Summary of dicussion during chop-nod analysis telecon of Dec. 12, 07

Hiroko, Lero, Mike, John, Giles

_________________________________________________________________

reduced-chi-squared procedures:

Mike reported that these are all written, and are currently being
debugged.

Mike brought up one detail that was overlooked in Giles' original
formulation of the problem (see analysis logbook) which is the computation
of the errors in Q and U.

John outlined the following procedures in sharp_combine:

compute Q and I -> compute q (propagate errors) -> smooth and bin q

How is Mike supposed to go backwards and compute the error in Q from the 
smoothed q and its error?  That is the dilemma.  Mike favors the solution 
of simply inverting John's error-propagation equation, but this ignores 
the smoothing step so its a bit dicey.

We decided that the errors in I probably make an insignificant 
contribution, since they are calculated from frame-rate errors.  So this 
means that we can probably ignore this problem.  What Mike is doing should 
be quite acceptable.  (We may want to revisit at a later time.)

John suggests further work on applying GAP to the i.p. analysis, when time
allows.

_________________________________________________________________

IRAS 20126 data from August 2007:

Hiroko reported progress as follows:

she tried "-bg 1 0" vs. "-bg 50 0"

for some reason "-bg 0 0" doesn't work

John reports that this is a known bug and the work-around is to not use
any -bg flag at all if you don't want any background subtraction.

Hiroko reports that with no pointing or tau correction, the peak is 18% 
lower than if we do apply the pointing and tau correction.  This is true 
regardless of whether you have 1 or 50 iterations of background 
subtraction.

next Hiroko will re-do this test, but this time introducing the two 
corrections one-at-a-time, and will also pay attention to the width of the 
peak, as well as its height.

Hiroko noted that some of the really weird discrepancies seen in the 
mountain-top analysis (3-sigma vectors from different days intersecting at 
right angles) have gone away after removing obviously bad data (bad hwp 
positions).

_________________________________________________________________

M82 data from four runs in 2006 and 2007:

Lero is working on compiling Darren's code for gaussian fits on his Mac.

Hiroko noticed we have no RGM for late April and will work on this 
problem.

Giles and Lero to confer on (a) intuitive reduced-chi-square and angular
discrepancies (b) what is the effect of an error in assumed coordinates
for M82 (c) advanced features in polsharp.

____________________________________________________________

The following items were not discussed on the telecon:

NGC 1333 IRAS 4A data from November 2007 (see Nov. 21 summary)

Mike's plans to make a smoothed tau for November 2007 

Giles' plans to work on i.p.

linked list-serves for SHARP

_____________________________________________________________