search

LSSTDESC / DC2-production

Configuration, production, validation specifications and tools for the DC2 Data Set.
BSD 3-Clause "New" or "Revised" License
11 stars 7 forks source link

Run metacal on dr1b to enable 3x2 validation test on Run 2.1i #372

Closed katrinheitmann closed 4 years ago

katrinheitmann commented 5 years ago

The 3x2pt team measured xi_plus from metacal analysis of Run 2.1.1i (25 sq degree). The results are shown below (capture from their Slack channel #desc-3x2pt). This is a very important validation result for DC2. However, the area for Run 2.1.1.i is rather small and the scatter in the test still very large. Run2.1i dr1b would increase the area by > 10x with the same depth and should help to improve the test results. The AGN problem of Run 2.1i is not of much concern for this test, so this will be very informative if Run 2.2i will be ok on the shear-shear measurement front. It would be therefore important to carry out this validation test before Run 2.2i production. The main task is to rerun metacal in Run 2.1i.

xip_plus_metacal

johannct commented 4 years ago

this is running at CC : https://srs.slac.stanford.edu/Pipeline-II/exp/LSST-DESC/task.jsp?refreshRate=60&task=53661482&refreshIsOn=true&refreshCount=39 weekly 19, new pipeline in order to correctly run post patch merging into tracts (and to solve the https server modification to the XML schema). Only metacal (no ngmix bd flux, which is very long to run)

johannct commented 4 years ago

Pipeline task is completed. Four tracts are missing :

@airnandez requesting transfer of rerun/metacal-dr1b-v2 and dpdd/coadd-dr1b-v1\:metacal-dr1b-v2/ under /sps/lsst/dataproducts/desc/DC2/Run2.1i/w_2019_19-v2

joezuntz commented 4 years ago

We now have 3x2pt measurements on 2.1i as well as 2.2.1i.

The lensing signal looks good! I'm looking into the GGL and clustering signals, which I'm confused about because one of my outputs says there were zero lensing galaxies, but does return a signal, so I've clearly messed up somewhere.

xi_plus and xi_minus plots below

xip_plus_metacal

xip_minus_metacal

rmandelb commented 4 years ago

Nice! The increased area has definitely made for less-ratty-looking signals. Looking forward to GGL and clustering as well, once you've solved the mystery of the sample with zero galaxies...

joezuntz commented 4 years ago

I have the GGL and wtheta plots checked now - see below.

The signal is pretty small in the GGL plot, which I think gives rise to the apparently noisiness and presumably comes from the poor redshift selection, which is uncalibrated. I can't check that hypothesis yet because we don't have a match to the input sample with the true redshifts.

gamma

w

joezuntz commented 4 years ago

Here's the estimated n(z), but that's completely uncalibrated and using a somewhat random trained model I had from earlier on:

nz

joezuntz commented 4 years ago

Hi all - the above plots are almost probably wrong, owing to an issue in GCR that I've described here (most recent comment): https://github.com/LSSTDESC/gcr-catalogs/issues/398#issuecomment-548458988

katrinheitmann commented 4 years ago

@joezuntz Is there an update on this? Thanks a lot!

joezuntz commented 4 years ago

I have a possible repair for the issues I've seen but still have to debug the new version, so nothing concrete yet, sorry.

katrinheitmann commented 4 years ago

Hi @joezuntz, any further updates on this? Thanks!

katrinheitmann commented 4 years ago

Since we have by now retired Run 2.1i I am closing this issue which has been quiet for ~6 months. We should open a new one for Run 2.2i and metacal as needed. If you feel strongly about keeping it open, please reopen it. Thanks!