Hi Ed,
Thanks for an expert view. Most of waht you say illustrates well the need not to make definitions rigid - that way they cannot be contradictory - and there has already been plenty of development of 'ranking' algorithms so that it should be easy to avoid ending up with everything as both a sub- and a super-set of everything else...
cheers
a
On Mon, 5 Mar 2007, Ed Shaya wrote:
>
>> Yes, it is - and I don't think that the IVOA should attempt to create rigid
>> definitions when there are many existing practices which make sense but
>> only in limited contexts. For example, I would say (following Padovani et
>> al. 2004 and refs therein) that anything with an X-ray luminosity > 10^35 W
>> and a very hard x-ray photon index is an AGN *in the GOODS fields* and
>> anything with an X-ray luminosity > 10^37 W is a QSO. If an object is only
>> detected in the X-ray, then there is no distance information to allow the
>> luminosity to be calculated, but the hardness of the X-ray photon index may
>> mean that you can define it as an AGN even though no optical host has *yet*
>> been detected. After all, a non-detection is only relative to some
>> arbitrary set of observational sensitivities.
>
> "define it as an AGN"? I think you need to differentiate between making a
> tentative identification and defining. Sometimes you can make a good guess
> at the type if you know only the luminosity in one band and have a
> statistical reason for arguing most sources in the field are at z>1 or so. A
> fleshed out ontology would tell you what other classifications are reasonable
> and under what conditions the identification would be correct (eg, "as long
> as z>1.6 and B<15, and variability is < .2 mag, then this is certainly a
> quasar").
>>
>> Hence I would say that it is probably uncontroversial to make QSO a sub-set
>> of AGN? I have never come across the use of QSO to describe something
>> which could not also be described as AGN although the converse is not the
>> case?
>
> I have. Primordial black holes have been hypothesized by several illustrious
> astronomers and a supermassive one feeding on the IGM would probably act like
> any other QSO. It just would not have a galaxy. Arguments for such objects
> are reasonable since it does seem quite hard to accrete so much matter in
> time. Recently, there have been several papers (just Google on "primordial
> supermassive blackholes") contemplating the idea that old elliptical galaxies
> began life as these primordial blackholes that later gathered a galaxy around
> them. I guess that this possibility, though highly speculative, has been the
> underlying reason for keeping the term QSO disjoint from AGN.
>
>> Another example which only makes sensse in a limited context is "stars are
>> points, galaxies are extended" which is fairly accurate (pace QSOs!) for
>> some optical data but falls down for e.g. X-ray surveys of distant objects
>> (or even some optical high-sensitivity, but not-so-high-resolution surveys)
>> where everything except nearby galaxies is point-like - and for
>> interferometry which can resolve some stars...
>
> Again, this is probabilistic identification. This time because there are
> only upper limits on the sizes and no distances. So, yes ontology alone is
> not useful for identification when you have insufficient data to positively
> identify. I don't get the point of coming up with use cases that are too
> hard for the human mind let alone a simple algorithm. On the other hand, if
> there is data that provide an upper limit of 2e8 km diameter, and a M_B =
> 4.8+/-.3 and B-V = 0.7+/-.2, then ontology could tell this is a G2V star.
>>
>> So either we have to add wavelength and resolution to any definition, or
>> (probably more practically) allow multiple definitions and build up a tree
>> of supersets and subsets based on accumulated application to real
>> catalogues etc., not the prejudices of those of us on this mailing list, if
>> that's possible.
>>
> Every astroObject has a position (set of coordinate values) in the
> N-dimensional observation space. We carve up this space into a reasonable
> set of subspaces which we call the object classes. Over most of this space
> the clustering of the properties demarcate for themselves acceptable ranges
> for each class. But it is true that in certain projections (ie, you have
> limited wavelength coverage, resolution etc) things can get messy.
>
>> Regarding individual objects, the IVOA's reasoning could _add_
>> classifications, e.g. QSO + extended optical galaxy at same position is
>> also AGN (although I would say that anyway as a case of inheritance) - but
>> that should somehow be distinguished from definitions taken straight from
>> the literature. But we should not _take classifications away_ even if they
>> seem to be contradictory (maybe within some time limit e.g. 20 yr).
>>
> I agree except for the last bit. We must take away contradictory definitions
> because the system will crash. We can deal with contradictions only in that
> individuals can have their private ontology versions that are somehow amended
> to their way of thinking. Our private versions might contradict each others.
>>
>>
>> Incidentally, microquasars are always referred to as such and are indeed a
>> separate phenomenon and not a subset of any of the above...
>>
>> best wishes
>>
>>
>> Anita
>>
>>
>> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
>> Dr. Anita M. S. Richards, AstroGrid Astronomer
>> MERLIN/VLBI National Facility, University of Manchester, Jodrell Bank
>> Observatory, Macclesfield, Cheshire SK11 9DL, U.K. tel +44 (0)1477 572683
>> (direct); 571321 (switchboard); 571618 (fax).
>>
>