Functionality

Note that XAssist sets the telescope and XAssist currently performs the following functions (each performed in a ``package'' which is shown in bold):

• Acquire Downloads data from an appropriate archive (ASCA and ROSAT only) or sets up links to existing data (i.e., downloaded manually or copied from a CDROM). For CIAO data, see cda.py at http://xassist.pha.jhu.edu.

• Reduce Initial data reduction (including recomputing gain corrections, if possible) and filtering of bad PHA/PI values to improve signal-to-noise, creation of full-band images along with a smoothed image for display purposes and a mask image giving the active detector area for detection purposes. For Chandra data, an ``exposure'' map is generated and that is used to generate a mask (detector masks are generated also by ciao tools but this approach is used since the exposure map is useful of course). XAssist will also attempt to fit for the rotation between the sky and detector coordinates in order to determine the CCD boundaries. N.B., in XAssist versions prior to 0.889 the PI selection was not optimal for XMM-Newton data.

• Detect Source detection (using a builtin routine for ASCA and ROSAT, CIAO wavdetect for ACIS) using a mexican-hat (a.k.a. wavelet) algorithm, and a sliding box algorithm (eboxdetect + emldetect in xmmsas) for XMM-Newton.

• Timeclean ``Cleaning'' of the background by clipping times when the background rate is very high or very low (based on the user-defined $\sigma$ parameter) to produce a gti file for use in subsequent analysis. The background level is determined by excluding each detected source. Note that if the mask created in the Reduce package is incorrect, then of course the background level will also be inaccurate.

• Simplespatial ``Simple'' source fitting (i.e., not including the PSF) to assess source significance and extent, based on a small stamp image around each source with models that include a sloping background and adjacent sources. The sources are fit with elliptical Gaussian models (surface brightness $\sim e^{- ( x - x_c )^2 / 2 \sigma^2_x - ( y - y_c )^2 / 2 \sigma^2_y}$ with and without the position and extent being ``free'', and the statistically most signicant fit is kept in memory (in general the extent in the X and Y direction are tied, in which case the source model is a circular gaussian). The fitting is done with ximgfit (see http://xassist.pha.jhu.edu/ximgfit) using the ``C'' statistic appropriate for the Poissonian regime.

• Assess Flagging of extended (currently only for HRI and ACIS data), asymmetric, underresolved, and crowded sources. Spatial fit errors are stored (and written out in the csv file in the top-level directory).

• Extract Source and background events lists, images, spectra, and light curves are extracted for each source. Spectral responses are generated for each source (rmf files are only retained for sources with >SpectralMinCnts counts for which spectral fitting is possible).

• Reassess The source flux is re-computed based on the ARF computed in the Extract package. In the near future, if the source and background region have sufficient counts, the source and background count rates will be determined from the extracted events lists (and the flux and source significance will be updated).

• Spectral Absorbed power-law models are fitted to sources using XSPEC (along with the TCL scripts available at http://xassist.pha.jhu.edu/downloads to automate the error search). Spectal fitting is only attempted for sources with more than 100 counts. Fit parameters are stored in the csv file in the top-level directory.

• Temporal Computes the K-S statistic comparing the source light curve to the ``local'' background extracted and to the ``global'' background determined in the timeclean package.

• Spatial Currently just extracts ``clean'' images of the FOV (meaning with the bgd. filtering done in timeclean), smooths the images with a gaussian kernel, for ACIS data also optionally adaptively-smoothing with csmooth, and then attempting to remove sources using a Poisson-deviate replacement system (the proceedure used in the Chandra M82 paper Griffiths et al. 2000 and similar to the CIAO thread on this point). Future versions will generate psfs and use them to re-fit the brighter sources in user-supplied bandpasses.

• Correlation HEASARC tables usno, gsc2, rc3, first, 2mass, veron2001, and zcat are called for each field and the tables are correlated with the source list for each detector.

Features under development (although may not be available for some time since formal funding has ended and development is geared mostly toward usage of XAssist for Chandra and XMM-Newton survey work):

• Fitting of sources including the PSF (ximgfit currently has semi-analytic models builtin for the HRI and PSPC, simulators will be used to generate PSFs for ASCA, Chandra, and XMM).

• Merging of qlook (http://xassist.pha.jhu.edu/xassist/qlook) functionality, specifically source position lookup and available data determination, into the full XAssist system so that a user can input a source name and XAssist will set itself up to process available ROSAT and ASCA data (and later also Chandra and XMM) for that source.

• A GUI and better batch processing support (including batch jobs that can be submitted by remote uses who do not want to download and install the software themselves).

• Simulation for generation of psfs and for use in more advanced analysis (such as more precisely determining detection threshold for extended sources).