- The dimension of indices is now treated as a kind of "effective" dimension
authorChristian Bauer <Christian.Bauer@uni-mainz.de>
Thu, 18 Jul 2002 19:24:33 +0000 (19:24 +0000)
committerChristian Bauer <Christian.Bauer@uni-mainz.de>
Thu, 18 Jul 2002 19:24:33 +0000 (19:24 +0000)
commitb882ec7bcc64b32d1a0eb13d646975d4aca6ac92
treebae84d5cf37a0b11048c8d5d4060f4739afc5e16
parent93997f8dd95626382a895b3800fc2846da4f109b
- The dimension of indices is now treated as a kind of "effective" dimension
  with respect to contractions; this allows for a limited support of subspaces,
  e.g. delta.i[2].j[2] * a.i[3] -> a.j[2] by simplify_indexed(), and
  delta.i[2].i[3] -> 2. Likewise, eta~mu[3]~nu[3] * eta.mu[4].nu[4] -> 3 etc.
  (The numbers in brackets are the index dimensions which are not visible
  in the output.)
  Symbolic ("N", "D", ...) dimensions are always assumed to be higher than
  numeric ones.
- eps0123 is gone, dirac_trace() generates ordinary epsilon tensors with
  4-dimensional indices.
check/exam_clifford.cpp
ginac/clifford.cpp
ginac/idx.cpp
ginac/tensor.cpp
ginac/tensor.h