These are a set of routines performing the reduction of tensor
integrals to scalar integrals. They are loosely based on the 
paper of van Oldenborgh and Vermaseren

@article{vanOldenborgh:1989wn,
      author         = "van Oldenborgh, G.J. and Vermaseren, J.A.M.",
      title          = "{New Algorithms for One Loop Integrals}",
      journal        = "Z.Phys.",
      volume         = "C46",
      pages          = "425-438",
      doi            = "10.1007/BF01621031",
      year           = "1990",
      reportNumber   = "NIKHEF-H/89-17",
      SLACcitation   = "%%CITATION = ZEPYA,C46,425;%%",
}
In particular on the idea of separating the physical and 
transverse spaces. This allows the reduction to occur by
relating the Tensor Integrals themselves, rather than the
form factors as in the Passarino-Veltman method.
The tensor integrals are functions
of the the external momenta, p_i not the off-set momenta, q_i.
The determination of the tensor integrals reduces to 
solving a series of linear equations of the form (say,for n=3)
where Gram(i,j)=p_i.p_j

                 (x1)     (b1)
     Gram(i,j) * (x2)  =  (b2) 
                 (x3)     (b3)

which are solved by LU decomposition and back substitution.