General Keywords

   

There are three modes for modeling a spectrum: to CALCULATE a model spectrum, COMPARE a model spectrum with an observed one, or to FIT a model to an observed spectrum.

     

CALCULATE mode: calculate a flux or count rate spectrum. All model parameters have to be fixed (see PARAMETER below). COMPARE mode: compare an observed count rate or photon flux spectrum with a model. All model parameters are either in a fixed or in a grid mode (see PARAMETER below). FIT mode: Fit an appropriate model to an observed count rate or photon flux spectrum spectrum. At least one of the model parameters has to be either in the mode free, interesting or uninteresting.

FIT      ! alternative: COMPARE, CALCULATE

The modeling modes given above can be used either for a photon flux or a count rate spectrum by FOLDing the photon flux with the detector response.  

FOLD     ! omit, if no folding with a detector response is required

The next four keywords are used only in the FIT mode: With ITERATIONS the maximum number of iterations steps can be chosen (default: 30 iterations).

ITERATIONS        30      ! max_number of iteration steps

  CONVERGENCE_LIMIT is a fractional epsilon, at which the change of the relative reduced chi-square stops the iteration procedure (default: 0.0001).

CONVERGENCE_LIMIT 0.0001  ! condition to stop iteration

  Five CONFIDENCE_LEVELS can be chosen for error calculation of the fitted parameters. The default values of 68.33, 95.45, 99.73, 99.9936, 99.999942% correspond to 1 to 5 standard deviations.  

CONFIDENCE_LEVELS 68.33 95.45 99.73 99.9936 99.999942 ! %

There are two possibilities to manipulate measured errors: a SYSTEMATIC_ERROR and/or a STATISTICAL_ERROR can be added linear or quadratic to the measured error of all data, respectively (defaults = 0).  

SYSTEMATIC_ERROR 0.01  ! relative error, linearly added
                         to the error of all data points

 

STATISTICAL_ERROR 0.01  ! relative error, quadratically added
                          to the error of all data points

  The STATISTIC_MODE determines the merit function, which has to be minimized: chi-square (default mode) or maximum_likelihood.      

STATISTIC_MODE chi-square ! altern.: maximum_likelihood

If the data file contains more than one observed spectrum (up to 250 DATASETs are allowed), the whole set or a defined subset can be processed:

DATASET start_no end_no step_no (Not yet tested.)

DATASET   1 1 1  ! defaults for start_no end_no step_no

In case of folding the model with a detector response the ENERGY_RANGE is determined by the energy grid of the detector response matrix and given values for this key word have no effect.   This energy grid can be used without folding, too. In this case the start and end value of the range must be zero. In all other cases an ENERGY_RANGE must be defined by at least two energy values (units are in keV) E_start and E_end:

ENERGY_RANGE E_start E_end [E_step]

An optional step width E_step can be used for convenience. If no step value is given the energy range is devided in 1000 equidistant bins, which correpond to the highest resolution.  

ENERGY_RANGE 0.1  2.5 0.02   ! E_start E_end E_step in keV

The different possibilities of input and/or output can be managed by the INPUT_MODE (default: source) and OUTPUT_MODE (default: parameter).    

INPUT_MODE          source    ! default: source
                              ! background/source+back
OUTPUT_MODE         spectrum  ! default: parameter

Command MODEL/SPECTRUM uses the following hierarchy of file name convention:

• default names inside the program are overwritten by names defined in this parameter file with the exception of the file name for the Raymond-Smith spectral table. The program is first looking for a name thbrte.tbl in the working directory. If it fails it accesses the table, which name is given in the parameter file under the key SPECTRAL_MODEL_FILE.  
• names defined in this parameter file are overwritten by file names given on the command line. That means that file names defined in a parameter file are optional or user defined default names.

The following keywords define file names (examples):

INPUT_FILE           crab_spc          ! observed spectrum
OUTPUT_FILE          crab_spc_o        ! modeled spectrum
CORRECTION_FILE      crab_spc_c        ! corr_effective_area
RESPONSE_FILE        crab_drm          ! det_response_matrix
SPECTRAL_MODEL_FILE  exsas_cal:thbrte  ! table of special model
                                       ! e.g. Raymond-Smith