Dear developers, I found out that the result of ex f=dirac_gamma(varidx(nu,dim)) *dirac_slash(k,dim) *dirac_gamma(varidx(mu,dim)) *indexed(n,varidx(nu,dim,true)) *indexed(k,varidx(mu,dim,true)); cout << "before canon_cl: " << f << endl; f=canonicalize_clifford(f); cout << "after canon_cl: " << f << endl; is before canon_cl: n.nu*(gamma~nu*k\*gamma~mu)*k.mu after canon_cl: 2*n\*k.mu*k~mu-2*n~mu*k.mu*k\+k\*n\*k\ I think it is strange that canonicalize_clifford can return something that contains k\*n\*k\ because this looks particularly uncanonicalized. I suggest the patch Index: ginac/clifford.cpp =================================================================== RCS file: /home/cvs/GiNaC/ginac/clifford.cpp,v retrieving revision 1.84 diff -r1.84 clifford.cpp 940c940 < ex canonicalize_clifford(const ex & e) ---

ex canonicalize_clifford(const ex & e_) 941a942 ex e=simplify_indexed(e_);

With this patch the test program gives before canon_cl: n.nu*(gamma~nu*k\*gamma~mu)*k.mu after canon_cl: n\*k.symbol8*k~symbol8 Best wishes, Chris