Elixir System Release History
The CFHT Elixir System is a vast collection of software tools that affect the pixel data and the FITS keywords in various ways. It would not be reasonable to provide a version number for each component. In consequence, a single general version of Elixir is stamped into the processed data FITS headers (machine readable format):
 EL_SYS  = '2.0               ' / Elixir System Version
 

The pixel processing part of Elixir - FLIPS - also loads a version number but only in a human readable HISTORY field for the record.

For clarity we will refer only in this page to the Elixir general revision version used for releasing data made available at CADC, not the actual Elixir version since it was first put in operation at CFHT in 2001 for the CFH12K mosaic camera.

Release date Elixir System versionNote
TBD 3.0 Optimized recipes for MegaCam
Nov. 2003 2.0 Native CFH12K recipes tuned to MegaCam
Aug. 2003 1.0 Basic CFH12K port, trial distribution


Elixir System Version 2.0:

We learned a lot over the course of the first 8 months of operation of the instrument, yet time was too short to derive the best pre-processing recipes as the 2003A semester was over and the PIs and CFHTLS users were waiting for the processed data. We are fully aware that some parts of the pipeline still need some refinements as the evolution from the version 1.0 to 2.0 was simply a tuning of existing CFH12K recipes whereas the implementation of new recipes is needed (e.g. fringes) and more data mining required to build the best models for the sky background modes correction, etc... Here's the list of the major problems affecting the data processed by Elixir 2.0 that are currently being looked at for the version 3.0.

  • Fringes (i' and z' band):
    It appears that there are second order terms most likely related to the airmass resulting in fringes residuals up to 1% for some images. Several fringe patterns need to be constructed for various observing conditions and applied to the science image using a principal components decomposition to find the best combination to eventually correct the fringes at the fraction of a percent (0.1 to 0.2%) as CFHT is committed to.

  • Sky background modes (all filters):
    After one year of operation, we now have a large collection of images to stack to decompose the various sky additive components and build a proper model for each of them.

  • Data splicing and flat-fielding (all filters):
    Even as good as it is (within specification), the slight non-linearity of some of the CCDs becomes apparent after data splicing (stitching together the two amplifiers from a same CCD) at a very low level. This can be minimized by taking flat-fields at a lower level (1/3 of the CCD dynamic). New flat-fields for 2003A and 2003B will be constructed from the existing data (when possible) using that new limit on the flux.

  • Photometric superflat (all filters):
    The photometric grids obtained early 2003 to correct the data aren't optimal and we've since acquired (fall 2003) new grids with longer exposure times and a lot more more stars. We expect a gain of 10 on the precision of the map (though the current one is already doing a good job with a photometry flatness across the image better than one percent).


Elixir System Version 1.0:

A straight port from the CFH12K recipes which proved inadequate in their capabilities to flatten properly the images (sky background modes correction) and the fringes.


How to find out if you have the latest version of a processed image

Note that the presence of a new number for the Elixir System Version keyword in a file that had previously been processed is a sure sign that it is a new version of the processed image. The Elixir date keyword is however of equal importance to track a new processed version of the image as a given file may have been reprocessed with a better set of new detrending frames but using the same Elixir software version.

In order to define precisely if the preprocessed file you have is the most up to date one offered at CADC, compare the date keyword "DATEPROC" in the FITS header:

 DATEPROC= '2003-11-05T14:56:45' / Processing Date