<html>
  <head>
    <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <div class="moz-cite-prefix">I agree with Marco and Mark, so just a
      last comment on this thread.<br>
      <br>
      On 02/09/2016 06:49 PM, Arnold Rots wrote:<br>
    </div>
    <blockquote
cite="mid:CAJXToE9MqXmABOAB0R+Etz8f5Wd_mMhOEY0GyOMr0Ke_eMcRPA@mail.gmail.com"
      type="cite">
      <div dir="ltr">
        <div>
          <div>
            <div>
              <div>That would be nice, but I suspect that people might
                find it too complicated.<br>
              </div>
            </div>
          </div>
        </div>
      </div>
    </blockquote>
    I guess most astronomers are used to the chi-square test so<br>
    it should not be that complicated to understand.<br>
    We just have to show them how it can be done iteratively<br>
    when cross-correlating more than 2 catalogues  <br>
    (weighted mean positions and associated errors is the answer).<br>
    <blockquote
cite="mid:CAJXToE9MqXmABOAB0R+Etz8f5Wd_mMhOEY0GyOMr0Ke_eMcRPA@mail.gmail.com"
      type="cite">
      <div dir="ltr">
        <div>
          <div>
            <div>Besides, the you really would want to do a proper
              Bayesian cross-match<br>
            </div>
            that handles complete collections of sources and takes into
            account the<br>
          </div>
          areas of coverage as well.<br>
        </div>
      </div>
    </blockquote>
    Well, although it may not be really clear in my mind,<br>
    for me the cross-match is just the selection of candidates<br>
    (based on a chi-square test if we account for the positional errors,
    fixing a significance level).<br>
    Then, computing and putting threshold on Bayesian
    probabilities/posterior likelihoods,<br>
    performing cuts based on color-color diagrams/... is part of the
    cross-identification<br>
    and can be made in a second step.<br>
    <br>
    fx<br>
    <br>
    <blockquote
cite="mid:CAJXToE9MqXmABOAB0R+Etz8f5Wd_mMhOEY0GyOMr0Ke_eMcRPA@mail.gmail.com"
      type="cite">
      <div dir="ltr">
        <div><br>
        </div>
          - Arnold<br>
      </div>
      <div class="gmail_extra"><br clear="all">
        <div>
          <div class="gmail_signature">
            <div dir="ltr">-------------------------------------------------------------------------------------------------------------<br>
              Arnold H. Rots                                         
              Chandra X-ray Science Center<br>
              Smithsonian Astrophysical Observatory                  
              tel:  +1 617 496 7701<br>
              60 Garden Street, MS 67                                   
                fax:  +1 617 495 7356<br>
              Cambridge, MA 02138                                    
                  <a moz-do-not-send="true"
                href="mailto:arots@cfa.harvard.edu" target="_blank">arots@cfa.harvard.edu</a><br>
              USA                                                   <a
                moz-do-not-send="true"
                href="http://hea-www.harvard.edu/%7Earots/"
                target="_blank"><a class="moz-txt-link-freetext" href="http://hea-www.harvard.edu/~arots/">http://hea-www.harvard.edu/~arots/</a></a><br>
--------------------------------------------------------------------------------------------------------------<br>
              <br>
            </div>
          </div>
        </div>
        <br>
        <div class="gmail_quote">On Tue, Feb 9, 2016 at 12:23 PM,
          François-Xavier Pineau <span dir="ltr">&lt;<a
              moz-do-not-send="true"
              href="mailto:francois-xavier.pineau@astro.unistra.fr"
              target="_blank"><a class="moz-txt-link-abbreviated" href="mailto:francois-xavier.pineau@astro.unistra.fr">francois-xavier.pineau@astro.unistra.fr</a></a>&gt;</span>
          wrote:<br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            <div bgcolor="#FFFFFF" text="#000000">
              <div>Going this way, why not taking into account
                elliptical errors:<br>
                <br>
                xmatch(ra1, dec1, a1, b2, pa1, ra2, dec2, a2, b2, pa2,
                thresold on the Mahalanobis distance)<br>
                <br>
                with:<br>
                - a: semi major axis<br>
                - b: semi minor axis<br>
                - pa: position angle<br>
                <br>
                returning the Mahalanobis distance, the weighted mean
                position and the associated elliptical error to be able
                to coherently chain cross-matches...<br>
                <br>
                Cheers,<br>
                <br>
                fx
                <div>
                  <div class="h5"><br>
                    <br>
                    On 02/09/2016 06:01 PM, Arnold Rots wrote:<br>
                  </div>
                </div>
              </div>
              <div>
                <div class="h5">
                  <blockquote type="cite">
                    <div dir="ltr">
                      <div>
                        <div>
                          <div>
                            <div>
                              <div>Alowing the two radii to be specified
                                separately would make it more<br>
                              </div>
                              explicit that they are associated with
                              specific catalogs and allows<br>
                            </div>
                            users to keep track of those specific
                            catalog-dependent values, particularly<br>
                          </div>
                          if they do cross-matching involving multiple
                          catalogs.<br>
