[vsnet-chat 7657] Re My favourite Kepler light curves

[J K]

You'll find that most of "your" favourite light curves were published quite some time ago in a major eclipsing object catalogue that appeared also as a pre-print in astro-ph.

The initial data release for Kepler was almost exactly a year ago, I remember it as I waded through many light curves to find only roughly half a dozen variable stars.  Eventually I found half a dozen, one of which appeared almost instantly on astroph in two to three papers by several professional teams, all of whom had different interpretations of it.  In the end three were publishable as not seen by others and I submitted them at the end of December last year, needing to resubmit them in a slightly revised form after some discussion on variability classification later on when the GCVS team refereed the paper.  Basically I had the EA/RS star as EA+BY, but they don't like EA+BY for some reason, even though it involved a red dwarf.  Brian Skiff has commented elsewhere on the problems of classifying "microvariables" due to lack of precedent of such dataset availability for stars.

The main point being was that I wondered where all the variable stars where.  It was not only that I was finding less than one would find in random searching in general stellar fields, but on examination stars that had been found as eclipsing binaries by the ASAS Kepler Field paper were not in the data, nor interestingly enough were any associated with ROSAT sources.  Examination of the Kepler objects in a planetarium application (via Guide 8, using a dataset and TDF with Kepler lightcurve outlinking that I later made publicly available on the Guide 8 user list so any one could examine these curves) I found distribution of objects was even enough, and a field containing a known eclipser or ROSAT source would have plenty of neighbouring sources, but no source for them.

Well, a little looking around showed that there was no variable star team for Kepler.  At least officially.  However, quasi-officially the CoRoT variable star group had seemed to have been given first pickings.  This was fair enough, the Kepler team could do what they wanted with their data, however it looked like the "public release" of Kepler data wasn't quite as public as declared.

Well, assumption is one thing, fact is another.  So I emailed the relevant heads of the CoRoT team and said to them it was a bit peculiar there were so few, if any, eclipsers in the Kepler data, and they cryptically replied that there were in fact plenty of eclipsers contained therein, which whilst not admitting they were being withheld, revealed they had to be, because I'd found 4 in 2000 lightcurves, which is too small.  And indeed, later on they published their eclipsing catalogue, and all the data for those objects became available in the Kepler data soon after.

And most of the lightcurves available from this new so called "citizen science" planet hunter scheme are published in it, and are available for viewing online as solved eclipsers.  Although the shortest period and bluest ones aren't eclipsers but pulsators, as the paper itself admits the likelihood of.  The paper's on arxiv and as, you know arxiv, you should be able to find it yourself.

Meanwhile Ron Janson has been submitting new variable stars from Kepler data to AAVSO VSX for several months, mostly rotators and a handful of pulsators, probably mostly nonradial pulsators.

This is because, like I, he is using the MAST STScI server, which carries quicklook lightcurve plotting (improved from first incarnation) and provides direct access to the FITS lightcurve tables.

Akin to most zooniverse style "citizen science" planethunter merely has to use humans to wade thro' data because AI and neural networks have never really lived up to their promise.  But as is also typical, no identification of the object is given, and the finder merely does the donkey work, learns nor is taught any science, and has no means of checking the objects against simbad or vizier or VO in general, so extra data and information cannot be used to ascertain the nature of the object, and of course the Kepler Eclipsing Binary Catalogue is not mentioned either (which contains most of the lightcurves people are finding in planethunter.org and calling exotic).

You just report your find, and the professionals will then follow it up if you find one of the missing transiting planets, you ain't needed no more, even though you could easily do the analyses yourself (especially of the variable stars) using online resources and freely available downloadable analysis packages, via the MAST STScI Kepler data server.  The main Kepler site even shows and explains how TOPCAT can be used to access the fits files.  You can actually use TOPCAT to fold lightcurves, and rescale them on magnitude, to check phaseplots, though it ain't obvious how at first, it can be learned, with pointers from the Kepler Mission data analysis pages.  I think I gave a small walkthrough on how to do that on that Guide User mailing list post a long time ago too.  I did a similar one for how to find and play with CoRoT data.

Of course planethunter.org also does the same unuseful thing as most of the professional papers on Kepler data, it uses "relative intensity" (sometimes referred in some papers as "normalised flux").  This doesn't help in identifying variables when it isn't instantly obvious whether a dip in a lightcurve is a few millimagnitudes (which Kepler can achieve) or a quarter or more magnitudes, which Kepler can also see.

Indeed, the apparent flare in one of "your" curves looks very small relative to the rest of the curve, and Kepler lightcurves are well known for 'glitches' that emulate flares, as well as spikes caused by cosmic rays, so you really need to convert flux to millimagnitudes, and the time axis to days, whence it can be checked whether such is a glitch, or a real flare of appropriate few tenths of a magnitude or more amplitude with a hour or up to quarter day duration, or just some small centimagnitude artefact lasting barely minutes to an hour.  Rather than just trying to figure out what the hell is happening from unprovenanced, unidentified, objects with a lightcurve on them that has a little spike in a dataset well known for producing glitch artefacts (CoRoT lightcurves suffer similarly, btw).

But then you'd have to read up on the literature and the Kepler mission and MAST STScI data archive and webpages and know what you are doing.

There is one arxiv paper of variable star interest that you keep missing out, you know, this one :-

http://arxiv.org/abs/1003.5293

the case has some relevance to Kepler lightcurves, as Brian Skiff has noted, the identification of variability class for these objects will be problematic.  In the one to several day range it will be very difficult to ascertain whether the objects are varying due to rotation (eg starspots or reflected hotspots or just hotspots or even just uneven surface brightness), orbital circumstance (eg grazing eclipsers or ellipsoidal stars) or pulsations (eg SPB, gDor or similar stars), all of which can produce very low amplitude variation of regular periodicity.

If it's difficult at the decimag level, imaging how much more difficult it is at the centi and even millimag level!

J