Re: Data Sets for Data Reduction

[Tom Droege <tdroege@veriomail.com>]

Andrew,

First, I just want to get some data reduced by any pipeline.  By this I 
mean an amateur paper.  This will show up a lot of problems - like what to 
actually put in the fits headers.

I would like to spend no more than a couple of weeks in discussion about 
how to do this, then I would like to get on with a first pass.   OK, I will 
make a goal.  A paper for IAPPP.  "Results of Processing Test Data from the 
Prototype TASS Camera."  ( his title will be changed.)  This does not say 
what the result will be.  We will know that after we process the data.  In 
my usual style, I like to write about what we did wrong as well as what we 
did right.  So some of the emphasis should be where we found the errors and 
what the problems are with the hardware, and (hopefully) how to fix them - 
or at least how to design a better Mark IV.  Then we should also present 
some really good data.

I would be delighted to have the data run through two pipelines.  It is 
much more than I could possibly hope for.  Perhaps the way to do this would 
be to split the data into two stacks (there is enough for a third stack if 
a third pipeline volunteer shows up) and then have the parties duplicate 
their stack and pass it on to the other group.

There are many wonderful features to having two (or more) independent 
pipelines:

1)  The results have a check.  (Still possible to be wrong)
2)  The results can be measured against each other for quality.  It will 
possibly point toward the best method.
3)  It forces some standards

At the moment, I consider 3 most important.  But 2) will be fun as the 
pipelines vie for perfection.  Also good science, I believe, as I can see 
little harm in trying to get the best possible result.

Tom Droege



At 08:16 PM 11/5/00 +0000, you wrote:
On Sat, 04 Nov 2000 10:36:01 -0600, you wrote:

 >I have recruited the start of a Data Reduction Group.  We will try to get
 >some useful information out of the data I have accumulated.  This is
 >represented by two growing data sets.  One which I will call T was taken
 >with TOM.  The M data set was taken with MICHAEL.

I would like a data set.

1) If there is any sign of a concensus (yes - I know: herding
cats!) that the Data Reduction Group is going to process one
set by each of the competing pipelines, then obviously I would
like a copy of that one set. May I suggest T? It overlaps the MKIII
stuff and I already have most of the catalogs sorted out ...

2) I can copy CD's (given time) so I could be part of
a distribution network. 2 sets for one would pay my dues
even if I didn't get the analysis done! (full PSF fitting
is running about 2 images a day average - time for a faster
computer or a more efficient program.)

Andrew Bennett
80 Avondale Cross Rd.,
RR2 Newport NS B0N 2A0
Canada

Andrew Bennett, Avondale Vineyard, Nova Scotia, Canada.