[TDA_wg] Outline of discussion at Hotwired IV
ahowell at lcogt.net
Thu May 21 20:39:09 EDT 2015
I have strong feelings on the naming issue since I've done it many
different ways, and I'm partly responsible for this naming mess.
At SNLS I helped come up with our system there. Then when I went to PTF I
pushed for a similar system, then I did the same at Pan-STARRS. However,
now that we have too many names, I'm arguing for a new and more regular
1st - reason the IAU system doesn't work.
SNe get too many names. First they often get an internal name with a
survey, then they get a PSN name (based on coordinates) once a CBET is
issued, then they get a supernova name if they are confirmed as a
supernova. This is too much to keep track of in a large survey when you
are dealing with hundreds of supernovae that could have several names
each. You start off in emails discussing something under one name, and it
may even get written into image headers, then people start calling it
something else (and more data may be taken under the other name), and then
for publication you switch everything again. Also, arguing with Dan Green
over hundreds of transients would be a full time job. Not to mention the
fact that the IAU circular response time is impossibly slow for the modern
This is why we went to the SNLS/PTF/PS1 system -- every new transient gets
an automatic name, one that it keeps all the way from detection to
publication. We don't need to have SN right in the name -- we can keep
track of this in a database.
Coordinates don't work well, because they are too long and people shorten
them arbitrarily. Then it is hard to find discussions in email. Also,
coordinates to some surveys with big pixels are not accurate. Amateurs can
also get coordinates wrong. Finally, sometimes transients are discovered
but their coordinates are not known well (previously GRBs, but soon
gravitational wave events).
Studies show that people can keep about 7 pieces of information in their
brain, and it is even better if it is "chunked" into substrings or
different kinds of info (e.g. why phone numbers and social security numbers
are hyphenated, license plates have numbers and letters, etc.). This is
why we went with, e.g. 15abc. That is the most compact way to write
something, and strings of a few letters are easier to remember than a long
list of numbers. Plus, it allows you to separate out the year (easy to
So to me the requirements a modern naming system should have are:
- Be as compact as possible so that people can remember / not shorten. You
can store types and coordinates in a database.
- Be issued immediately upon discovery so that you don't have to change
- Not be based on the transient subtype so that again you do not have to
To achieve this, I imagine a name server. You send it a discovery at a
coordinate with an error. If a transient exists, it returns the name. If
not, it issues one. These would be in the form 15abc (etc.) start with
15a, add letters as needed. This is to be decided, but in principle
surveys could still append their name on the front, and then when we find
out something is a supernova, we could replace it with that, e.g. PTF 15abc
would become SN 15abc.
On Wed, May 20, 2015 at 3:40 PM, Allen W. Shafter <awshafter at gmail.com>
> This seems like a sensible solution/compromise to me.
> I agree that a longer, formal name of the form [survey][UT date]:[RA, Dec
> in J2000] with both temporal and spatial info will be important for machine
> handling, and for transient follow-up (nice to have date of discovery in
> addition to position readily available). For more informal reference, a
> link for translation to a shorter name as suggested would also be useful. I
> don't think a single naming scheme can be satisfactory for all situations.
> On Wed, May 20, 2015 at 2:20 PM, George Djorgovski <
> george at astro.caltech.edu> wrote:
>> Folks, for what it is worth, we went through the naming of events with
>> the PQ survey, and continued with CRTS. Since events - as opposed to
>> objects with which they may be associated - are spatiotemporal things, we
>> though that the naming should reflect that. We came up with a clunky
>> [survey][UT date]:[RA, Dec in J2000]
>> For example: CSS150520:132803+085623 is an event that was observed by the
>> CSS survey on UT May 20, at those coordinates. Given the typical surface
>> density of transients, that is plenty unambiguous, and obvious extensions
>> of a precision (time or coordinates) could be added if needed.
>> This format was designed to reveal the useful information, as opposed to
>> the now prevalent XYZ15abc format, which hides most of the information, for
>> example “can I observe it tonight?". However, it is too clunky to say, so
>> conversationally we would say, e.g., “we observed 150520”, because there
>> aren’t too many interesting ones that get a lot of attention, but always
>> use a full designation in the publications. On the other hand, XYZ15abc is
>> very easy to pronounce and remember.
>> Even with the sources (not events) a common practice is to use nicknames,
>> e.g., we would say “we observed SDSS 1435” instead of its full official
>> name, SDSS J143507.47+132117.5, and in a given context everyone involved
>> would know what we mean.
>> I guess we could have both, with a translation easy to find somewhere,
>> CSS15pqr = CSS150520:132803+085623, which then may also get additional
>> names, e.g., = SN2015xy
>> The long names are for the computers, the short ones for the humans, more
>> or less.
>> It is probably safe to say that whatever the IAU decides, astronomers
>> will do whatever they please…
>> Cheers, George
>> TDA_wg mailing list
>> TDA_wg at timedomainastronomy.net
> Allen W. Shafter, Chair
> Department of Astronomy
> San Diego State University
> San Diego, CA 92182-1221
> Ph: (619) 594-6170, Fax: (619) 594-1413
> TDA_wg mailing list
> TDA_wg at timedomainastronomy.net
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the TDA_wg