$$l^{\mu} = l^{\mu}_{\parallel} + l^{\mu}_{\bot}$$ be a D-dimensional vector living in a Minkowskian space; the only non-zero components of $l^{\mu}_{\parallel}$ are the first four, while the only non-zero components of $l^{\mu}_{\bot}$ are the last $D - 4$.
Let moreover be $\gamma^{\mu}$ an element of a generic D-dimensional Clifford algebra, and define $$\gamma^5 = i\gamma^0\gamma^1\gamma^2\gamma^3;$$ the element $\gamma^5$ therefore commutes with all the elements with index different from $\mu = 0,1,2,3$.
Finally, using Feynman slash notation let's consider the trace $$tr[-2\gamma^5\not{l_{\bot}}(\not{l_{\parallel}} - \not{k})\gamma^{\lambda}\not{l_{\parallel}}\gamma^{\nu}(\not{l_{\parallel}} + \not{p})] ,$$ where $k^{\mu}$ and $p^{\mu}$ are assumed to be D-dimensional vectors behaving like $l^{\mu}_{\parallel}$, i.e. having only the first four component different from zero. According to Peskin & Schroesder's "An introduction to QFT", page 663,

To evaluate this contribution [...] we must retain one factor each of $\gamma^{\nu},\gamma^{\lambda},\not{p}$, and $\not{k}$ to give a nonzero trace with $\gamma^5$.

This is equivalent to say that $$tr[-2\gamma^5\not{l_{\bot}}(\not{l_{\parallel}} - \not{k})\gamma^{\lambda}\not{l_{\parallel}}\gamma^{\nu}(\not{l_{\parallel}} + \not{p})] = tr[2\gamma^5\not{l_{\bot}}\not{k}\gamma^{\lambda}\not{l_{\parallel}}\gamma^{\nu}\not{p}].$$ I don't get why this should be true. I tried to solve the problem assuming specifical values for the indexes $\nu$ and $\lambda$ and working on the fact that $\gamma^5$ commutes with $\gamma^{\alpha}$ if $\alpha > 3$ and anticommutes otherwise, but got nowhere; either each term of the trace becomes zero, or everyone survives.
So the question is: how can I prove the last relation I wrote?


The proposed final equality is in fact not correct - the correct interpretation of this statement given by P&S is in the text, c.f equation (19.59) where the end resolution of the trace is presented. To see this, one needs to be up on their Dirac algebra in $d$ dimensions, particularly with regards to those relations involving $\gamma^5$.

The trace as initialised consists of 18 terms. The relevant non vanishing contributions come from those terms proportional to $\ell_{\perp} \cdot \ell_{\perp}$. One can systematically reduce the number of terms through employment of the following relations $$\text{Tr}(\gamma^5 \gamma^{\mu} \gamma^{\nu}) = 0 \,\,\,,\,\,\text{Tr}(\gamma^5 \gamma^{\mu} \gamma^{\nu} \gamma^{\rho} \gamma^{\sigma}) \sim \epsilon^{\mu \nu \rho \sigma}$$ and $$\text{Tr}(\gamma^5 \gamma^{\mu} \gamma^{\nu} \gamma^{\rho} \gamma^{\sigma} \gamma^{\alpha} \gamma^{\beta}) \sim 4 \delta^{\mu \nu} \epsilon^{\rho \sigma \alpha \beta} + \text{perms.} $$ with $\mu,\nu,\rho,\sigma,\alpha, \beta \in \left\{0,\dots,3\right\}.$

A sacrificing of Lorentz invariance but with the ability to utilise the familiar definition of $\gamma^5$ in $d$ dimensions (as proposed by ‘tHooft and Veltmann circa 1972) means that the following further relations hold $$\left\{\gamma^5, \gamma^{\mu} \right\} = 0\,\,\,\,\text{and}\,\,\,\,[\gamma^5, \gamma^{\mu}] = 0$$ where the former holds if $\mu \in \left\{0,\dots, 3 \right\}$ and the latter if $\mu$ is evaluated in any of the other $d-4$ dimensions.

The important realisation is that in the trace of $\gamma^5$ with four or six other gamma’s in $d=4$, if we make one or more of these gamma’s carry an index outwith $\left\{0,1,2,3\right\}$, the trace vanishes by virtue of their proportionality to the epsilon symbol, as given above, non zero for e.g $\epsilon^{0123}$ and its permutations. This is why the non vanishing contributions come with $\ell_{\perp} \cdot \ell_{\perp}$, we have removed the dependence of remnant $d$ dimensional $\not{\ell_{\perp}}$ through explicit $\not{\ell_{\perp}} \not{\ell_{\perp}}$ terms to produce the scalar product.

As you will see, this leaves us with five terms, four containing a trace involving $\gamma^5, \gamma^{\lambda}, \gamma^{\nu}$ and either $\not{k}$ or $\not{p}$ but under anticommutation of these objects they all cancel in a pairwise fashion. The remaining term is that within (19.59) and the resulting loop integration reduces to (19.56), with a sole dependence on $\ell_{\perp} \cdot \ell_{\perp}$, as mentioned above.

References for identities in respective appendices: 'Field Theory in Particle Physics', B.De Wit & J.Smith, Vol. 1, 1986 and 'Renormalisation', J.Collins, 1984


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.