                          <br>
                        </div>
                        One might even consider returning the overlap
                        ratio (area of intersection<br>
                        over area of smallest circle) as a (admittedly
                        somewhat bogus) match probability .<br>
                        <br>
                      </div>
                      <div>Cheers,<br>
                        <br>
                      </div>
                      <div>  - Arnold<br>
                      </div>
                    </div>
                    <div class="gmail_extra"><br clear="all">
                      <div>
                        <div>
                          <div dir="ltr">-------------------------------------------------------------------------------------------------------------<br>
                            Arnold H. Rots                              
                                       Chandra X-ray Science Center<br>
                            Smithsonian Astrophysical
                            Observatory                   tel:  <a
                              moz-do-not-send="true"
                              href="tel:%2B1%20617%20496%207701"
                              value="+16174967701" target="_blank">+1
                              617 496 7701</a><br>
                            60 Garden Street, MS
                            67                                     
                            fax:  <a moz-do-not-send="true"
                              href="tel:%2B1%20617%20495%207356"
                              value="+16174957356" target="_blank">+1
                              617 495 7356</a><br>
                            Cambridge, MA 02138                         
                                           <a moz-do-not-send="true"
                              href="mailto:arots@cfa.harvard.edu"
                              target="_blank">arots@cfa.harvard.edu</a><br>
                            USA                                        
                                      <a moz-do-not-send="true"
                              href="http://hea-www.harvard.edu/%7Earots/"
                              target="_blank">http://hea-www.harvard.edu/~arots/</a><br>
--------------------------------------------------------------------------------------------------------------<br>
                            <br>
                          </div>
                        </div>
                      </div>
                      <br>
                      <div class="gmail_quote">On Tue, Feb 9, 2016 at
                        11:50 AM, Tom McGlynn (NASA/GSFC Code 660.1) <span
                          dir="ltr">&lt;<a moz-do-not-send="true"
                            href="mailto:tom.mcglynn@nasa.gov"
                            target="_blank">tom.mcglynn@nasa.gov</a>&gt;</span>
                        wrote:<br>
                        <blockquote class="gmail_quote" style="margin:0
                          0 0 .8ex;border-left:1px #ccc
                          solid;padding-left:1ex">Arnold,<br>
                          <br>
                          Assuming you mean for a match to occur
                          whenever the two circles overlap, then
                          wouldn't<br>
                          <br>
                              xmatch(ra1,dec1,rad1, ra2,dec2,rad2)<br>
                          <br>
                          be equivalent to<br>
                          <br>
                             xmatch(ra1, dec1, ra2, dec2, rad1+rad2)<br>
                          <br>
                          if we're returning only integer values 1 and
                          0.<br>
                          <br>
                          I suppose one could define this differently
                          with<br>
                             xmatch(ra1,dec1,rad1, ra2, dec2,rad2)<br>
                          returning the fraction of the circle defined
                          by ra1,dec1,rad1 which<br>
                          is enclosed in the second circle.  Not sure I
                          really see the use case for that though.<br>
                          <br>
                              Tom<br>
                          <br>
                          <br>
                          Arnold Rots wrote:<br>
                          <blockquote class="gmail_quote"
                            style="margin:0 0 0 .8ex;border-left:1px
                            #ccc solid;padding-left:1ex"><span> I don't
                              think I am the only one who is not
                              enamored of these<br>
                              brute-force unnuanced cross-matches, but I
                              wonder whether<br>
                              it would be helpful to improve them by
                              allowing two radii.<br>
                              <br>
                              Cheers,<br>
                              <br>
                                - Arnold<br>
                              <br>
-------------------------------------------------------------------------------------------------------------<br>
                              Arnold H. Rots Chandra X-ray Science
                              Center<br>
                              Smithsonian Astrophysical Observatory
                              tel:  <a moz-do-not-send="true"
                                href="tel:%2B1%20617%20496%207701"
                                value="+16174967701" target="_blank">+1
                                617 496 7701</a><br>
                              60 Garden Street, MS 67   fax:  <a
                                moz-do-not-send="true"
