[AstroPy] Re : Cross matching two source lists

[Tom Aldcroft]

What about using a KDTree with the sources represented as unit vectors
in 3-space (i.e. on the surface of a sphere).  Then the metric is
(roughly) Euclidean over the whole sky as long as you are correlating
sources that are relatively close.  Once you have a matched source
list you could post-filter using exact spherical computations if
needed since this method slightly underestimates the distance.  Just a
thought.  :-)

- Tom

On Fri, May 11, 2012 at 6:32 AM, Jerome Caron
<jerome_caron_astro at ymail.com> wrote:
> One option could be to transform the (alpha,delta) with the projection law
> described here, pp5-6
> http://lanl.arxiv.org/pdf/astro-ph/9907229v1.pdf
> It is shape-preserving, and I understand size distorsions are small even for
> large fields.
> Jerome
>
> PS: my apologies Thomas, I sent the email too fast to you but not to the
> list.
>
> De : Thomas Robitaille <thomas.robitaille at gmail.com>
> À : Sergio Pascual <sergiopr at fis.ucm.es>
> Cc : astropy mailing list <astropy at scipy.org>
> Envoyé le : Vendredi 11 mai 2012 12h04
> Objet : Re: [AstroPy] Cross matching two source lists
>
> Hi Sergio,
>
> Thanks for your reply! So it looks like your algorithm only works for
> e.g pixel or arcsec coordinates, not e.g. longitude/latitude or
> ra/dec, correct? (since for spherical coordinates, even for a small
> field of view, the spherical angular distance is not the same as the
> euclidean distance, i.e. 1 second of RA is not 15 arcseconds except at
> the equator).
>
> For a small field of view, one could do a KDTree using (RA * cos(Dec),
> Dec), but I'm interested in finding an algorithm that works for large
> areas of sky.
>
> Cheers,
> Tom
>
> On 11 May 2012 11:10, Sergio Pascual <sergiopr at fis.ucm.es> wrote:
>> Well, I don't know if it's exactly what you are looking for, but,
>> after some discussion in this list I created a package that
>> implementsthe Groth algorithm for catalogue matching:
>>
>> A Pattern-Matching Algorithm for Two-Dimensional Coordinate Lists
>> Eduard J. Groth 1986 AJ 19, 5
>> http://adsabs.harvard.edu/abs/1986AJ.....91.1244G
>>
>> The code is here: https://bitbucket.org/sergiopr/gmatch
>>
>> Regards
>>
>> 2012/5/11 Thomas Robitaille <thomas.robitaille at gmail.com>:
>>> Hi everyone,
>>>
>>> I am looking for a Python package that can do arbitrary and efficient
>>> cross-matching of source lists, i.e. some kind of kd-tree that uses
>>> spherical rather than euclidean distances. Is anyone aware of such a
>>> package?
>>>
>>> Thanks,
>>> Tom
>>> _______________________________________________
>>> AstroPy mailing list
>>> AstroPy at scipy.org
>>> http://mail.scipy.org/mailman/listinfo/astropy
>>
>>
>>
>> --
>> Sergio Pascual     http://guaix.fis.ucm.es/~spr    +34 91 394 5018
>> gpg fingerprint: 5203 B42D 86A0 5649 410A F4AC A35F D465 F263 BCCC
>> Departamento de Astrofísica -- Universidad Complutense de Madrid (Spain)
> _______________________________________________
> AstroPy mailing list
> AstroPy at scipy.org
> http://mail.scipy.org/mailman/listinfo/astropy
>
>
>
>
>
> _______________________________________________
> AstroPy mailing list
> AstroPy at scipy.org
> http://mail.scipy.org/mailman/listinfo/astropy
>