Copositive geometry of Feynman integrals

From MaRDI portal





The paper connects the theory of copositive matrices (corresponding to copositive quadratic forms, that is, quadratic forms that are nonnegative on the positive orthant or equivalently, linear functionals that are nonnegative over the cone of nonnegative matrices) to Feynman integrals in physics, themselves related to scattering amplitudes. The goal of the paper is to give mathematical foundations to this connection.\N\NThe paper associates to a Feynman graph a full dimensional closed convex semialgebraic cone, the copositive cone \(\mathcal{C}_G\) of such graph. It is known that testing copositivity is NP-hard, since there are NP-hard problems from combinatorial optimization that reduce to this test. In the context of this paper, this means that testing whether a vector belongs to \(\mathcal{C}_G\) is a hard problem. This relates to the hardness of the convergence of Feynman integrals. Some certificates for the membership in the copositive cone can be given via a combination of linear and semidefinite programming, however this fails in high dimension (\(n>4\)). Moreover we also know from a work of Bodirsky, Kummer and Thom that the copositive cone is not a spectrahedral shadow.\N\NThe paper studies the copositive cone for the special case of the Symanzik polynomial of a Feynman graph. The algebraic boundary of this cone is characterized via Landau discriminants, a polynomial of degree exponential in the number of internal particles of the Feynman graph. Finally copositive geometry is developed in a support-based framework such as the one in the book by Gel'fand-Kapranov-Zelevinsky.











This page was built for publication: Copositive geometry of Feynman integrals

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6951653)