[lsst-dm-stack-users] In response to Mario's request
Dave Monet
dgm at nofs.navy.mil
Mon Oct 22 09:34:39 PDT 2012
Here is a brief discussion of how far I have gotten
trying to use the LSST DM Stack on recently arrived
DECam images.
1) I followed the instructions on the Wiki back in
September about how to download (from the tarball)
and install the DM Stack. I got as far as
the final line, but this ended up taking more than
a week before I gave up on it. I followed Mario's
suggestion to do
eups distrib install -t Summer2012 obs_lsstSim
eups distrib install -t Summer2012 obs_sdss
So far as I can tell, this worked.
2) I installed git() from some free download site.
It needed something else, and it must be installed
without root privilege or putting it in obvious
places /usr anything. So far as I know, the installation
was successful. I can to the following:
lethe> git --version
git version 1.7.12.GIT
3) I found viz12-lupton-tutorial.pdf based on Andy's
suggestion that there was a Hands On session using
the DM stack at the Seattle LSST-Gaia meeting. I
got as far as page 4. In response to
import utils
I get a failure that this particular version of
utils.py is not in my path. However, it was
possible to proceed with the next few lines
without these functions so long as I was willing
to type. I didn't have the Hands On files, but
I have various other local files. I could do
import lsst.afw.image as afwImage
import lsst.afw.display.ds9 as ds9
im = afwImage,ImageF(some complete path)
So I think that some of the DM stack is present
and mostly working.
4) I moved on to other things, but had a desire to
return to the DM Stack to process the DECam images.
Robert Lupton and Paul Price pitched in, but Paul
sent me the appended note. To me, this reads as if
it were in a dialect of regional Martian that I simply
do not understand. I asked Paul for clarification, but
he was too busy. I have no idea what "tickets" are about
and how to update based on these. I have no idea how
to proceed with the git() to get the obs_decam package.
This just underlines the complete discontinuity between those
who work with the DM Stack on a daily basis, and folks like
me who need serious hand-holding. Honest. I would like to
be a Team Player, and I would like to use utilities that are
known to work and have support, but I just have no idea how
to proceed.
Pointers, examples, or clues would be gratefully accepted.
-Dave
-------- Original Message --------
Subject: Re: Plan B for DECam images?
Date: Tue, 2 Oct 2012 15:24:32 -0400
From: Paul Price <price at astro.princeton.edu>
To: Dave Monet <dgm at nofs.navy.mil>
CC: Robert Lupton the Good <rhl at astro.princeton.edu>
On Oct 2, 2012, at 8:49 AM, Dave Monet wrote:
> I hope Paul is taking notes about how he is feeding the DM stack. That would be an ideal introduction
> on how to do that for clueless folks like myself.
You need a relatively recent version of the stack. LSST will be pushing
something out soon-ish, or I could try to get you access to our HSC
build somehow. Then you need a couple of recent updates: tickets/2352
on daf_persistence and afw; and tickets/2353 on pipe_tasks; and master
on meas_astrom and pipe_base. Then get the obs_decam package
(git at git.lsstcorp.org:contrib/price/obs_decam.git).
Download the DECam observations (in MEF format) from the NOAO science
archive and run:
$ ingestImages.py decam /path/to/repo/ --create --mode=link
/path/to/data/*.fits.fz
That will suck the images into a new repository at /path/to/repo/ .
I've specified "--mode=link" which means it's set up as symbolic links
to your raw data; you can "copy" or "move" if you desire. If you've
already got files in the repo, leave off the "--create" flag.
Once that's done, set up your astrometry.net catalogue. I'm using
2MASS, courtesy of Dustin (if you need the configuration files for
running it with the LSST code, let me know and I'll hook you up; RHL:
it's on /tigress-hsm/astro/hsc/products/astrometry_net_data/2mass):
$ setup astrometry_net_data 2mass
Then reduce the data with:
$ processCcd.py decam /path/to/repo/ --id visit=135633..135662 -c
calibrate.doPhotoCal=False
The range of visits (135633..135662) is all the Monet 1 observations.
I'm turning photometric calibration off because I only have J-band
magnitudes from the 2MASS catalogue. If you only want to do a single
CCD, use, e.g., "--id visit=135633 side=N ccd=4" (to do the N4 CCD from
exposure number 135633). It looks like there's a bug in my geometry file
that's offsetting and chopping off 50 or so rows for the 'N' CCDs; I
hope to fix that soon. If you've got multiple cores and your office is
cold, you can parallelise by adding, e.g., "-j 8" to the command-line.
I hope to have some measurement files for you later this afternoon ---
I'm burning through the data now, and I'll go back and update the 'N'
CCDs if I can.
P.
More information about the dm-users
mailing list