Feynman amplitudes and Landau singularities for one-loop graphs (Q719222)

This paper discusses finite one-loop Feynman integrals and its thresholds from a perspective of variations of Hodge structures. It gives a nice example for the interplay between mathematics and physics. From the physics side, the theory of one-loop Feynman integrals is well understood. One-loop Feynman integrals can be grouped according to the number of external particles \(n\). In particular it is known that all one-loop integrals with \(n>4\) can be reduced to \(n \leq 4\) integrals. If the integral is finite, we can work in four space-time dimensions and the statement above follows from the fact that in a vector space of dimension \(4\) one cannot have more than \(4\) linearly independent vectors. The generalisation to dimensionally regulated one-loop integrals is also known, the only modification in \(D=4-2\varepsilon\) space-time dimensions is that at \({\mathcal O}(\varepsilon)\) one-loop pentagon (\(n=5\)) integrals appear. Therefore the number of basic one-loop Feynman integrals is finite and by now all of them have been calculated. For finite integrals or dimensionally regulated integrals up to order \({\mathcal O}(\varepsilon^0)\) we have the following observation: All results are of the form of a sum of algebraic functions of the masses and the momenta times two transcendental functions, whose arguments are again algebraic functions of the masses and the momenta. The two transcendental functions are the logarithm and the dilogarithm. The paper by S. Bloch and D. Kreimer restricts itself to finite integrals, which implies \(n\geq 3\) (to avoid ultraviolet divergences) and non-zero internal masses (to avoid infrared divergences). It relates the occurrence of the logarithm and the dilogarithm to a mixed Hodge structure. It is very well written and offers the reader valuable insight how mathematics and physics are interleaved. Let me mention a few technical points: The authors separate the discussion into the case \(n \geq 4\) and \(n=3\). The reason for this is given by the fact, that for \(n\geq 4\) only the second Symanzik polynomial appears in the denominator, while for \(n=3\) both the first and the second Symanzik polynomial appears in the denominator. The authors call the case \(n=3\) exceptional. In view of a generalisation to higher loops I would rather call this case ``generic'', and the cases with only one polynomial ``special''. The reader interested in the case where two polynomials appear in the denominator will find very valuable information in Section 10 of this paper. In the discussion of the case \(n\geq 4\), Lemma 6.3 is essential. To the best of my knowledge, this has already been proven in [\textit{G. Duplančić} and \textit{B. Nižić}, ``Reduction method for dimensionally regulated one-loop N-point Feynman integrals'', Eur. Phys. J. C, Part. Fields 35, No. 1, 105--118 (2004; Zbl 1191.81116), \url{arXiv:hep-ph/0303184}].