Name

---

    MATCH_ND

Purpose

---

    For each arbitrarily-dimensioned point in one vector, determines the closest
    point in a second vector. Method is to take the list returned by MATCHALL_ND
    and narrow it down to the closest match.

Category

---

    Astro

Calling Sequence

---

    Result = MATCH_ND(P1, P2, MaxDistance)

Inputs

---

    P1: N1 x D array of D-dimensional coordinates.
    P2: N2 x D array of D-dimensional coordinates.
    MaxDistance: Maximum D-dimensional distance.

Keyword Parameters

---

    MINDIST: Optional output containing an array of the actual distance
              to the closest match for each element of P1.
    ONE_TO_ONE: Enforces one-to-one matching. By default, matching can
                be many-to-one, i.e. one entry in P2 can be the
                closest match to several entries in P1. If
                /ONE_TO_ONE is given, then each multiple entry in P2
                is first assigned to its closest point in P1. Then any
                entries in P1 that have lost their match are assigned
                the next closest point within MaxDistance. This process is
                iterated until all points in P1 are matched to a
                unique point in P2 or there are no more points
                within MaxDistance.

Outputs

---

    The function returns an array with one entry for each P1
    element containing the index in P2 that is closest, or -1 if
    there are none within MaxDistance.

Example

---

    na = 10
    nb = 1000
    a = randomn(seed, na, 3)
    b = 2. * randomu(seed, nb, 3) - 1
    matches = match_nd(a, b, 0.3)
    iplot, /scatter, /iso, b[*,0], b[*,1], b[*,2]
    iplot, /overplot, /scatter, a[*,0], a[*,1], a[*,2], sym_index=6, $
      sym_color=[255,0,0]
    for i=0L, na-1 do if matches[i] ne -1 then $
      iplot, /overplot, [a[i,0], b[matches[i],0]], [a[i,1], b[matches[i],1]], $
        [a[i,2], b[matches[i],2]], sym_index=0, color=[255,0,0]

Modification History

---

    Written by: Jeremy Bailin
    10 June 2008 Public release in JBIU as WITHINSPHRAD
    24 April 2009 Vectorized as WITHINSPHRAD_VEC
    25 April 2009 Polished to improve memory use
    9 May 2009 Radical efficiency re-write as WITHINSPHRAD_VEC3D borrowing
                  heavily from JD Smith's MATCH_2D
    13 May 2009 Removed * from LHS index in final remapping for speed
    23 August 2010 Modified to only return closest match as WITHINSPHRAD_CLOSEST
                  (importing some stuff back from MATCH_2D)
    8 Sept 2010 Renamed MATCH_SPH. Added /ONE_TO_ONE option, and modified
                  to explicitly call MATCHALL_SPH and then cull.
    9 Feb 2011 Forked as MATCH_ND to handle Euclidean case of arbitrary dimension.
    13 April 2011 Fixed to work around cumulative total 1-element array bug.