Minutes for the Terameeting of September 2sd 2005.

Present: Mireille, Chiara, Emmanuel, Fred, Laurent, Jean-Christophe, Yannick

Important matter: All

 It is decided the Chiara's Terapix-canonisation will be on September 23 ... Per Omnia Secula Secularum...

Cleaning the T0002 release: Mireille, Laurent, Jean-Christophe, Yannick

 The situation regarding the T0002 images is still unclear. Mireille listed the images that must be removed or added . For example the CFHTLS_W_g_090526-031860_T0002.* fiels must be droped and only CFHTLS_W_g_090526-031900_T0002.* files must be archived at CADC. Likewise, all stacks including Megacam images obtained before June 1st 2003 should be removed (unless the contamination of stacks by images before june 1st is negligible). So, in order to clarify we will proceed as follows: (1) Laurent will send a list of wrong stacks to Yannick ; (2) Yannick will control each stack form the list and will send a list to Jean-Christophe that keep those that must removed at CADC ; (3) Jean-Christophe will contact Luc Simard at CADC and will send our request to him.

T0003 set up : QualityFITS: Fred

 Bachall and Soniera stellar count model almost implemented in Qualityfits (hereafter Qfits); the (ra,dec) -> (l,b) transformation still to be implemented.
 parallelisation of Qfits still in progress.

 Since it will take about one week to Laurent to test/validate the whole pipeline, Qfits must be delivered quickly. It is decided Fred will deliver a new Qfits by Tuesday September 6, 2005 , 11:59:59 am at the latest. If not, then T0003 will run with the old Qfits version.

T0003 set up : SWarp : Emmanuel

 The current SWarp version is ok for the Qfits-in of the STEP1 process. Emmanuel is working on the "line-mixing" problem that occurs with the MULTI_THREAD> 1 option. Emmanuel will try to fix it within the month, so that it will be used for the image stacking step of T0003.

T0003 set up : SCAMP : Emmanuel

 The chromatic aberration correction is not yet validated for a survey release usage. This has no impact on Qfits. Emmanuel will try to release the new version by October (2005!).

T0003 set up : Weightwatcher and SExtrator: Emmanuel

 New release put on Terapix pages by August.

T0003 set up: cluster: Fred

 Fred moved from clix to clix2. the new clix is now clix2 !!. The old clix is now clix2 test you IQ!. Linux 2.6.13 kernel installed on al nodes but ftpix and fcix. Beware: the Web server path is no longer /var/www/html/ but moved to /var/www/localhost/htdocs/ . So please check all your .htaccess data and all your symbolic links.

 Fred moved the dbterapix DB from clix to the dbterapix node. Laurent will compile the DB and check the dbterapix DB works on the 64-bit dbterapix node. Gerard (Tissier) did already check the data base can be implemented on 64-bit processors, so Laurent does not think he will face on severe problem (finger crossed).

 Fred, Laurent and Jean-Christophe must install the cron job on the new clix nodes.

 Fred will set a rotation process in order to backup the dbterapix database and the clix node (most critical nodes).

T0003 set up : disk space: Laurent, JC, Fred

 The transfer from input-FC bay to the cluster disks will start by Tuesday September 5 (2005) at 12:00:00.0001 . All cluster disks has been cleaned, but pix6:/raid2 (Dennefeld' data) and mix5:/raid1/bertin/SDSS (EFIGI SDSS data). The SDSS data will be transferred to mix7 when this node will be installed (mix7 will not be in the cluster for T0003, but should host EFIGI data and likely all first semester - 05B - WIRCAM data).

 Yannick and Fred will move the biggest jpg and png images located on clix (6 GB total).

T0003 set up : SWARP keywords for Laurent : All !

The T0003 keywords that must be stored in final stacks are:

OBJECT =	Spica-generated stack name (see ICD document)
ORIGIN =	"TERAPIX"
DATE_OBS =	Obs. date of the oldest Megacam image of the stack list or nothing if mixed data
TELESCOP =	"CFHT 3.6 m" or nothing if mixed data
INSTRUMENT =	"Megaprime" or nothing if mixed data
DETECTOR =	"Megacam" or nothing if mixed data
OBSERVER =	"QSO team" or nothing if mixed data
PHOT_C =	"30.0"
PHOT_K =	"0.0"
SEEING =	Seeing in arc-sec from QFITS-out

To add these keywords in the final stack, Emmanuel will include at the beginning of the SWarp step the relevant existing CFHTLS input keywords automatically (TELESCOP,INSTRUMENT,DETECTOR and OBSERVER), and will create empty field keywords for those that do not exist yet or that must be different from the input CFHT keywords (OBJECT,ORIGIN, PHOT_C,PHOT_K,SEEING). SWarp will also include automatically the DATE_OBS keyword derived from the oldest Megacam image of the running sample. This way, there will be no need to open/write twice the stack and weightmap stack images at the very end of the processing, just to add the keywords.

The Keyword will be expained in the T0003 explanatory document.

A new meeting will be organised in order to update the Keywords for T0004 (released worldwide).

T0003 set up : USNO usage in SCAMP: Jean-Christophe

 Jean-Christophe will implement the USNO query to CDS for SCAMP in case of failure of the Terapix node hosting the USNO catalogue. Dead line : Monday September 4.... done after the Terapix meeting, see "Nouveau aclientTerapix" email from JC.

PI data: Mireille, Jean-Christophe, Yannick

 Jean-Christophe will first transfer Seymour and Tully PI data (top priority)

 Need to get all metadata attached to PI data, in particular the CFHT validation number. Done: JC got the FITS table that contains the field we need. The JC's email is below:

Toutes les metadata se trouvent dans un table binaire fits ayant le nom de fichier mcmd* (voir mail joint).

Il me semble que ces fichiers mcmd sont aussi disponibles au CADC. Dois-je les telecharger ?" (question to Laurent?)

 Mireille started an evaluation of the 05AH18 u* band data (PI: Sanders). Most images have astrometric keyword problem and SCAMP failed to recover data from the 05AQ05 QRUNID. Following the August 26th Tera-meeting, Henry will try to set the .ahead file using the script written by Fred that automatially select all images of the same QRUNID. This may take some time, not an easy task.

Webservice: Jean-Christophe

 JC is working on the library that will include all steps of the Webservice: check disk space; create the script for pbs, run pbs, check the progress of the processing. The dead line for demo is September 15. Not sure the Web interface will be ready. JC will present to Terapixian the Webservice, as it stands now, next Monday, 14:30 . A version working with a cgi is almost operational for the ADASS meeting, but will likely not be V.O compliant.

WIRCAM: Chiara, Yannick, Jean-Christophe

 JC will install the 2MASS catalague on one Terapix node, as it was done for the USNO catalogue. Emmanuel believes 2MASS will be more useful than USNO for the photometric calibration.

 Chiara FITSVERIFY-ied the WIRCAM images and found important keywords were still missing: CRVAL, CTYPE are the most important. EPOCH and DATE_END were also missing. She sent emails to Loic Albert.

 Chiara will ask Loic Albert whether new (most recent) WIRCAM images are available

 Qfits can work with WIRCAM data provided mofications are done the the FITS header (astrometry). However Qfits fails with a NAXIS = 3 declararation. Fred will look at this problem.