Terapix Star Formation Region IC 1396, © 2001 CFHT
Teraminutes for the meeting of Friday, February 6th 2004
Article
by HJMCC - Updated June 9th, 2005

Teraminutes for meeting of Friday 6th February Present: everyone

Panorapix (Fred): Still problems with the transformation between x,y RA,DEC. and reading WCS information correctly. Note from note from Henry: zooming out on big images still crashes panorapix. *Deadline*: before Fred leaves for Rio. MEF, WCS, overlays, postscript, jpeg outputs will be implemented. Fred assures us that these features are easy to implement. Emmanuel is doubtful. Emmanuel bets Yannick a bottle of Chablis 1er cru that these features won't be ready before the deadline!

QualityFits (Fred): Reverse images in some stacks: when QualityFits is executed again, the next time around there are no is no stacked images. Perhaps it's Swarp which is responsible for the switched images, although the version of SWarp hasn't changed. QualityFITS for stacked images: we need to have a map of PSF in this case (and not just one big ave. PSF in the centre) Emmanuel report on visualisation problem with QualityFits 1.07 and 1.08: icons and corresponding png are shifted by 1 icon.

Use of WCS lib should be implemented once done for pano.

Scamp (Emmanuel): Problem in SWarp prevents projection of astrometric solution (in all cases). Several special cases (like the integral-CCD dithers in the very wide survey) cause problems which aren't fully handled yet. Outlying CCDs which have problems are not handled correctly yet: there is module which handles this case but it isn't activated yet (computes the mean WCS for each CCD and uses this for CCDs which are badly wrong. Emmanuel promises that in 15 days (Feb. 26th) there will be a version of Scamp which includes the photometric solution. Discussion about dead CCDs: conclusion: image processing software (Emmanuel's) should handle bad CCDs. Complex and it will take time to implement this.

Defectix (Anthony) Not much progress in the code. Recovery rate of simulated objects from fake images 80%. Choice between PCA and ICA. Better to start with simple PCA prior to move on complex and computer intensive ICA. Yannick will provide masks+image examples to Anthony next Monday.

Deep stacks (Henry): Colour g'r'i'z' stacks for the d3 field shown (at least the would have been if there was any network in the meeting room). Discussion about the astrometric residuals: seems that the small dithers almost certainly contribute to the size of the astrometric residuals in the border regions (confirmed in the case of a2029, where the dithers are much larger and the astrometric residuals are a factor of two smaller). Next up: photometric calibration. Need to check with Yannick whether color gradient effects seen on the stacked images are differential refraction residuals.

Spica (Laurent): On going: waiting for Scamp and the new version of QualityFITS before reprocessing more data. Cutting images, and what to do for very large images. Make image quantum of 1.1 deg2. Deadline: *next Friday* to decide. Try to find out from CFHTLS Steering Group what they need. Possibly there should be an overlap?

ICD (Jean-Christophe): The document is progressing well, with inputs from Jean-Christophe, Yannick and Laurent. Will be finalised once Jean-Christophe will be at CADC.

Skywatcher (Jean-Christophe): Little progress. Sky projections with PLplot implemented and Jean-Christophe is now handling WCS, so important progress expected once Jean-Christophe back (if Canadian bears leaves him...)

Dbclient (Gerard): Now handles catalogues up to 7MB: Understanding the code not straightforward, some stuff not optimal.

Webservice (Gerard): Under investigation.

Sophie Maurogordato cluster CFH12K (Gilles): Data loaded onto the cluster, Gilles is checking it out. Mireille provided advices to Gilles and checked no pre-processing necessary. Gilles can move on straight to Astrometrix.

JP Picat Coma CFH12K data: Mireille will process them this week.

Hardware ========

Naming the machines: For the science cluster we will adopt the name 'cosmix' and for each of the sub-nodes cos1, cos2 ...; for the data reduction cluster we will have 'pix' and pix1, pix2 ... for the sub-nodes.

Discussion about installing mandrake clustering. This is not an upgrade but a *complete reinstall*: all system software will have to be removed from all machines. This will happen in March, after Fred returns from Rio (March 7)

Discussion follows about need to have centralised secure data storage: probably we will have a specific meeting about this at some point in the future. Gilles will find out about what solution has been adopted at the IAP for this.

Opteron update: The arima cards have been tested, and they work (in AMD64). Mr. Landat will send us a complete machine with the cards and disks for testing ourselves.

Next tera-meeting: Friday Feb. 13, 14:30 (is this a good idea?).


Site Map  -   -  Contact
© Terapix 2003-2011