aglc vs. dmextract

The CIAO tool, dmextract, can also be used to bin light curves. There are both major and minor differences between this and the aglc functions and analysis methods. There are planned enhancements to be made to dmextract; in this case, aglc can be considered a prototype for the future dmextract.

The operational difference is that dmextract is file-based, whereas aglc is memory-based. In aglc, files can be read once and binned many times, for different selection criteria (such as vs "hardness" region). This can result in a large performance benefit for large event files.

Another advantage of memory-based analysis is that arbitrary operations can be applied to results using the S-Lang infrastructure. With each histogram, for example, reverse indices are saved. This allows rather complex filtering to be applied. For example, bins with count-rates above an arbitrary threshold can be selected, and the reverse indices used to access some other property of each event binned, such as wavelength in order to bin spectra.

Some current limitations of dmextract (but which will be removed in future versions) are:

• Multi-chip GTI computation: grating spectra span many CCDs; dmextract only uses the GTI from one chip. GTI's can differ slightly from chip-to-chip, or greatly for a heavily piled zero order on S3.
• dmextract does not yet support phase binning.

Some limitations or differences of aglc:

• aglc does not recognize data-model filters on input files (but it will recognize standard cfitsio filters, which are similar). Alternatively, S-Lang filters (e.g., via the where() intrinsic function or the vwhere() interactive "gui-let") can be employed.