                                href="tel:%2B1%20617%20495%207356"
                                value="+16174957356" target="_blank">+1
                                617 495 7356</a><br>
                            </span> Cambridge, MA 02138 <a
                              moz-do-not-send="true"
                              href="mailto:arots@cfa.harvard.edu"
                              target="_blank"><a class="moz-txt-link-abbreviated" href="mailto:arots@cfa.harvard.edu">arots@cfa.harvard.edu</a></a>
                            &lt;mailto:<a moz-do-not-send="true"
                              href="mailto:arots@cfa.harvard.edu"
                              target="_blank">arots@cfa.harvard.edu</a>&gt;<br>
                            USA <a moz-do-not-send="true"
                              href="http://hea-www.harvard.edu/%7Earots/"
                              rel="noreferrer" target="_blank">http://hea-www.harvard.edu/~arots/</a>
                            &lt;<a moz-do-not-send="true"
                              href="http://hea-www.harvard.edu/%7Earots/"
                              rel="noreferrer" target="_blank">http://hea-www.harvard.edu/%7Earots/</a>&gt;<br>
                            --------------------------------------------------------------------------------------------------------------

                            <div>
                              <div><br>
                                <br>
                                <br>
                                On Tue, Feb 9, 2016 at 10:21 AM, Mark
                                Taylor &lt;<a moz-do-not-send="true"
                                  href="mailto:M.B.Taylor@bristol.ac.uk"
                                  target="_blank">M.B.Taylor@bristol.ac.uk</a>
                                &lt;mailto:<a moz-do-not-send="true"
                                  href="mailto:M.B.Taylor@bristol.ac.uk"
                                  target="_blank">M.B.Taylor@bristol.ac.uk</a>&gt;&gt;

                                wrote:<br>
                                <br>
                                    Hi Marco and DAL,<br>
                                <br>
                                    I support the introduction of a new
                                XMATCH function for two main<br>
                                    reaons:<br>
                                <br>
                                    Usability:<br>
                                       A dedicated crossmatch function
                                should be easier<br>
                                       for users to use and remember,
                                and all round less horrible,<br>
                                       than the current recommended(?)
                                cross-match idiom:<br>
                                          1 = CONTAINS(POINT(coordsys,
                                lon1, lat1),<br>
                                                       CIRCLE(coordsys,
                                lon2, lat2, radius))<br>
                                       (where coordsys is a string that
                                should probably be 'ICRS', or<br>
                                       maybe should be empty, but anyway
                                is unlikely to make much<br>
                                       difference to the result).<br>
                                <br>
                                    Implementability:<br>
                                       It may make it easy/possible to
                                provide standard ADQL syntax for<br>
                                       efficient sky crossmatching in
                                database backends that otherwise<br>
                                       can't do it, because they have
                                trouble implementing the ADQL<br>
                                       geometry functions (because they
                                lack pgSphere).<br>
                                <br>
                                    In my opinion it should look like
                                this:<br>
                                <br>
                                        1 = XMATCH(lon1, lat1, lon2,
                                lat2, radius)<br>
                                <br>
                                    The alternative would presumably be<br>
                                <br>
                                        1 = XMATCH(POINT(coordsys, lon1,
                                lat1), POINT(coordsys, lon2,<br>
                                    lat2), radius)<br>
                                <br>
                                    which from the point of view of
                                usability is not much better than<br>
                                    the status quo.  Although I believe
                                the annoying and disingenuous<br>
                                    coordsys argument to the geometry
                                functions is scheduled for removal<br>
                                    from ADQL, my understanding is that
                                it's not intended for this<br>
                                    (minor) revision.  Even without the
                                coordsys arg, I think the<br>
                                    POINTless form looks less
                                intimidating for users.<br>
                                <br>
                                    I would have thought that from the
                                point of view of implementability<br>
                                    as well the POINTless form presents
                                fewer constraints.<br>
