Teraminutes for meeting of Friday 13th. Absent (and in Canada): Jean-Christophe.

Panorapix (Fred) problem in the corners of the images (possible pointer offset?): fixed during the Terameeting! WCS display information now works. Overlays, Postscript for monday, before leaving on the aeroplane.

QualitFITS(Fred) problem in the WCS information resolved.

Dbclient (Gerard) perhaps 2GB bug found! Perhaps comes from pointer/allocation of memory for the catalogues. This class has been totally re-written, and tests are underway. We should now early this coming week whether the 2 GB bug dead!

SCamp (Emmanuel) 1/4 crashes, Extra degree of freedom added; in which it is assumed that we don't always have the same camera: (distortion term assumes that we don't always deal with the same camera). Emmanuel feels good! And is on target for a 'full release' of Thursday February 19th 2004.

Defectix (Anthony): After discussions with Emmanuel over the week, Anthony moved to an ambitious project of vision model. Anthony started first with a PCA on images. The software works by small blocks of pixels, pass neurones on the mask two times. First time get info, second time run analyse on PCA, run PCA on each block, look a the variance in the block. 16x16 boxes, look for outliers.

Deep stacks (Henry) Testing and photometric calibration of deep griz-D3 stack underway. Impressive agreement between F02-virmos counts and i* Megacam counts. Absolute magnitude scales seem to be better than 0.05mags. Colour-colour diagrams also good in gr-ri plane. Perhaps a small problem in ri-iz: under testing. Potential problem in w(theta) at large angular scales: also under testing.

Wide W3 stacks (Mireille) stacks made over 4x1 deg2 but no way to visualize the images! On the alphas, and using his PC, Henry was able to visualise the 3 GB cut-down image, but no-one else was. Probably a problem with panorapix or the Xserver, or both. Should be addressed during next tera-meetings.

Spica (Laurent): RA/DEC boxes added ; versioning added. problem with survey processing mode tracked down to SCAMP. Not much progress as still waiting for scamp.

ICD document (JC/Laurent): discussion about sending config files to the CADC. CADC want config files each time. Unfortunately these file names have to be different each time. Discussion about visibility of the configuration files. TERAPIX should not write an interface for the CADC.

Cutting images and delivery of very large (>1 Megacam pointing) images: strategy not decided yet. Canadians want images 1x1; overlap preferred but still debated. Some of the French community want the large images. Question about final data products: what information should be included in the final catalogues? How should flag-information be decided.

Data Sophie Maurogordato (Gilles) not much progress here.

MDA-ACI proposal deadline march 22. Yannick and Emmanuel will contact STICK.

Hardware ========

Opterons (Emmanuel). Combination of problems, both hardware- and software-related prevent the opterons from working: firstly, a mandrake release canditate for amd64 is still not ready yet; still haven't received the right motherboards from Mr. Landat. Maybe a test machine can be made to work with just the internal disk? For the time being, tests will be made with smaller images (Henry notes he is still unable to produce a combined chisquared image of the four D3 images, on any machine.)

DNS changes: the following changes will be implemented in the DNS, Gilles will see Georges about this:

new name/old name: cosmix/pix3 testix/pix2 wifix/pix1

Labelling machines in control room isn't obvious (sometimes the name on the KVM switch label doesn't match the actual machine). Names should be affixed to all the machines.

Next Meeting: Friday Feb. 20 at 14:30.