
    
    plot data residuals

    
    

    
    Fig.2: XSPEC data, model and residual plot

15. determine one-side errors on the best-fit parameters
    
    error 2.706 1 2 3

    The first number indicates the delta chi-squared value.

RGS

1. Start XSPEC
   
   xspec
   
   
2. Load spectrum data (in this case RGS1 background subtracted spectrum, first order) into first internal dataset:
   
   data 1:1 /xvsas05/sasval/data/procred/Mkn421/P0099280201R1S001SRSPEC1001.FIT
   
   
3. Load the response matrix into the first internal response set. The SAS task rgsrmfgen put both response and ancillary data into the response matrix file.
   
   resp 1 /xvsas05/sasval/data/procred/Mkn421/P0099280201R1S001RSPMAT1001.FIT
   
   
4. Loading a second dataset into second internal data position (in this case data corresponding to RGS2):

data 2:2 /xvsas05/sasval/data/procred/Mkn421/P0099280201R2S002SRSPEC1001.FIT

5. Load the corresponding response matrix:
   
   resp 2 /xvsas05/sasval/data/procred/Mkn421/P0099280201R2S002matrix1001.FIT

   A few differences need to be noticed at the data loading level

   • rgsrmfgen (the SAS tasks to generate RGS response matrices, which is by default automatically launched as the last stage of rgsproc, create a response matrix, which include both the redistribution matrix and the effective area vector (ancillary file). There is therefore no need or loading any *.arf files in XSPEC when analyzing RGS spectra
   • RGS spectra are already by default created background-subtracted. There is therefore no need to load a background file
   • the syntax data n:n allows to keep the fit parameter (fully or partly) independent between the two datasets during the fitting procedure. The alternative syntax data n forces the same model parameters to be fitted to all datasets

6. Create the output graphic plotting window
   
   cpd /xs

7. Define how data has to be plotted, e.g. in energy bins (wavelength would be a valid alternative):
   
   setplot energy

8. Relax getting a look into your data
   
   plot data
   

9. Ignore bad channels in data:
   
   ignore bad

10. Ignore energy intervals where the pn matrices are currently not calibrated (e.g.: beyond 1.9 keV)
    
    ignore 1.9-**

11. Load a model to be fitted (in this case a simple power law model)
    
    model powerlaw
    
    (you will have to hit return several times for every parameter of the model)

12. Fit the parameters
    
    fit 100 1e-1

    The former number indicates the maximum number of iterations before the minimization routine stops. The latter number indicate the minimum difference in chi-squared between two minimization runs, before the routine stops

13. set the rebinning of your choice (for plotting purposes only!)
    
    setplot rebin 3 4096

    The former number indicates the maximum number of sigma to be accumulated in a rebinned channel. The latter, the maximum number of channels to bu summed.

14. plot data with model adjustment and e.g the residuals  (alternatives, ratio data/model, absolute chi-squared, etc)
    
    plot data residuals

Notice that the help system in XSPEC is started with help, and works similarly to the old good VMS help system.