Consider $d$-dimensional gamma matrix structures. I have an expression like $$ \sum_{h_2=\pm}\text{Tr}(\not{\xi}_2\not{p}_3\bar{\not{\xi}}_2\not{p}_1), $$ where $\not p=p^\mu \eta_{\mu\nu}\gamma^\nu$ are the momenta contracted with the gamma matrices and $\not \xi=\xi^\mu \eta_{\mu\nu}\gamma^\nu$ are the helicity vectors contracted with the gamma matrices. $h_2$ is the helicity corresponding to $\xi_2$. $\bar{\not\xi}_i$ is the same helicity vector as $\not\xi_i$, but with opposite helicity.

We have $p_i^\mu\eta_{\mu\nu}\xi^\nu_i=0$.

I know things like the anti-commutation relation $\{\gamma^\mu,\gamma^\nu\}=-2\eta^{\mu\nu}$ or the completeness relation $$ \sum_{h_i=\pm}\not{\xi}_i\bar{\not{\xi}}_i=2\cdot\text{Id}, $$ but I don't know how to handle the $\xi$'s with another expression between them. The final expression should not be depending on the $\xi$'s anymore. The process I'm looking at is a tensor structure involving fermion and gluon scattering. The helicity vectors belong to gluons.

I would be very happy if anybody could help me with the evaluation of the first expression.

Edit: After I have mentioned it in the comments, the term I have a problem with in particular is $$ \text{Tr}(\not{p}_3\not{\xi}_3\not{\xi}_4\not{p}_4\not{p}_1\not{p}_3\bar{\not{\xi}}_3\bar{\not{p}}_4\not{p}_4\not{p}_1). $$ As far as I know, the solution of this expression depends on the dimension $d$. Hence, I do not think the usual commutation relations lead to a result. Currently, I am spending my time figuering out how FeynCalc, mentioned by Luc J. Bourhis, works. But for the possibility of somebody providing more ideas, I wanted to mention this in the original question as well. What I have been told recently, is that $$ \sum_{h_i=\pm}\text{Tr}(\dots\not\xi_i\not p_j\bar{\not\xi_i}\dots)=f(d)\cdot\sum_{h_i=\pm}\text{Tr}(\dots\not p_j\dots) \text{ for }i\neq j $$ with $f(d)$ being some rational function of the dimension $d$. I have some guesses about $f$, but I will not write them here, because I have many dobts concerning them.

  • $\begingroup$ I would just use $$\newcommand{\slashed}[1]{#1\!\!\!/}\mathrm{Tr}(\slashed{a}\slashed{b}\slashed{c}\slashed{d})=4(ab)(cd)-4(ac)(bd)+4(ad)(bc)$$ $\endgroup$ – user154997 Jul 7 '17 at 11:41
  • $\begingroup$ Thank you! I must admit that I have forgotten about that identity. Still, what I was hoping for, was an identity for the $\xi$'s. The actual expression I wanted to handle is $$\text{Tr}(\not{p}_3\not{\xi}_3\not{\xi}_4\not{p}_4\not{p}_1\not{p}_3\bar{\not{\xi}}_3\bar{\not{p}}_4\not{p}_4\not{p}_1)$$ which is much more complicated. I was hoping there would be an identity for the polarisation vectors such that this expression simplifies... I am looking at light like momenta with $\sum_{i=1}^4p_i=0$ btw. $\endgroup$ – Fred Jul 7 '17 at 15:07

For the calculation in your question, you could just use:


For the longer one in the comments, you could do what every does these days: use a program! I am afraid I don't remember a specific result for such expressions with polarisation but I can show you how to proceed by hand. Note that all the formula I will use are valid in dimensional regularisation, with dimension $d=4-2\epsilon$ typically then.

So I will move $\not{\bar{\xi}}_3$ to the left.

$$\begin{gather}\text{Tr}(\not{p}_3\not{\xi}_3\not{\xi}_4\not{p}_4\not{p}_1\not{p}_3\bar{\not{\xi}}_3\not{p}_4\not{p}_4\not{p}_1) = \\ -\text{Tr}(\not{p}_3\not{\xi}_3\not{\xi}_4\not{p}_4\not{p}_1\bar{\not{\xi}}_3\not{p}_3\not{p}_4\not{p}_4\not{p}_1)\\ -2(\xi_3p_3)\text{Tr}(\not{p}_3\not{\xi}_3\not{\xi}_4\not{p}_4\not{p}_1\not{p}_4\not{p}_4\not{p}_1) \end{gather}$$

You can then repeat that. When $\not{\bar{\xi}}_3$ will pass through $\not{\bar{\xi}}_4$, you will get a factor $(\xi_3\xi_4)$, and then you will be left with

$$\sum_{h_3=\pm} \text{Tr}(\cdots \not{\xi}_3\bar{\not{\xi}}_3\cdots)$$

which you can simplify with the completeness relation.

You will still be left with traces of 8 gamma matrices of course. They can be evaluated with the recurrence trick:

$$\begin{align}\text{Tr}(\not{a}_1\cdots\not{a}_{2n})&= (a_1a_2)\text{Tr}(\not{a}_3\cdots\not{a}_{2n})\\ &-(a_1a_3)\text{Tr}(\not{a}_2\not{a}_4\cdots\not{a}_{2n})\\ &+\cdots\\ &+(a_1a_{2n})\text{Tr}(\not{a}_2\cdots\not{a}_{2n-1}) \end{align}$$

I think it would be clever to use one of the leftover $\xi$ for $a_1$, after a circular permutation of the trace.

As I mentioned above, I recommend using a program. Nowadays everybody seems to use the Mathematic package FeynCalc. I toyed a bit with it a few years ago and I found it pretty comprehensive if a bit short on the documentation but there is a significant community to ask questions to. Note that I used to rely on an ancient package called Tracer. I would not recommend it (not even sure it still runs on recent Mathematica). I have just noticed that FeynCalc has even a command Polarization but I haven't tried it!

  • $\begingroup$ They are valid in $d=4-2\epsilon$ dimensional regularisation. $\endgroup$ – user154997 Jul 7 '17 at 17:02
  • $\begingroup$ Thank you very much for this answer. This is what I've been afraid to do, but I guess I won't really have a choice. So thank you very much! Unfortunately, I haven't yet found a good program to handle it in $d$ dimensions, but anyway, I will probably be able to handle it. I will wait one or two days to see whether somebody else has an even more beneficial answer, but I guess that's it then. $\endgroup$ – Fred Jul 8 '17 at 11:59
  • $\begingroup$ What CAS do you have access to? Mathematica? Maple? None? $\endgroup$ – user154997 Jul 8 '17 at 12:02
  • $\begingroup$ Mathematica, Maple, Matlab (all from university), maybe more. But I only have experience with Mathematica. $\endgroup$ – Fred Jul 8 '17 at 12:04
  • $\begingroup$ Great! I amended my answer with some recommendations then. $\endgroup$ – user154997 Jul 8 '17 at 14:43

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.