Light curves in the IVOA
seaman at noao.edu
Tue Jun 22 19:29:45 PDT 2010
> the usual conflict between doing it right and doing it well enough....
System engineering is an exercise in the art of satisficing, not optimizing. There is no conflict. Doing it right is the same as doing it well enough. A Platonic ideal of the virtual observatory is an oxymoron. The question is one of impedance matching to maximize the effect of our activities.
> We need low cost of entry. We need users, adopters, and a clientele. We
> also need to be prepared for when they get frustrated with their limited
> system, ready to provide the next step up.
Precisely - get 'em hooked and then net the big fish.
> People like Josh Bloom at UCB are doing this stuff and making it work. We would be stupid to ignore this.
> Maybe Tweets (140-characters) are a relevant example. Huge buy-in, limited
> content. (Full disclosure: I have never read nor posted a tweet, and I
> cannot imagine why I would ever want to do either.)
Tweets are "natural language" - not a good metaphor for any of our XML standards. VOEvent packets are brief, but can be richly constructed out of simple building blocks.
> From a fundamental point of view, I don't like Roy's or Rob's suggestions,
Roy and Rob were suggesting Josh Bloom's SimpleTimeSeries be used for VOEvent. I'm not sure what you don't like, since you just suggested the same thing :-)
By all means pursue broader standards as well, and VOEvent will link to them externally when they are available. It would be inappropriate, however, not at all "doing it right", to embed extremely complex semantics internally within a VOEvent packet.
> and do like Doug's, but I am guessing that this is probably what we need to
> do. Doug pointed out something very important, however: significant
> collections of TD data is out there now, compliant with SSAP, and one can
> interpolate that is was not all that difficult. Perhaps providing some
> tools to map simple time series data into SSAP could bridge the gap.
John Brewer, in Josh's group, has been working on such tools.
> On 6/22/10 7:35 PM, "Rob Seaman" <seaman at noao.edu> wrote:
>> In particular, VOEvent use cases emphasize the "Simple" in
>> SimpleTimeSeries. A primary goal of v2.0 is a schema that is hardened
>> (and validate-able) for battlefield usage by third party troops. Much
>> effort has gone into SimpleTimeSeries to make it compatible with that
>> On Jun 22, 2010, at 4:07 PM, Roy Williams wrote:
>>>> Not surprisingly I support Raul's suggestion that a time series
>>>> interface be based upon SSA and its underlying data model.
>>> I entirely agree that the formats derived from the SSAP and STC
>>> should be subsetted for light curves and other relevant time series.
>>> I would also suggest that IVOA endorse other representations that
>>> are already in use. I would like to suggest the simple time series
>>> from Berkeley: http://dotastro.org/simpletimeseries/
>>> It is closing in on compliance with STC, and is expected to be
>>> incorporated into VOEvent.
More information about the dal