XSPEC v9 Known Bugs

Questions and comments to
xanprob@athena.gsfc.nasa.gov

9.00a Plots of multiple datasets with an energy x-axis and use of "setp rebin" to bin up to a certain number of sigma produce some garbage points.
Fixed in v10.00.

9.00b The ascac model requires the XFLT0001 keyword to be an integer followed by two reals. The documentation only mentions the integer. The function actually does nothing with the two reals but will register an error if they are not there.
Fixed in v10.00.

9.00c The pexrav and pexriv models go haywire if the response matrix is changed in the course of an XSPEC session. New versions of these models are available.
Fixed in v10.00.

9.00d Equated parameters are not propagated. eg if par1 = par2 and a command is given to set par2 = par3 then par1 ends up not equated to anything while it should be set to par3.
Fixed in v10.00.

9.00e The "plot summ" command does not work correctly.
Fixed in v10.00.

9.00f When multiple datagroups are in use, the gain fitting is enabled, and the model contains a null component then the "addcomp" command causes total confusion.
Fixed in v10.00.

9.00g The gain command does not work correctly with ROSAT PSPC data (but does work correctly with ASCA data).
Fixed and released as v9.01.

9.00h The untie command leaves the parameter value correct but not the step size, lower limits, or upper limits.
Fixed in v10.00.

9.00i The renorm command does not work correctly if there are additive components with parameters linked to multiplicative components.
Not fixed yet.

9.00j The descriptions of the plrefl and pliref models are misleading and incorrect. ALL users of these models should read these errata supplied by Andrzej Zdziarski.
These models will be replaced in the next release.

9.00k The Chebyshev polynomials used in the models cp6mkl and cp6vmkl are incorrectly defined. The result is that only positively rising polynomials are allowed, a subset of the available phase space.
Fixed in v10.00.

9.00l There is a minor error in the plotting of spectra against wavelength. The wavelength scale is in error by a factor of 0.99982.
Fixed in v10.00.

9.00m An ignore or notice command whose first channel is 0 causes XSPEC to get confused. If you do this you need to exit from the program and restart. Note that XSPEC always uses 1 for the first channel of a spectrum.
Fixed in v10.00.

9.00n The message written by "abun ?" is misleading. The default for relative abundances is angr.
Fixed in v10.00.

9.00o After a large number of calls to the model command when using table models bizarre things start happening due to the available number of i/o channels being exceeded.
Fixed in v10.00.

9.00p The "weight gehrels" command with more than one dataset causes xspec to prompt for the first dataset name. The workaround is to enter the name here and xspec will work correctly.
Fixed in v10.00.

9.00q Setting the MY_XCOMS environment variable does not work.
This is not a bug - the variable must have a trailing /.

9.00r There is an error in the comptt model. A fixed version of the function source routine is available.
Fixed in v10.00.

9.00s The script command does not work as advertised. Just giving the command "script" generates a file called xspec.log not xspec.xcm. Also the scripting cannot be turned off using the command "script none".
Fixed in v10.00.

9.00t The second parameter of the zvfeabs model is not the metal abundance but the metal abundance abundance times the value of the first parameter.
Fixed in v10.00.

9.00u Response matrices with F_CHAN relative to 0, not 1, have a silent one channel offset when read into XSPEC. The XTE HEXTE matrices are the only ones known about at present which produce this bug.
Fixed and released as v9.02

9.00v The save command loses any additive constant in a parameter link e.g. parameter 1 = parameter 2 + 0.5 is saved as parameter 1 = parameter 2.
Fixed in v10.00.

9.00w If the flux command resets the energy limits for a dataset then the new energy limits are used for all subsequent datasets.
Fixed in v10.00.

9.00x If multiple datasets are in use and one is replaced from a new file with more channels than the original then all subsequent datasets are trashed.
Fixed in v10.00.

Keith Arnaud, Lab. for High Energy Astrophysics, NASA/Goddard Space Flight Center

9.00a	Plots of multiple datasets with an energy x-axis and use of "setp rebin" to bin up to a certain number of sigma produce some garbage points. Fixed in v10.00.
9.00b	The ascac model requires the XFLT0001 keyword to be an integer followed by two reals. The documentation only mentions the integer. The function actually does nothing with the two reals but will register an error if they are not there. Fixed in v10.00.
9.00c	The pexrav and pexriv models go haywire if the response matrix is changed in the course of an XSPEC session. New versions of these models are available. Fixed in v10.00.
9.00d	Equated parameters are not propagated. eg if par1 = par2 and a command is given to set par2 = par3 then par1 ends up not equated to anything while it should be set to par3. Fixed in v10.00.
9.00e	The "plot summ" command does not work correctly. Fixed in v10.00.
9.00f	When multiple datagroups are in use, the gain fitting is enabled, and the model contains a null component then the "addcomp" command causes total confusion. Fixed in v10.00.
9.00g	The gain command does not work correctly with ROSAT PSPC data (but does work correctly with ASCA data). Fixed and released as v9.01.
9.00h	The untie command leaves the parameter value correct but not the step size, lower limits, or upper limits. Fixed in v10.00.
9.00i	The renorm command does not work correctly if there are additive components with parameters linked to multiplicative components. Not fixed yet.
9.00j	The descriptions of the plrefl and pliref models are misleading and incorrect. ALL users of these models should read these errata supplied by Andrzej Zdziarski. These models will be replaced in the next release.
9.00k	The Chebyshev polynomials used in the models cp6mkl and cp6vmkl are incorrectly defined. The result is that only positively rising polynomials are allowed, a subset of the available phase space. Fixed in v10.00.
9.00l	There is a minor error in the plotting of spectra against wavelength. The wavelength scale is in error by a factor of 0.99982. Fixed in v10.00.
9.00m	An ignore or notice command whose first channel is 0 causes XSPEC to get confused. If you do this you need to exit from the program and restart. Note that XSPEC always uses 1 for the first channel of a spectrum. Fixed in v10.00.
9.00n	The message written by "abun ?" is misleading. The default for relative abundances is angr. Fixed in v10.00.
9.00o	After a large number of calls to the model command when using table models bizarre things start happening due to the available number of i/o channels being exceeded. Fixed in v10.00.
9.00p	The "weight gehrels" command with more than one dataset causes xspec to prompt for the first dataset name. The workaround is to enter the name here and xspec will work correctly. Fixed in v10.00.
9.00q	Setting the MY_XCOMS environment variable does not work. This is not a bug - the variable must have a trailing /.
9.00r	There is an error in the comptt model. A fixed version of the function source routine is available. Fixed in v10.00.
9.00s	The script command does not work as advertised. Just giving the command "script" generates a file called xspec.log not xspec.xcm. Also the scripting cannot be turned off using the command "script none". Fixed in v10.00.
9.00t	The second parameter of the zvfeabs model is not the metal abundance but the metal abundance abundance times the value of the first parameter. Fixed in v10.00.
9.00u	Response matrices with F_CHAN relative to 0, not 1, have a silent one channel offset when read into XSPEC. The XTE HEXTE matrices are the only ones known about at present which produce this bug. Fixed and released as v9.02
9.00v	The save command loses any additive constant in a parameter link e.g. parameter 1 = parameter 2 + 0.5 is saved as parameter 1 = parameter 2. Fixed in v10.00.
9.00w	If the flux command resets the energy limits for a dataset then the new energy limits are used for all subsequent datasets. Fixed in v10.00.
9.00x	If multiple datasets are in use and one is replaced from a new file with more channels than the original then all subsequent datasets are trashed. Fixed in v10.00.