New in Version 9

The following list summarizes the major changes to XSPEC since the previous edition of the user's manual.

• An abund command to switch between different Solar abundance tables.
• A goodness command to perform a Monte-Carlo goodness-of-fit calculation for non-background-subtracted data.
• A systematic command to apply a uniform systematic uncertainty to the model.
• An untie command to unlink two parameters that have been linked together.
• A weight command to change the weighting scheme used in the calculation of chi-squared. The new option is gehrels which uses , an approximation for small number statistics.
• The step command has a new option step best which resets non-stepped parameters to the best-fit after each grid point. The opposite is step current.
• par1 = par2 + number sets par1 to be equal to par2 plus an additive constant.
• ignore and notice work with energies if given real numbers as their arguments.
• Multiple spectra can be stored in the same FITS file and read individually. The command data filenm{N} reads spectrum N in filenm.pha.
• The behaviour of the data command is changed so that data 1 filename does not remove files 2-** but data filename does.
• model none removes all models.
• The following new models have been added and old ones improved:
  • An ionized absorber (absori from Magdziarz and Zdziarski) has been added.
  • A model (ascac) has been added for simultaneous analysis of ASCA spectra with corrections for energy-dependent telescope scattering.
  • Power-law differential emission models (cemkl and cevmkl) using the MEKAL code have been added.
  • A model (comptt) for Titarchuk's comptonization of a Wien law spectrum has been added.
  • Sixth-order Chebyshev polynomial differential emission models (cp6mkl and cp6vmkl) using the MEKAL code have been added.
  • MEKAL models (mekal and vmekal) have been added. These are new versions of the MEKA model using Fe L calculations supplied by Liedahl.
  • A pair plasma model (nteea from Magdziarz and Zdziarski) has been added.
  • Models of exponentially cut-off power-laws reflected from neutral matter and ionized matter (pexrav, pexriv from Magdziarz and Zdziarski).
  • A photo-electric absorption model (phabs) using the latest relative abundances and cross-sections has been added.
  • A model (plabs) has been added for absorption as a power-law in energy, useful for things like dust.
  • A model (redge) for recombination edge emission has been added.
  • A model of comptonized reflection from neutral matter (tita_a from Magdziarz and Zdziarski) has been added.
  • A model for redshifted absorption with variable iron abundance (zvfeabs) has been added.
  • The plasma emission models now use relative Solar abundances as set by the abund command. The default is from Anders & Grevesse and the new option is from Feldman. The changes in this new abundance table are as follows : C : , N : , Ne : , Ar : , Fe: .
  • All absorption models except wabs and zwabs use the same relative abundances as the plasma emission models as set by the abund command.
  • The bremss, vbremss, and zbremss now have a normalization which is not temperature-dependent.
• For local models the lmodel.dat file needs an extra line starting with the word "null" than before. See the example file in the release.
• A number of changes have been made to table models.
  • All table models now use a FITS format. The program ctable converts old format table models into FITS.
  • Multiple table models can be used simultaneously.
  • The Raymond-Smith model now runs off a table in the standard table model FITS format.
  • An etable model reads an additive table model type file and then takes the exponential of the negative of the input.
  • The mtable model now defaults to 1 rather than 0 when there is no tabulated model for part of the response energy range.
• The C-statistic has a new normalization suggested by John Castor which provides a goodness-of-fit.
• All vectors for a given dataset are plotted in the same colour.
• XSPEC can be called as a subroutine (through xspec/src/subr/xscmnd.f)
• Model functions can give a result depending on which file the model is being calculated for. This has meant a change in the subroutine interface, the variable WORKSP has been removed and a new integer IFL has been included - see files in xspec/src/functions for examples. For most models IFL will not be used but must still be in the subroutine call.
• All VMS Fortran structures have been removed from the XSPEC source code.
• A few bugs have been fixed.
  • The gain fit command is a little more robust (but should still be used with care). It now handles features in the detector efficiency curve correctly provided that they are included in the arf or no arf is used and only a response matrix is given.
  • There was a minor error when calculating errors on grouped data when both the data and background contained bins with zero counts.
  • The wavelength plots were in /keV rather than /A as claimed.
  • ignore bad now works for arbitrarily many ignore ranges.