The most attractive method underlying Dixon's talk - for most of the audience - were his transcendentality expansions. Write down all L-loop Feynman diagrams and evaluate the relevant integrals. The coefficients of different terms will be rational multiples of polynomials in "pi" and "zeta(3)" and similar values of the zeta function. Trust me that zeta(3), zeta(5), zeta(7) are transcendental numbers. Even if you don't believe me, you may abuse the language and define the degree of transcendentality "DT" of a number so that
- DT(p/q) = 0 ... for rational p/q
- DT(pi^k) = DT(zeta(k)) = k ... for positive "k"
- DT(uv) = DT(u)+DT(v)
Some coefficients don't have the form of polynomials in "zeta(k)". For example, "ln(2)" can occur as a prefactor, too. I've convinced Dixon to consider "ln(2)" to have "DT=1" until proved otherwise because
- ln(2) = 1 - 1/2 + 1/3 - 1/4 + ...
The homework exercise for the readers of the Reference Frame is to find a physical interpretation of the large transcendentality expansion - for example using the language of the AdS bulk description. In what modified theory or for what kind of processes do the leading transcendentality diagrams dominate? A mathematician has found correlations between the appearance of "zeta(3)" and "knots" in various Feynman diagrams - but is there a truly physical understanding of the diagrams? The solution to this homework could be analogous to, but more subtle than, 't Hooft's topological expansion of the Feynman diagrams in the large N limit.
Relevant references for this work of Dixon are mentioned by informed physicists in the fast comments.