Summary of dicussion during chop-nod analysis skypecon of March 12, 2008

Mike, Hiroko, John, Lero, Giles

_________________________________________________________________

skype:

this time it was not echo-ey.  I guess headsets help.  

_____________________________________________________________

NGC 1333 analysis (Nov. and Dec. 2007 runs):

Mike gave us an update on the status of things and we discussed where to go from here.  The main 
items of progress are that the detailed analysis shows that we are OK on item 3, and that Mike is 
working on getting crush going locally (this presumably relates to item 5).  Mike will send his 
latest report on item 3 out to (at least) Giles and Martin soon.

The numbers in the previous paragraph refer to the summary of "remaining issues" given in the 
previous skype-con summary.  Giles suggests that Item 2 can probably be ignored, and that Items 4 
and 5 may resolve themselves naturally once we look at Item 1.  So we are closing in.

John notes that for Nov. we are at 50% dead pixels, for polarimetry.  Giles expected 30%, so maybe 
we should look into this someday.

Lero asks if we can avoid areas of dead pixels.  John explains that locating point sources in good 
areas does not require "math at 14,000 feet".  So this in principle is feasible.  Giles suggests a 
software update to the IRC script that would allow one or two corners of the dither to be "skipped 
over".  This goes on our list of future software updates.

_____________________________________________________________

M82 analysis - April 2007 runs:

Lero sent lots of new results for one day in early April (3rd UT).  He got rid of the "weird 
stuff" mentioned in the previous skype-con summary.  But he did this by restricting to 4-6 files 
out of 16.  Of the remaining files, Lero got rid of them because they were either near the edge of 
the array or located in a "sea of bad pixels".  (This "sea" is several rows in bottom half of 
array.)  Giles suggests that when M82 is lost in the "sea" the data is worthless anyway, but when 
M82 is simply near the edge of the array it may be OK to use FAZO and FZAO from adjacent (in time) 
files.  This could be good data that can be salvaged.  Pointing drifts on 20 minute time scales 
may be quite small.

For reference, we decided 3 arcsec pointing errors are tolerable.  Better to keep such data than 
to chuck it.

Note that the ps files Lero sent are smoothed.  In the fits file one can access the original 
unsmoothed data.  John points out the sometimes the unsmoothed data has the "drizzle map" effect.  
Giles thinks the four-file case shows two peaks in the unsmoothed fits but not in the smoothed 
image (the one given in the ps).  John suggests using the unsmoothed I (or less aggressively 
smoothed I) as a reference while working to get the FAZO and FZAO right.  Once we have FAZO FZAO 
we can smooth more to get more vectors.

Issues with background subtraction were not discussed (see previous skype-con summary).

_____________________________________________________________

not discussed on skypecon: sharpcombine on mac