Teraminutes for the meeting of Friday, January 30th.

Absent: Emmanuel, Mireille

Panorapix (Fred) (1) Version demonstrated which can display MEFs individually: version to display all MEFs simultaneously not available yet. 39s to load all CCDs. Fred and Anthony have modified the CFITSIO library and have put this release on the Terapix CVS. (2) IMPORTANT: Can Panorapix display all CCDs simultaneously yet? No. This seems to be a lot of work. How is this carried out in other applications? Note to Emmanuel: does the Astro-wise specification document indicate that simultaneous display of all images has to be performed by Panorapix?

QualityFits (Fred) Some images have CCD images inverted. Fred is going to fix this problem in the coming week.

After discussions with Henry who needs the PSF mapping, Fred will use the WCS library to modify QualityFits and run the PSF anisotropy mapping faster than in the early QualityFits release.

Defectix (Anthony) CFITSIO new library available. Routine tested which adds rectangles to images and then searches for them afterwards. 80% of these rectangle defects have been identified on the images. Note bad for very first tests.

*Action* Yannick will provide masks created on real images to train the software.

SPICA (Laurent) (1) SCAMP is not ready, which is slowing down data processing. In the meantime, can we use Astrometrix in the pipeline? We also need Photometric calibration as output provided by Photometrix. Astrometrix is not easy to install (and has bad error trapping). ***Meeting proposed with Emmanuel, Yannick, Laurent and Henry on Wednesday to discuss these issues** Real need to release data to the community. (2) Manual validation of images processed by QualityFits. Henry and Yannick will grade images processed by QualityFits. (3) Need to define how we handle very large images, and mosaic of multiple MEGACAM pointings (for instance for the Wide survey). Do we SWarp all the images together? Yannick proposes we define a 'data release' quantum of 1.1 deg^2. Image center defined by the barycenter of all images.

We finally decide the following: SPICA identified the object keyword and handle data as follows:

- For D1, D2, D3 and D4, we stack and do not cut since each D*s has 1 deg^2 - For W1, W2 and W3, we stack each 50 deg^2 and make 3 50 deg^2 images that we will cut into 50 1 deg^2 quanta for CADC - For the Very Wide, each field is composed of 3 image triplets, so we just stack and do not cut into 1 deg^2 quantum.

Bad CCDs in QualityFits: when the number counts are computed, how are the bad CCDs identified? Can these CCDs be flagged somehow and removed from the number counts calculations.

Would like to add keyword headers describing the health state of the CCD. JC/Fred also needs to look at the mean value in each CCD for the seeing calculation and reject bad CCDs (look for objects which are at zero). ** action** Jean-Christophe will spot CCDs prior to feed the DB and will flag bad CCDs (those with pixel values =0)

ICD (Jean-Christophe) [see also action above] WeightWatcher output needs to have the ORIGIN, SOFINST... headers added (the same as in SWarp). Needs to be made explicit where the images were created (increase visibility of TERAPIX).

dbclient (Gerard) Bertrand's corrections to dbclient on Kiravix checked by insure, looking for various memory leaks/optimisations. try dbclient memusage (suggestion from Fred, already suggested by Emmanuel during previous Tera-meeting). Tests being carried out on the new node11 testix which is now a pure development node for DBclient.

Gerard contacted P. Fernique in order to produce SExtractor web server for running SExtractor from Aladdin.

Security matters (Laurent, Gerard), Intranet: SPICA passwords will be made the same as those used to access the web server administration page. The pipeline account password will be changed as well.

Other matters: VOtables (Yannick). Many acronyms explained by Yannick. Too many to list.

Deep stacks (Henry) (1) r*, i* stacks shown. Astrometric problems discussed. Henry will try to SWarp an image with independent astrometry to see what happens. - griz stacks will be ready shortly. REMINDER TO FRED: we need to find out how to increase the shell variable 'descriptors' cluster-wide, otherwise SWarp will fail with very large numbers of images. (2) Photometric calibration of i* stack indicate that it reaches to at least i* 25AB. (3) For star-galaxy separation on the stacked image, Henry will use the PSF model extracted by PSFex extracted by QualityFITS.

**Important: Henry keeps track of its photometric calibration steps. Each step will have to be implemented in SPICA soon.

Hardware (Fred): Sub-nodes for the science cluster aren't ready yet because of a problem with the PCI cards.

Mr. Landat has been contacted wrt the Opteron motherboards (which don't work). Option+ will test new cards next week and provide us with test cards if they are successful.

Next Tera-meeting Friday 6th of February at 14.30 (note again the new time).