# The examples below follow the csh syntax, and for convenience, we'll also # assume the following commands take place in a $c2d directory. # Modify for your machine appropriately. # This script lives as genv/scripts/install_python set c2d=`pwd` # you may need to install your own python if you don't have root access to /usr/lib/python2.2 wget http://www.python.org/ftp/python/2.2.3/Python-2.2.3.tgz tar zxf Python-2.2.3.tgz cd Python-2.2.3 ./configure --prefix=$c2d make # takes about a minute make install set path=($c2d/bin $path) # add new python to path rehash which python # observe the new python location !!! # numarray (the new one) 1.0 is now also available # numeric is the old one, Numeric-23.3.tar.gz being the last one - don't use it cd $c2d wget http://aleron.dl.sourceforge.net/sourceforge/numpy/numarray-1.0.tar.gz tar zxf numarray-1.0.tar.gz cd numarray-1.0 which python # just making sure where it goes python setup.py install # pyfits (needs numarray) - notice the hardcoded python version path !!! cd $c2d wget http://www.stsci.edu/resources/software_hardware/pyfits/current/pyfits.py cp pyfits.py $c2d/lib/python2.2/site-packages/ # cfitsio (needed by pCFITSIO) cd $c2d wget ftp://heasarc.gsfc.nasa.gov/software/fitsio/c/cfitsio2500.tar.gz tar zxf cfitsio2500.tar.gz cd cfitsio ./configure --prefix=$c2d make make install # pCFITSIO (needs cfitsio sources, and numarray) cd $c2d wget http://panoramix.stsci.edu/~npirzkal/python/pcfitsio/downloads/pCFITSIO-0.99.2.tar.gz tar zxf pCFITSIO-0.99.2.tar.gz cd pCFITSIO-0.99.2 ### alas, you can either edit setup.cfg with the value of $c2d/cfitsio ### vi setup.cfg python setup.py build_ext --include-dirs=$c2d/include --library-dirs=$c2d/lib python setup.py install cp pcfitsio.so ../lib/python2.2/site-packages/
setenv CVSROOT :pserver:$USER@cvs.astro.umd.edu:/home/cvsroot (if you don't have a $USER account with us, use anonymous instead) cvs login (no password, only needed if you have not done this before) (anonymous uses no password, so just hit RETURN) cvs co -d genv SIRTF/genv cd genv ./configure (this should also report if you have a proper version of python) source genv.rc (now "genv" has been loaded in your current shell) (read the INSTALL file if you need to update your python installation)
more description to follow here: gls list c2d data on the TX archive gget get an AOR via the c2d index number gput put your possibly modified AOR back to the TX archive ...
CAVEAT: using just the corners, and not a full WCS description, can cause some RA/DEC selections to be inside, when they are really outside, and vice versa. A new scheme is being worked out storing a more complete WCS description (sans higher order SSC pixel correction terms that seem to be up to 1/2 pixels correction terms).
c2d_findsource $C2D_ARCHIVE/.corners.log 203.93 -75.94 1 0 2 or python -i $C2D_ROOT/scripts/c2d_findsource $C2D_ARCHIVE/.corners.log 203.93 -75.94 1 0 0 >>>Usage:
The arguments are: - a corners.log file (the ingestion team makes it, and will not be needed in a next version) - RA (decimal for now, H:M:S shortly) - DEC (decimal for now, D:M:S shortly) - bcd/icd/ecd identification (1=bcd, icd=2, ecd=3) - channel (0=all, 1,2,3,4 is the requested channel) - view (0=just compute and list, >1 means views this selected frame from the list)In the first example the command is only run once, which is terribly expensive since reading the big $C2D_ARCHIVE/.corners.log file takes 10-30", depending on your cpu. If you run it in interactive mode, the data will be in memory and much quicker to interact with. An example of the two will illustrate this:
ds9 & <--- you probably want to start ds9 before time c2d_findsource $C2D_ARCHIVE/.corners.log 203.93 -75.94 1 0 0 <--- a small benchmark Found 197652 fields in /sirtf1/ARCHIVE/TX_archive//.corners.log # Matched 36 fields bcd=1 channel=0 There are 36 frames 11.220u 0.300s 0:11.82 97.4% 0+0k 0+0io 646pf+0w <--- takes about 12 seconds on "preakness" python -i c2d_findsource $C2D_ARCHIVE/.corners.log 203.93 -75.94 1 0 0 <--- the same thing, for interactive work Found 197652 fields in all-corners.log # Matched 36 fields bcd=1 channel=0 <--- two python objects, a and b, were created in your session There are 36 frames >>> <--- this is the python prompt >>> a.show() <--- a is a python object, show() the function fieldname=all-corners.log nfields=197652 ra=13:35:43.200 dec=-75:56:24.000 nfound=36 >>> a.list() # Found 36 fields for ra=203.93 dec=-75.94 1 /home/teuben/sirtf/data/cham_archive//IRAC/BCD/0005741312/1/SPITZER_I1_5741312_0_0_1_bcd.fits .... 36 /home/teuben/sirtf/data/cham_archive//IRAC/BCD/0005741312/4/SPITZER_I4_5741312_9_0_1_bcd.fits >>> a.ds9(4) <--- show field 4 from this list >>> a.ds9(4,zoom=5,arcsec=4) <--- and with different defaults >>> help(a.ds9) <--- use python builtin help >>> a.find(204,-76,channel=3,bcd=2) <--- select only channel 2 and the ICD # Matched 6 fields bcd=2 channel=3 [.....] <--- a python list of 6 filenames >>> b=a.find(204,-76,channel=3,bcd=2) <--- do it again >>> b <--- list the 3rd one (python starts arrays at 0) '/home/teuben/sirtf/data/cham_archive//IRAC/BCD/0005741312/1/SPITZER_I1_5741312_2_0_1_bcd.fits'The current implementation of c2d_findsource uses the corners of a field (hence corners.log), which is ok for small fields like IRAC, but can still cause problems for high dec sources where the curvature was ignored. I'm looking into a better WCS based algorithm, but it's bound to be slower.