Minutes for the terameeting of Friday July 29th 2005

Present: Fred, Yannick, Laurent, Jean-Christophe. Mireille, Chiara, Fred, Anthony, Henry

 Fred and Mireille's birthday! Mireille brought a nice bottle of "Cremant" as well as nice cake-lets from Moisan.

 A nice picture of the TERAPIX group's V.3.0 was taken by Emmanuel and is available here Fred did take picture control of the picture, as quality assessment of the new Emmanuel's camera...

Status of machines and upgrades (Fred)
 Fred was very busy with on going hardware orders that turns out to be more complicated than expected (paperworks and mistakes created by the IAP). This has delayed all other tasks.
 Switch from clix2 -> clix1 not finished yet
 parallelization of qualityfits should be completed this afternoon
 system software upgrades mostly completed on all machines: all machines should have the same kernel revision and software versions.
 fpix2 mix2, mix5, still need to be rebooted. Pending problem with these machines when you reboot because there is a problem with ypind
 not clear how the pix machines can be upgraded to gentoo

Urgent matters to be completed before Fred leaves
 reboot of the machines [completed]
 parallelization of qualityfits [still pending]

PBS:Jean-Christophe
 This allows the distribution of jobs across the network.
 Intelligently allows one to select nodes to execute jobs based on available resources at each node.
 PBSpro offers some interesting features (automatic restarts after hardware crashes) but costs around 2000euros/an
 PBS could be useful for webservices
 Yannick reminds we will have a meeting by September 15 (with a pre-meeting by September 1st) to assess JC's progres and decide between these alternatives

SCAMP (Emmanue)l
 Emmanuel has made some tests on simulations to find out what the absolute astrometric accuracy is
 Emmanuel finds that the 'best possible' astrometric solution is around 4mas/rms: this was achieved by removing blended objects from the sextractor catalogues;
 This is still around a factor of two higher than the theoretical errors: no explanation yet.
 In the new version of SCAMP the error added to the the blended objects in the catalogues has been increased, which increases the overall accuracy of the solution
 In the end it seems that there is a 4mas source of white noise which we dont understand

Sextractor (Emmanuel)

 In the new version of SExtractor, based on simulations, the positional accuracy for faint galaxies and stars is the same.

PI data
 On the 25 August all data will be removed from the nodes: Henry has until that date to list all data to preserved.
 Note: All PI data should pass through Fred and Yannick for packaging so that the format distributed to users is as homogenous as possible.

Pipeline development (Laurent)

 After 25 August all development relating to the T0003 will be frozen.
 Configuration files will also be frozen after this date.
 Side question: what is the status of the local USNO catalogue? We have to be able to process data without depending on access to the CDS.

 JC has to check with Laurent that TERAPIX "internal" access works. There are two ways to handle this: Either Emmanuel modifies SCAMP accordingly or Laurent includes it into the pipeline command lines/ configuration files.

 This needs to be decided as soon as possible, and no later than August 25!!

Status of orders
 mix6 has been ordered and could arrive from the week starting the 17 August
 8tb storage machine also ordered
 cosmix3 (4xcore) has been ordered;
 cosmos4 (clone of cosmix2) w. 1Tb (this machine will be paid for by J. Bergeron and A. Omont)
 order for the PCs not ready yet because of the re-organisation at LDLC
 The 24" monitors should arrive at the end of August.
 No UPS have been ordered for these new machines.

Defectix and double pass mode (Chiara, Yannick, Antony)
 How can we remove the sources from the wircam images for background estimation in the WIRCAM? pipeline?
 Yannick suggests that DEFECTIX could be used for this purpose. Some evaluation is need to see if this the best approach.

Other Matters
 Distortion in the MEGACAM images has been completely removed from the instrument by reversing the L2 mirror and adding a 2mm spacer/
 Masks for the deep area for release candidate T0002 should be added.