Terameeting Friday April 22nd 2005

Emmanuel: various bugs Investigations with valgrind for swarping very large numbers of images is inconclusive: the line reported by valgrind where the program crashes is not meaningful. More work needed.

Swarp, resampling bug, Emmanuel This is seen when shifting images which have the wrong central ra,dec value.

 It seems that while resampling certain images, there is a problem with inversion of pixel values: this could be a problem with the value of COMBINE_BUFSIZE. Emmanuel recommends simply increasing the value of COMBINE_BUFSIZE. Jean-Christophe will re-run the image resampling with a larger value of COMBINE_BUFSIZE.

Emmanuel: scamp bug: pattern matching Some stacks have an image which doesn't match with the astrometric catalogue; removing this stack and re-running scamp reveals another image with a bad astrometric solution: so this probably comes from a bug in scamp. Emmanuel will investigate this later in the week.

Status of the release (Laurent)

 100 images in very-wide (1 week to go)
 10 days to finish the deep stacks
 10 days to finish the wide stacks

Photometric recalibration - Henry will continue to work on the photometric intercalibration between the images. This program will have to be tested and working before Henry leaves for Marseille on Tuesday evening. It is currently top priority. Henry will then show Mireille how to use the tools so that she can make progress on this during the following week.

Quality assessment (Mireille, Henry, Yannick): Mireille asks for a list of statistics to be checked for each image.

 Henry will supply the procedure to produce the catalogues for each stack (star-galaxy separation and application of masks), then Mireille will generate masked catalogues for the 18 patches in the wide data.

Hardware (Fred)
 The ftp server has been ordered: the machine will have 4Tb of disk (opteron bi-processor)
 new rack for mix5 has arrived
 Gilles will turn off the alphas and the new rack will placed there
 Gilles will take pix10 to make tests (not clear if we can recover data from this machine: it should be treated as lost).
 For mix4 (ex mix2) fred seems to have narrowed down the problem to originating with the one of the RAID cards.

Other points: -JC has downloaded USNOB2 catalogue on dbterapix. Some tests necessary before this can be fully updated.