TERAMINUTES for the meeting of 31st January 2005

Present: : Chiara, Henry, Emmanuel, Jean-Christophe, Laurent, Anis, Antony, Yannick

 Some discussion about the proposed visit of JC to Garching: Yannick is ok.

WIRCAM Chiara

 There is a problem concerning the computation of the photometric offsets between the quadrants.
 Changing the 2MASS extraction parameters changes a lot the derived photometric parameters. The results derived from the Chiara's stack are different from the Cuby's stack. There are not too many 2MASS stars in each quadrant, either. The result is not very stable. Could this be a bug in SCAMP?
 Emmanuel suggests that it would be better to use another field to equalize the zero-points.
 Yannick also suggests that it would be better to compute the offsets field-to-field manually. Chiara will try and will check the reliability of chip-to-chip calibration derived by Scamp by comparing with the brute force manual calibration.

VeryWide-V0003 Yannick, Laurent, Emmanuel

 Some of the stacks in the very-wide are totally off. Emmanuel suggests that this is a bug in scamp.
 These images have not been erased, and Emmanuel will try to investigate
 There are problems with the stability of the raid partitions on mix3 and mix4
 There are 3880 Very Wide image in V0003. With the parameters used for Scamp (max radius investigation area = 3.5 deg) each SPICA run takes 45mn. So, the release need 40 days. In order to improve this, Laurent will run several Spica on each mix(3,4,6). Expected gain 30-50%. Yannick suggest we also reduce the max radius down to 2 deg., so you can expect a total gain of about 2/3. If so, the V0003 can be released in 15-20 days. We first need to check the reduction of max-radius does notr degrade the SCAMP astrometric calibration.

Transfer of rc-T0003 to CADC Yannick, Laurent

 CADC have downloaded the 21GB of data for the release candidate but they have not yet been made available to the community;
 Yannick has proposed to make the data available to the community of registered users, but we're still awaiting a response from Veillet (we need the list of registered users)
 For the three images u* which were not properly processed by SPICA, Henry will provide the u*-image files to Laurent and they will be copied of output fc disks directly, so we do not need to delete the SPICA products.

SCAMP Emmanuel

 Emmanuel is working an implementating of merged catalogue output in SCAMP

 Also work being done on implementing the error return codes in SCAMP.

 Correction made in size of input arguments in scamp, which allows for very large command-line arguments

 There is a problem with pattern matching at large airmassess, which is why the megacam.ahead doesn't always work

 can't always use all the reference catalogue for pattern matching, and this can be a problem sometime in crowded fields. Adjusting the NMAX param fixes this problem.

Other matters Emmanuel, Chiara, Yannick
 Problem with the training for images which are have stars and good seeing; Lancsos-3 kernel produces lots of spurious artefacts which cause problems; will try with nearest-neighbour interpolation. Yannick will provide a new stack using nearest-neighbour instead of Lanczos3 using the same sub-arcsecond seeing Wide images in both cases.

2MASS catalogue Jean-Christophe
 Apparently no problem to install this on local; CDS will send script and code by the end of Feb.

Webservice Jean-Christophe
 Emmanuel asks JC to provide a tool to allow the uploading of many images.
 For this Fred will have to implement 'write only' ftp access.

PI data Yannick
 Chiara on Cuby's data, waiting for Sanders WIRCAM.
 03BH98 data from Tully almost done. Waiting for Laurent's green light about Qfits-in then Mireille can go ahead. Tully at IAP asked news about the processing status of these data.
 Dougados: Jean-Christophe must contact CFHT/CADC to speed up thr process. Monin from LAOG asked news about these data.

Shaders Anis
 Some tests made under windows, apparently that works well
 Emmanuel and Yannick asked for Anis to provide a practical prototype displaying shader functionality with astronomical images.

EFIGI Emmanuel, Antony
 New data for training sets : 2000 images
 Classification of CFHTLS images: Yannick will class 168 galaxies. Already done by Emmanuel and Pascal (Fouque).