Re: Analyzing Data

[Tom Droege <tdroege2@earthlink.net>]

Patrick and all,

I will no doubt get these stars on other nights.  Then one has the problem 
of night to night variability for who knows what reason.

What I think we need is a progressive refinement process.  First look at 
the single nights and make up lists of "possible variables".  Then this 
list can be small enough that a person can look at several combined nights 
and see if a star still looks "possible".

Note that I did not just look at a814.  When it looked promising, I looked 
at a bunch of stars around it of similar magnitude.  These were relatively 
constant.  I think any search algorithm should do the same.

My first goal is to get a very large list of measurements.  Most of these 
are pretty boring.  Then I would like to flag the interesting 
objects.  Then make everything available and let interested parties make 
their own decisions about what to do with the data.

Hmmmm!  Perhaps the best thing is to just make two lists.  One with 
everything, and one with the constant stars deleted.

>The two other stars show a clear trend (they also have a high Spearman
>RCC), but there should be some confirmation of this on other nights.  I
>think the "cost" of observing a single star like this in detail with a
>telescope is too high (there are not that many active
>telescope/observer combinations!).  There are other stars which look
>more promising for detailed study at this time.  In my opinion,
>"confirmation" should first come from surveys like TASS which can deal
>with a large number of stars simultaneously.

I did not mean to imply that these stars were candidates for 
observation.  I was just investigating why most stars found were below mag 
11 or so.   There should be more at fainter mags since there are more 
fainter mag stars.  So I think they are there if we try to find them.  I 
think we should try to find them.  I think we should further try to screen 
them to try to pick up any possible spectacular object.  So I think we need 
to look at the faint end pretty hard.

>Besides that, if I get to adapt the WS script to include other
>variability indicators in a satisfactory way, I will certainly send it
>to the list.  If the number of available observations for each star
>were, say, doubled, the task to identify variables would be a lot
>easier.

These three stars were picked from a single exposure series.  We should get 
them several more nights.  I have over 300 observations of some stars.  All 
data is not perfect.  Some nights are noisier than others.  Some 
observation series for one night are offset from an observation series for 
another night for some stars.  Other stars for the same night do not show 
the same offset.  Is this a true variation or instrumental?  How do we 
tell?  Just because star A is constant for night 1 and night 2, and star B 
has an offset between night 1 and night 2 is not enough for me to say that 
it has changed.  At least not yet.  I have to be pretty certain that I 
understand everything.

I sure hope that some of you will take on thoughtful analysis of the 
data.  I am still busy building cameras and such.   I would like to get a 
paper out.

Tom Droege

At 03:50 PM 8/6/02 -0700, you wrote:

>--- Tom Droege <tdroege2@earthlink.net> wrote:
> > Here are three files to try out better search techniques.  I recall
> > they
> > had WS statistics of 2.5, 2.0, and 1.2.
> >
> > They all look like they are short term variables to me.  a814 seems
> > the
> > most questionable yet had the highest ws number.
> >
> > I am hoping that one of you (Patrick?) will provide us with software
> > that
> > will dig stars such as this out of the crud.
> >
>
>
>For star 814 there are simply not enough observations (12 only).  The
>relatively high WS statistic is due to the two fainter observations at
>the end.  Removing those reduces WS to 1.1.  In this case, WS (and
>probably all other indicators as well) is not very reliable.
>
>The two other stars show a clear trend (they also have a high Spearman
>RCC), but there should be some confirmation of this on other nights.  I
>think the "cost" of observing a single star like this in detail with a
>telescope is too high (there are not that many active
>telescope/observer combinations!).  There are other stars which look
>more promising for detailed study at this time.  In my opinion,
>"confirmation" should first come from surveys like TASS which can deal
>with a large number of stars simultaneously.
>
>Besides that, if I get to adapt the WS script to include other
>variability indicators in a satisfactory way, I will certainly send it
>to the list.  If the number of available observations for each star
>were, say, doubled, the task to identify variables would be a lot
>easier.
>
>Patrick
>
>
>
>__________________________________________________
>Do You Yahoo!?
>Yahoo! Health - Feel better, live better
>http://health.yahoo.com