Ajit described the  Independent  Component Analysis  for HLT  track
finding.   The principal  component  analysis has  two phases:  linear
transformation  and  chi2  minimization.  It  is  very  fast.  Is  the
performance of the Independent  Component Analysis comparable? Yes, it
is more or  less the same. The same technique could  be used for seeds
in  the normal  track reconstruction.  Could  this method  be used  in
another  place apart  from TPC?  Probably in  MUON detector.  The slow
tracks in ITS are more difficult question