Description of David Asners' analysis so far...
            Harry Nelson, 12/03/97
------------------------------------------------------------------------
Skim Cuts:
  1)all code identical for half the sign permutations of the
    tagging slow pion and the K and pi: the half of the
    permutations with same sign fast K and fast pi are right
    now cut.
  2)event level cuts: had to be KLASGL=10, which is based
    on number of tracks and visible energy
  3)no quality cuts whatsoever on tracks such as KINCD  etc.
  4)the magnitude of the momentum of the particles must exceed:
    (this is really meant to be a vertex error cut)
    a) K:              100 MeV
    b)pi:              100 MeV
    c)tagging slow pi:  50 MeV

  5)Event written out if there is at least one permutation of
    tracks where:
    a)m(Kpi) within 210 MeV of the D0 mass
      and m(Kpipi)-m(Kpi) within 7.5 MeV of nominal (5.8 MeV)

    [[[SOON, WILL ALLOW...

    b)or m(Kpi) within 40 MeV of D0 mass
      and  m(Kpipi)-m(Kpi) within 40 MeV of nominal (5.8 MeV)]]]

    c)and The momentum of the Kpipi system must exceed 1.9 GeV

  6)The .rp files from David's summer skim (where a few of the
    cuts were tighter than above) are at:
    /cdat/lnssb1/disk7/asner/skim/4sK_4sL_kpi.rp.gz.
    This is for the 4SK and L data sets, which correspond to an
    integrated luminosity of 1.5 1/fb.
    His new skim, with the cuts above, is at:
    ~marsh/skims_il1ds/kpi_4sM.rp.gz.
    This is for the 4SM alone, about 0.5 1/fb.

Standard Cuts - made at the driver/(and/or)ntuple level
  1)Stiff Tracks: (K and pi from D0)
     a)KINCD=0
     b)Track Quality passing TRKMNG
     c)neither Z-ESCAPE nor DREDGE
     d)>15 hits in DR and VD
     e)no specific distinction between 
       number of rphi and z hits
     f)no silicon cuts
     g)Successful Kalman Fit
     h)no Chi^2 of track fit cut
     i)no Z0CD cut
     j)no DBCD cut
     k)no cos(theta) cut
     l)no momentum cut
  2)Soft Track: (the tagging slow pion)
     a)KINCD not required
     b)Track Quality passing TRKMNG
     c)can be Z-ESCAPE, cannot be DREDGE
     d)no requirement on number of hits in DR and VD
     e)>= 2 hits per view in the silicon
     f)Successful Kalman Fit
     g)no Chi^2 of track fit cut
     h)no Z0CD cut
     i)no DBCD cut
     j)no cos(theta) cut
     k)no momentum cut
  3)Event Hypothesis
     a)all Stiff tracks assigned both
       pi and K hypothesis.
     b)vertex constraint applied to all 
       stiff track hypothesis permutations,
       CL of vertex > 1%.
     c)M(KK), M(pipi) computed with vertex
       constraint for each surviving stiff track hypothesis;
       if M(KK) or M(pipi) within 28 MeV of the
       D0 mass, all permutations involving those two
       tracks are cut out.
     d)M(Kpi), and `flipped' M(piK) computed with vertex
       constraint for each surviving stiff track hypothesis;
       if M(piK) is within 28 MeV of the D0 mass, the
       M(Kpi) permutation involving these tracks is cut
       cut out.
     e)An event vertex is reconstructed for all surviving
       stiff track Kpi hypotheses.  The algorithm, described
       in 6) below, is designed to eliminate bias in the direction
       of the Kpi;  the CL of the vertex fit must exceed
       0.1%.
     f)The D0 is refit through the event vertex, and the
       CL of the fit must exceed 0.1%.
     g)each soft track candidate is refit through each
       event vertex; the CL of the refit must exceed 1%.
       dm=M(Kpipi_s)-M(Kpi)-m_pi, and the
       momentum of the Kpipi_s system, p_D*, is computed.
     h)Multiple candidates are now allowed.
     i)p_D*>2.0 GeV is required
     j)two scatterplots in M(Kpi) and dm are made
        1)pi_s^+(K^- pi^+) combined with pi_s^-(K+ pi^-) `right sign'
        2)pi_s^-(K^- pi^+) combined with pi_s^+(K+ pi^-) `wrong sign'
     k)about 2400 events are in the right sign
     l)the wrong sign has many events, from background...  PID needed
  4)Particle ID
     a)lepton id applied - if any track is either 1) or 2),
       the permutation is rejected:
        1)Electrons, |E/p|>0.8
        2)Muons, DPTHMU >= 3.00
     b)pi/K separation
        1)pulseheight and width corrections being redone.
        2)misid matrix being measured with right sign.
  5)Event vertex algorithm.
     a)The Kpi candidate is combined into a `D0 pseudotrack'
     b)If there are at least one stiff track that in addition:
         1) have >=2 silicon hits per view
         2) that are consistent with passing through
            the beam centroid with Chi^2<20 (2 DoF)
         3) are not part of the D0 candidate,
       then the successful stiff  tracks  are combined with 
       the D0 pseudotrack, and fit to
       a common vertex, and the weight matrix is determined.
       No cut is made on goodness of fit.
       If there are no successful stiff tracks, then the
       D0 trajectory is extrapolated back to the vertical
       beam position, and the event vertex is taken from
       the D0 trajectory at that point.
     c)The weight matrix from a successful common vertex
       in b) is rotated to a coordinate system
       where one axis is along the D0 direction, and the
       two remaining directions are transverse to the
       D0 direction.
     d)The 5 weight matrix elements involving the axis
       along the D0 direction are effectively set to zero.
     e)The point most consistent with both the beam centroid
       and the common vertex with reduced error matrix,
       and the D0 direction is found: this is the event vertex.
       This event vertex comes with a weight matrix, even
       if only the only track used was the D0 pseudotrack.
------------------------------------------------------------------------