Variable star data mining

[Michael Koppelman]

The most basic thing one can do is find new variable stars. This
generally involves looking at some standard deviation-like statistic
of the magnitudes of lots of stars and examining the ones that vary a
lot. If the light curve looks legit, it gets added to the list.
Existing variable stars are not examined. VSX is an excellent resource
to determine if a star is a known variable or not.

Next is figuring out the type of variable. This involves knowledge of
stellar astrophysics. By looking at the light curve, color, proper
motion, spectra and cross-correlation with other catalogs, you can
make an educated assessment of the variable type.

Then you can group them together and publish them in JAAVSO, IBVS or
similar publication. Make sure you double-check the literature and VSX
throughout the process. The higher density the paper, the better.
Papers about single new variable stars are rarely warranted anymore.
If any of your candidates are peculiar, you can do follow up
observations to try to nail it down. If you are lucky you might find
an object worthy of professional collaboration with large-telescope
spectra. Or you could solicit AAVSO observers or AAVSOnet for
followup.

There are many other approaches, too. One could take a look at the
intersection of catalogs e.g. x-ray sources within n arc seconds of a
blue star with a small proper motion. You could look at hot or cool
stars, near or far stars, disk or bulge stars.

One can data mine for data about existing stars as well. Poorly
studied stars can be researched with the same sources used to look for
new variables. You could, for example, look for period or color
changes or nail down the type on a star which has so far been
ambiguous.

So, for people new at this, which I am one, it's really just a matter
of getting familiar with how to interact with some of the data sources
available and practice some of these techniques. Find a whole bunch of
existing variables. Practice finding EW's or RR's. Play with cross-
correlating different catalogs. Download some catalogs and write some
code. Then when APASS and Kepler and LSST open their fire hoses,
you'll be ready.

Just got back from the AAVSO fall meeting. It was fun and it tends to
get one energized. I thought I'd take a stab at some fundamental
concepts in data mining as it relates to variable stars. Hopefully it
will help get some wheels turning.

Cheers,
M.

[Michael Koppelman]

Don't know if this has been linked here before but our friend John
Greaves pointed out this excellent information from our own Patrick
Wils:

http://www.vvs.be/wg/wvs/NSVS/Eclipsers.html

M.


On Nov 8, 9:44 pm, Michael Koppelman <michael.koppel...@gmail.com>
wrote: