subroutine kutl_vertex_same(z1, Vz1, z2, Vz2, chisq)
*
* begin_doc
*
*  Finds the chisquare for 2 vertices to be the same.
*
*  Note: chisq is set to -1 if the routine cannot evaluate the combined
*        covariance matrix.
*
*  Input Parameters:
*  z1             double precision array
*                 (x,y,z) of first vertex
*
*  Vz1            double precision array
*                 3x3 covariance matrix of first vertex
*
*  z2             double precision array
*                 (x,y,z) of second vertex
*
*  Vz2            double precision array
*                 3x3 covariance matrix of second vertex
*
*  Output Parameters:
*  chisq          double precision variable
*                 chisquare of fit
*
*  Other routines:
*
*  Notes:
*
*  >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
*   The method used is the standard constrained fit with 3 constraints.
*   See the discussion in CBX 91-71 for a derivation of the solution.
*   The constraints can be written z1 + z2 = 0 or Dx + d = 0, with
*
*           |1  0  0 -1  0  0|                   | z1 |
*       D = |0  1  0  0 -1  0|      d = 0    x = | z2 |
*           |0  0  1  0  0 -1|
*
*   The chisquare can be written
*
*        chisq = (z1 - z2)(t) * VD * (z1 - z2)
*
*   where
*
*        VD = (D * Vx0 * D(t))(inv) = (Vz1 + Vz2)(inv)
*
*   The chisquare is the same as that obtained by finding the average
*   the vertices as in kutl_vertex_average.
*  >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
*
*  Author:   Paul Avery      Created:  Tue Sep 23 21:28:38 EDT 1997
*
*  Major revisions:
*
* end_doc