                                    However, I'm not a database
                                implementation person, so I might be<br>
                                    wrong about that.<br>
                                <br>
                                    Mark<br>
                                <br>
                                    On Tue, 9 Feb 2016, Marco Molinaro
                                wrote:<br>
                                <br>
                                    &gt; Dear DAL members and ADQL fans,<br>
                                    &gt; to go on with the ADQL-2.1
                                working draft<br>
                                    &gt; one issue is left, from Sydney
                                interop,<br>
                                    &gt; to be discussed.<br>
                                    &gt;<br>
                                    &gt; In the DAL splinter at the
                                interop<br>
                                    &gt; it was agreed to add an XMATCH
                                function<br>
                                    &gt; of binary type and definition<br>
                                    &gt;<br>
                                    &gt; 1 = XMATCH(a,b,radius)<br>
                                    &gt;<br>
                                    &gt; However no agreement was
                                reached about<br>
                                    &gt; the 'a' and 'b' parameters,
                                whether they<br>
                                    &gt; should be points (ADQL:POINT)
                                or RA&amp;Dec<br>
                                    &gt; couples (floating point
                                values).<br>
                                    &gt;<br>
                                    &gt; Both choices have advantages
                                and disadvantages.<br>
                                    &gt; Points are more into the logic<br>
                                    &gt; of a sky cross-match but
                                require geometric<br>
                                    &gt; types to be directly available
                                to the DB.<br>
                                    &gt; Coordinates couples are
                                directly available<br>
                                    &gt; in whatever DB and would also
                                let the XMATCH<br>
                                    &gt; function work for non-orthodox
                                coordinates<br>
                                    &gt; matching, but of course loosing
                                the sky matching<br>
                                    &gt; logic.<br>
                                    &gt;<br>
                                    &gt; As I said (also due to time
                                constraints) no<br>
                                    &gt; agreement was found in Sydney.<br>
                                    &gt;<br>
                                    &gt; What's your opinion on this,
                                and why?<br>
                                    &gt;<br>
                                    &gt; Cheers,<br>
                                    &gt;     Marco<br>
                                    &gt;<br>
                                <br>
                                    --<br>
                                    Mark Taylor   Astronomical
                                Programmer   Physics, Bristol<br>
                                    University, UK<br>
                              </div>
                            </div>
                                <a moz-do-not-send="true"
                              href="mailto:m.b.taylor@bris.ac.uk"
                              target="_blank">m.b.taylor@bris.ac.uk</a>
                            &lt;mailto:<a moz-do-not-send="true"
                              href="mailto:m.b.taylor@bris.ac.uk"
                              target="_blank">m.b.taylor@bris.ac.uk</a>&gt;<br>
                                <a moz-do-not-send="true"
                              href="tel:%2B44-117-9288776"
                              value="+441179288776" target="_blank">+44-117-9288776</a>
                            &lt;tel:%2B44-117-9288776&gt;<br>
                                <a moz-do-not-send="true"
                              href="http://www.star.bris.ac.uk/%7Embt/"
                              rel="noreferrer" target="_blank">http://www.star.bris.ac.uk/~mbt/</a>
                            &lt;<a moz-do-not-send="true"
                              href="http://www.star.bris.ac.uk/%7Embt/"
                              rel="noreferrer" target="_blank">http://www.star.bris.ac.uk/%7Embt/</a>&gt;<br>
                            <br>
                            <br>
                          </blockquote>
                          <br>
                        </blockquote>
                      </div>
                      <br>
                    </div>
                  </blockquote>
                  <br>
                  <br>
                </div>
              </div>
              <span class="HOEnZb"><font color="#888888">
                  <pre cols="72">-- 
François-Xavier Pineau
CDS, Observatoire Astronomique de Strasbourg
11, rue de l'Université
F - 67000 Strasbourg</pre>
                </font></span></div>
          </blockquote>
        </div>
        <br>
      </div>
    </blockquote>
    <br>
    <br>
    <pre class="moz-signature" cols="72">-- 
François-Xavier Pineau
CDS, Observatoire Astronomique de Strasbourg
11, rue de l'Université
F - 67000 Strasbourg</pre>
  </body>
</html>