D..3 January, 2008 (AE Version 3.172)

• By default MARX, rather than the mkpsf tool, is now used to model the HRMA PSF. You can specify that AE revert to using mkpsf via the new option (/PSF_FROM_LIBRARY, §7.6).

  Note that there appears to be a systematic astrometry difference between MARX and mkpsf--the centroid of the two PSFs (before blurring) are generally offset from each other. I find this offset is near zero on-axis and increases linearly with off-axis angle at a rate of $\sim$0.12 sky pixels per arc minute. I do not, of course, know which PSF has the more correct astrometry. One implication of this offset is that one should avoid the situation where a source's position was determined using a PSF from one system (e.g. via AE's ``correlation position'' estimate or via image reconstruction), and the source's extraction region was constructed using a PSF from the other system. In a multi-observation project, one should not extract some observations with one flavor of PSF and some with the other unless you have considered very carefully how that would affect your analysis.

• Post-HRMA blurring effects are now modeled individually rather than via a single Gaussian smoothing as was done in previous versions (§5.1.2). I made this change because I have two concerns about the Gaussian method. First, I have theoretical concerns about the use of Gaussian models for two of the blur components. Second, even if the Gaussian model is adequate, as of January 2008 the recommended parameter value to use in that model (e.g. the DitherBlur parameter in MARX) has no solid calibration provenance.

• Spectral fitting with the C-statistic now employs a completely different model family (cplinear) for the background, significantly improving the fit for weak sources (§5.12). To support this innovation, AE multi-ObsId spectra now represent background scaling using the AREASCAL FITS keyword/column rather than the more common BACKSCAL keyword/column (§5.6). Thus, if you wish this AE to fit sources extracted by an earlier version of AE, then you must first re-run the MERGE stage to rebuild the spectra.

• The tool ae_better_backgrounds (§7.14) for constructing background spectra in crowded fields has been improved and further tested. I use it routinely myself, and now recommend it whenever a point source's extraction region may contain significant contamination from another point source.

• Our own recipe for using AE (§7.1) has been significantly revised.

• A tool that assists with time-resolved spectroscopy (ae_timerange_extract) is now available (§7.20).

• A tool to manually group spectra using the AE grouping algorithm (ae_group_spectrum) is now available (§7.20).

• A tool to plot radial profiles of a source's data and PSF (ae_radial_profile) is now available (§7.20).

• A tool that partially automates the complex process of building AE-compatible PSFs from ChaRT (ae_chart_interface) is now available (§7.20).

• Fitting scripts can now pause after the source is fit, but before fluxes and output files are produced, so the observer can interact with XSPEC to refine the fit (see the /INTERACTIVE option in §7.18).

• Please note that fitting scripts and data products are now stored in a sub-directory named spectral_models (§7.18).

• Please note that the parameters expected by the tool ae_make_catalog (§3, §7.2.2) have changed to be more consistent with the tools ae_source_manager and ae_standard_extraction--the source list file name parameter is now optional, defaulting to ``all.srclist''.

• Please note that the use of MARX to generate PSFs requires that an additional CIAO data product, the aspect file, be provided to AE (§3 and §7.6, keyword ASPECT_FN).

• Please note that for some of the multi-ObsId source properties stored in the obs.stats file (§7.16) the method for combining the corresponding single-ObsId properties has changed. For example, PSF_FRAC was defined as the smallest value found among the observations; now it is a weighted average.

• The CHECK_POSITIONS stage (§7.17) has a new option, THETA_RANGE, that can be used to limit which sources are processed.

• The SHOW_REGIONS stage (§7.8) has a new panning mode.

• This manual now includes a suggestion for how users of the bash shell can run AE (§6.1).

• I believe the earliest version of IDL that will run AE is 6.1; I am currently using 6.4.1.

• The XSPEC scripts distributed with AE now require XSPEC version 12.3.1 or higher.

• The MARX simulator is now listed as one of the packages AE requires (§6).

• AE has not yet had much run-time under CIAO 4.0; please report any problems you encounter.

• Although I try to avoid them, there may very well be backward compatibility problems, beyond those mentioned above, between this version of AE and data products produced by earlier versions. Please contact me if you have specific concerns.