Dear Vladimir,
However, if this is still not sufficient you may derive something from clifford class adding something more involved than int for the commutator_sign.
thanks, this makes sense. I guess this is what one has to do at the end if one wants to implement new things in the handling of Dirac algebra. Apart from spinors, one could also add a non-naive renormalization scheme for Dirac gamma^5, where the anitcommutator of {g^5,g^mu} is non-zero in D-dimensions. However, this would be a rather formidable undertaking ... Cheers, Vladyslav On 27/08/14 18:53, Vladimir V. Kisil wrote:
Dear Vladislav,
On Wed, 27 Aug 2014 17:12:23 +0200, Vladyslav Shtabovenko <v.shtabovenko@tum.de> said: VSh> However, this doesn't seem to be possible since the eval_ncmul
That is true, if you are going to use precooked Dirac gammas. However, the clifford class has the member
int commutator_sign; /**< It is the sign in the definition e~i e~j +/- e~j e~i = B(i, j) + B(j, i)*/
This allows to implement either commuting rules or anti-commuting rules for a particular clifford object. In particular, some while ago I have experimented with a Lie algebra implementation as a clifford object of GiNaC (the code is available on request). Also, clifford objects with different representation labels simply commute. This combination gives a significant freedom for implementing various algebraic rules.
However, if this is still not sufficient you may derive something from clifford class adding something more involved than int for the commutator_sign.
Best wishes, Vladimir