I have a position, how do I find which ARG fields contain it (if any)?

• by JeanTate

  This question, in many forms, has come up many times in RGZ Talk. And the answer - from RGZ Team Members - is that it's not possible; there's no 'reverse directory'. But is it true? Is it truly impossible to get an answer (an ARG field, or 'there is no such field'), even for a weaker answer (e.g. 'highly unlikely that there is such a field', rather than absolute certainty)?

  I don't think so.

  And in this post I will show one way to answer this question.

  While there's no public info on how the ARG IDs are created - these are the names of the ARG fields, e.g. ARG00033zs - the first RGZ paper does say "The data is stored in a mongoDB database structure with each click on the image recorded for each step", so it may be that the centers (positions) of the ARG fields, and their IDs, are also recorded in a mongoDB database. If so, it should be quite straight-forward to write code to query the database, to find field(s) which include a given position (RA, Dec), or return a statement to the effect that "the ARG database does not contain an ARG field which has that position within it".

  I have developed a spreadsheet-based method to answer this question; I will describe how it works. It does not always find an ARG field, and I do not have a good way to estimate how likely it is that such a field actually exists (despite a failure to find one).

  Let's take J0836+0532, the host of a newly discovered SDRAGN, as a concrete example. This is SDSS J083655.86+053242.0, and is located at (RA, Dec) = (129.2328, 5.5450).

  ARG field IDs can be converted into integers; e.g. ARG00033zs can be represented as 145144 (think base36, ten digits and 26 characters; if you'd like more detail, just ask!). When I plotted the Dec of the center of a set of ARG fields against their IDs, I found this (sorry, I don't seem to be able to post the image) ... so it seems that for at least some subset of ARG fields, the ID is closely correlated to the Dec of the field center.

  That suggests a fairly simple method for finding an ARG field, given a Dec value (if such a field exists):

  • start with a pair of (ARGID, Dec) values, with the target Dec in between the two starting Dec values
  • do a linear interpolation, to get an estimated ARGID of the target
  • check how close this is to the target
  • repeat until the ARG field center is 'close to' the target Dec; 'close to' should be, ideally, within 3', as the ARG fields are 3'x3'
  • manually search ARG fields 'next to' the candidate, out to at least 3' from the target;

  Let's see how this works with J0836+0532. Two pairs are (ARGID, Dec): (153980, 7.274), (167216, 3.75); our target is 5.5450. Interpolating gives 160474, or ARG0003ftm. That field's center is (176.229, 5.530). Repeat, using (160474, 5.530) and (167216, 3.75): 160417, which is ARG0003fs1, and its center is (130.296, 5.544). This is sufficiently close that we can start looking at neighboring fields, such as ARG0003fs0 and ARG0003fs2, etc.

  Checking just a few neighbors, and we find ARG0003frv, at the center of which is SDSS J083655.86+053242.0:

  

  Here's the FIRST field (the one you get when you click on the "FIRST" link):

  

  And the NVSS one:

  

  There are no Comments on ARG0003frv, nor any Discussions; so either no zooite has had this classify yet, or those who did did not notice that it is an SDRAGN. 😦

  Questions?

  Posted September 8, 2015 9:05 AM

• by JeanTate in response to JeanTate's comment.

  I found this (sorry, I don't seem to be able to post the image)

  I managed to create a URL that displays as an image:

  

  Posted September 8, 2015 6:14 PM

• by KWillett scientist, admin, translator

  Hi Jean,

  Nice detective work - I hadn't seen this post before. That's an interesting correlation - I actually didn't know that's how the Zooniverse IDs were generated. It's mostly unplanned - the filenames of the galaxies as we put them in do have the coordinates in their name, and so there was an intrinsic order to them as we generated the ID numbers. There is a second ID that we use for internal bookkeeping that's a much longer hexadecimal string and really is more randomly generated, but it's clearly not the same for the Zooniverse ID.

  Just as an addition - one of the reasons (not the only one) that searches like that aren't enabled is because they really can be computationally intensive. To find a particular subject from coordinates requires a call to our API; multiple people doing spatial queries can noticeably affect the performance of the whole system. In addition, we don't have the spatial indices pre-defined that would make queries like that possible (that's more of a database issue).

  Please do keep exploring, including the trick above, if you're curious about whether particular galaxies or areas of the sky have GZ or RGZ coverage. If you have particular questions that you can't answer yourself, the science team can try to help on individual galaxies as our time permits.

  Posted November 3, 2015 8:17 PM

• by JeanTate in response to KWillett's comment.

  Thanks Kyle.

  I hadn't seen this post before

  Hmm ... I didn't know that, and had taken your lack of response quite the wrong way.

  Just as an addition - one of the reasons (not the only one) that searches like that aren't enabled is because they really can be computationally intensive. To find a particular subject from coordinates requires a call to our API; multiple people doing spatial queries can noticeably affect the performance of the whole system. In addition, we don't have the spatial indices pre-defined that would make queries like that possible (that's more of a database issue).

  My own understanding of APIs and databases is very rudimentary, but I would think it rather straight-forward to develop a secondary database, with Dec as the primary key, and containing only Dec, RA, and ARG ID (and, perhaps, whether the ARG field is retired/currently being classified/not yet). And only a small modification to the API would allow searches on Dec to be done via Talk's Search*. There would be more data in the RGZ cloud, but a pretty trivial addition I should think, given how small this secondary database would be.

  But, as DZM never hesitates to explain (over in Zooniverse Talk), there are no resources available to do anything with any v2 Talk, so no matter how easy such an add-on would be, it won't happen.

  whether particular galaxies or areas of the sky have GZ ... coverage

  I haven't looked at GZ, but I expect that the IDs are likely coded by (sub)project; that would make more of a challenge to work out the correlation ...

  *or perhaps not; RGZ Talk's Search is clearly woefully weak/awful, compared with standard online search tools, even those available many years' ago

  Posted November 3, 2015 10:19 PM

• by KWillett scientist, admin, translator

  See http://radiotalk.galaxyzoo.org/#/boards/BRG0000003/discussions/DRG0000dr0 for a solution.

  Posted November 10, 2015 9:12 PM