Re: GSC 3493-1324

[Thomas Droege <tdroege2@earthlink.net>]

Looking at the tass error curve, the expected one sigma effect for a mag
11.3 star is about 0.07 mag.  One does not get very excited by 3 sigma
deviations.  In this data base, millions of measurements will be 3 sigma
away from the "true" value.  So one could be just "lucky" for a few
measurements.  So 0.2 mag is just not very significant for this data.  The
data you are comparing to probably has another error.  Combined, the data I
have seen just does not seem significant.  Interesting, perhaps, and enough
to get you to take out a telescope and make relative measurements.  

The way we are reducing the tass data (Michael please comment) gives us a
pretty good absolute photometry and tells us the error when we compare it
to enough standards.  I have done this using Brian Skiff's big photometry
table.  When I compare all the stars in Brian's table against what the tass
pipeline says they are, the error comes out consistent with the curve in
S&T 10.  I need to write this up.  A good project for a string of cloudy
days.  

There is another way we could take data.  (We have done this).  We could
follow a star as long as possible.  Then the relative error from
measurement to measurement is much smaller. (It can be as low as 0.005) 
However this does not tell us what the absolute measurement is.  When we
follow the star on a different night in a different frame position we might
get a curve that does not line up with the earlier one.  That is if we
follow a fixed star on two nights in two positions in the frame then there
can be a night to night offset.  This is due at least in part because we do
not have enough well measured stars in our reference catalog.  

I think it is misleading to present data with a small scatter as if the
scatter indicated the absolute accuracy.  I think the present way the data
is presented is more indicative of it's real value.  But it does not look
as nice.  I suppose we should round off  the data to two places to better
indicate the accuracy, and possibly to one place above  mag 11 or so.  The
decimal places are there for the computation.  One should read what the
author's say about the data before using it.  (Or in this case badger them
to write a difinitive paper.  ;^)  )

Seems to me there are plenty of stars in the tass data base where the
variation is 6 or 7 sigma and that would be significant.  OK, use the tass
data to give a hint of variability and then go out and take relative
measurements against a single star to tie things together night to night.  

The tass pipeline does not give results that allow a relative comparison,
so you have to do a lot of work if you want to pull relative measurements
out of the tass data.  I am not even sure it can be done without getting
the raw images.  (Michael comment?)  I hope that these can be made
available some day. 

Let me try to clarify this.  Suppose that we have two fixed stars that we
want to measure.  On day one, we measure the two stars against a set of
catalog stars and get a certain difference between them.  On another day,
we measure the two stars in a different position in the frame.  This
results in using a different set of catalog stars to measure their
magnitudes.  The result is that the difference for fixed stars can change. 
(A lot more than indicated by the statistics of the measurement.)  At least
part of the error in S&T 10 is due to this effect. Many of the reference
stars have errors in the range of 0.1 mag.  Further, we are using B and V
to derive an I reference since I data is just not available. So even though
we use a lot of reference stars, we can be unlucky on any one measurement. 
So we have an error in absolute measurement due to the errors in the
available catalog sets.  There are also other errors which we do not yet
understand.  

One way to get at the real picture is to take a lot of data.  Then we cn
try to reduce the errors by calibration.  This is the reason that I am
taking data in the way that tass Batavia is being run.  I am open to better
schemes. 

Tom Droege 




> [Original Message]
> From: Sebastian Otero <varsao@fullzero.com.ar>
> To: <tass@listserv.wwa.com>
> Date: 6/6/2004 1:03:15 AM
> Subject: Re: GSC 3493-1324
>
> > A one day error is not likely for this date.
>
> Well, I think it is more unlikely that the observations fit exactly (and I
> mean exactly) at the secondary eclipse (as Patrick said).
> The observations are made at HJD:
>
> 3110.8440 11.275
> 3110.8456 11.259
> 3110.8473 11.278
>
> instead of
>
> 3109.8440 11.275
> 3109.8456 11.259
> 3109.8473 11.278
>
> These are already converted to HJD. It would be a good idea to make this
> convertion beforehand as in the ASAS database. (NSVS is worse: MJD !!
> ;-((  )
>
> See the magic happening in the new lightcurve.
>
> Actually this is good news: TASS data weren't bad, on the contrary, they
> were very good datapoints so you were right on this regard and I was right
> on the period.
>
> So I think the discussion is over. We all learn something from here.
>
> Cheers,
> Sebastian.
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> That is nothing funny has
> > happened to the computer clocks this year except that  I have caught the
> > TOM1 computer jumping several hours.  This seems to happen when the
> > computer has been sitting idle for a day or so.  If the computer jumped
> > several hours, then the pipeline would simply not find stars.  This also
> > holds true for a one day off clock.  The star fields would be unlikely
to
> > match close enough and no data would come out of the pipeline.  Since
> there
> > is lots of data, one can look at known variables in the data and compare
> > them to known phases.  Should give a clue.
> >
> > While anything is possible, I think a one day error is unlikely.
> >
> > Tom Droege
> >
> >
> > > [Original Message]
> > > From: Patrick Wils <patrickwils@yahoo.com>
> > > To: tass@listserv.wwa.com <tass@listserv.wwa.com>
> > > Date: 6/5/2004 8:41:36 PM
> > > Subject: Re: GSC 3493-1324
> > >
> > > >   I looked at the sample comparison star Michael mentioned,
> > > > GSC 3493-1088, on the same night (and other nights).  It does
> > > > NOT become dimmer on the night of JD 2453109, in either V or I.
> > >
> > > Well, maybe the magnitudes are OK, but the date may be off.  If those
> > > three points are shifted 1 day to JD 2453110, they fall exactly in
> > > phase with the secondary eclipses according to Sebastian's period !
> > >
> > > Patrick