Re: Modified WS

[jg <jg@jgws.freeisp.co.uk>]

WAHOOO!!!!

Lovely stuff!!!

I intend to bung the overlapping RA bits of "July" and "August" (when
latter comes out) together and beat the living daylights out the
resultant dataset with this new code ;)

Patrick,

Thoughts on min paired numbers of obs?

I was going to stick to 100+, which'd probably be 30+ per night, but of
course you can end up with several nights at ~30 and one night with only
ten, so this latter would be automatically excluded in your code, but
would the other nights still be taken????  In other words, your code may
automatically allow some pre-filtering to save me having to pre-process
the data.


NB HD 155229 data all saturated in "July" dataset.  I forgot to look at
this point.  Makes a big diff. to viewpoint.  However, if the Michaels
have already looked at it and have nothing better to look at at the mo',
fair enough.  [Let's rename it from TASS to 2MASS... ...oh, that's
already been used, anyway data's mostly Tom's at the mo' ;^)]

Cheers

John G.

Patrick Wils wrote:
> 
> I have modified the WS program (see attached file) so that it also
> displays the following, besides the existing WS statistic:
> - the farthest outlier (expressed in standard deviations). Observations
> which deviate more than 4 standard deviations from the nightly mean,
> have been removed for the other calculations.
> - the mean of nightly WS values (a minimum of 10 observations is
> required for a night to be included): this should be better suited for
> short period variables, because a difference in nightly means has been
> removed by this procedure, e.g. the WS value for the Mira star SS Her
> decreases from 85 to 0.7 !  But, I bet there is still a lot of room for
> improvement here.
> - the range in V (maximum nightly V - minimum nightly V) expressed in
> units of the maximum nightly standard deviation
> - the same for I
> 
> The last two values are large for long period variables.  It seems that
> the I value is often inexplicably large compared to the V value, which
> might indicate a problem in the night to night calibration.  The V
> value seems more acceptable.  The star in the July data set with the
> largest V range is John's eclipsing binary GSC 396-1222, and SS Her is
> in third place.
> 
> All suggestions for improvement are welcome.
> 
> Patrick
> 
> __________________________________________________
> Do You Yahoo!?
> HotJobs - Search Thousands of New Jobs
> http://www.hotjobs.com
> 
>   ------------------------------------------------------------------------
>                   Name: wsv31.pl
>    wsv31.pl       Type: Perl Program (application/x-perl)
>               Encoding: base64
>            Description: wsv31.pl