Andrew's Pixel Problems

[Tom Droege <tdroege2@earthlink.net>]

In order to survey the whole sky, one has to cover a large area in each 
image.  Otherwise it will never get done. This is what determined the 
design of the Mark IV.  Get as much silicon looking at as much sky as was 
possible within my budget.   As an amateur effort, we already have more 
silicon area looking at more sky than any similar (amateur) effort that I 
know of.  I think we beat many professional surveys.  There is every 
indication that we can keep at it longer than a professional can 
afford.  The Mark IV design produces star images that cover only 1 or 2 
pixels.  That is just what happens.  I really don't see any solution to 
this.  We could defocus, but that may produce other problems.

So what to do?

I don't see much that we can do.  I think that we should do what we can do 
and get on with it.  Measuring the whole sky in a consistent way has some 
advantages.  There has been some discussion on the AAVSO group on filter 
transformations.  This is another insolvable problem.  Doing the whole sky 
with the same set of filters and optics at least allows comparison of one 
star with another *in the same survey*.  As I see it, filters are an effort 
to characterize a spectrum by taking the energy in a few regions and 
measuring it.   If you have two stars with the same spectra, then I can 
imagine transforming one set of filters to another.  It is easy to imagine 
stars with different spectra that just do not transform.  The result is 
that you compare two measurements taken with different filter sets at your 
peril.  (This probably holds for different optics or for different 
locations or for different regions of the sky to name a few.) .

Thus I see a use for a whole sky survey which contains variability 
data.  It should allow comparisons.  Others seem to back me up in 
this.  They want the data.

Listening to what Andrew is discussing indicates that it will be really 
hard to detect planets with the Mark IV.  Eclipsing binaries will also be 
difficult.  We will have to take a lot of measurements of each star to sort 
these things out.  In the end we will probably only generate lists of suspects.

At the other end, we can probably do a pretty good job on the slowly 
varying red stars.  A large catalog of these should be useful.  Here with 
the present data collection scheme we have 56 measurements to average.  If 
we take such a set once a year, then we should generate measurements of 
high precision.  Here we can use tricks like throwing out n high and n low 
readings to avoid cosmic rays and weak pixels.  For this some movement from 
frame to frame is desirable.

My present data taking scheme was meant to be a "straw man" strategy.  Do 
something so we can find what is  wrong with it.  Possibly experts could 
devise an absolutely perfect first strategy.  I doubt it.

I am only slightly discouraged.  I think we should do what is possible with 
this set of optics.  I think it will be a worthwhile catalog.  It will just 
not do everything that we might like.

I thus argue that we should get on with the analysis with the data we have 
now.  Possibly we can do better in the future.  I keep hoping that an 
individual or group will step forward and say "let me have your 500 disks 
from the engineering run and I will reduce it to a catalog."  If not, in a 
few months I will start pushing it through Michael's pipeline.  I just have 
to learn a little more Unix.  I would do this at the risk of not producing 
any more Mark IVs.  I doubt that I can both build Mark IVs and run a 
pipeline.  I feel strongly that I must start getting catalog data.

Tom Droege