XSPEC job list
Investigate having models as stand-alone programs. This would allow users to run their own models and hook them into XSPEC without any relinking. XSPEC would pass parameters and energies and the model function program would pass back fluxes (use Eric Mandel's code from SAOtng ?).
Need errors on fluxes and equivalent widths - this requires sampling the confidence volume and finding the max and min of the interesting value within it - should be able to do this with a Tcl script.
The fakeit command needs to check whether the data files and response matrices used are consistent (same number of channels and so on).
Change history output to FITS. Need to decide for what this history output is required.
Write an XSPEC subroutine interface.
Improve "fakeit" to make it more useable for bootstrap statistical estimation.
Ability to escape out of eg steppar or fit and return to the XSPEC prompt.
Double precision on critical arrays.
Generalise the parameters that can be fit. Should be able to define as parameters 1) the background file scalings, 2) the correction file scalings, 3) some complicated function of other parameters.
Check which objects are read/written by which parts of the code - does this reflect modularity ?
Add unfolded plot option to give model/unfolded data without the initial photoelectric absorption.
Extend the test command file to give the system more of a work-out.
Check behaviour of steppar if a parameter that is equated to another is specified. Should either work correctly or give a warning.
A plot summary should produce on one page the d, res, ecd, ecr plots plus useful numbers at the bottom. (Done except for numbers)
step option specified relative to position of minimum ?
AFT reports that "gain,,,1" generates segmentation violation.
Flux command also gives fluxes of individual components.
Add an option to STEPPAR to allow return to user control if a new minimum is found.
Check model valid energy range is consistent with the response matrix.
eqw needs some way of checking whether there is any sensible model component number.
Check whether an ARF has the same energy ranges as the RMF.
Way of turning off limits on parameter values.
New models it would be nice to have : Masai thermal, NEI.
A command to give delta chi-squared appropriate for N interesting parameters at confidence level C.
Plot grouped background on same plot as data.
Need to fix eqw command so works correctly when parameters are linked across components.
Improve history in FITS files created by fakeit.
Plot /b models.
Rename defined models so that parameters can be specified by name eg is model is "po + po" then rename the components "soft" and "hard" and refer to parameters as eg "soft.index".
Place model functions in a separate library.
Option to dump models into the FITS table format.
Fix the "fakeit background_file" option so that it works.
Options to plot in MeV or GeV (also normalization for power-law models etc ?).
The lumin command should not require a dummyrsp but should redefine the energy array internally. Also can lumin work off any redshifts defined in the model ?
Reduce size of test*.mod files in session.
Additional suggestions should be sent to Keith Arnaud, whose home page contains other XSPEC-related links.