Self-dual noncommutative \(\phi^4\)-theory in four dimensions is a non-perturbatively solvable and non-trivial quantum field theory (Q2249777)

Quantum field theory as defined by Wightman axioms relies on the manifold structure of space-time. In giving up the manifold one has to adapt the axioms. A reasonable replacement is not yet available. Although noncommutative geometry has a clear concept of compact manifolds with Euclidean signature, it is not known how to describe non-compact manifolds with Lorentzian signature. Therefore, because of the functional integral realization of the Euclidean quantum field theories on noncommutative manifolds, such models are easy to define. For this reason, the Euclidean approach is of central importance in the paper under review. At purely imaginary time, the Wightman functions become the Schwinger functions of a Euclidean quantum field theory. The advantage of the model constructed in this paper is that without knowing that the model derives from a true relativistic quantum field theory, it is possible to rigorously prove the existence of a Euclidean quantum field theory, or the existence of a statistical physics model. Many authors studied quantum space-time arising from space-time uncertainty relations at the Planck scale and showed perturbative one-loop renormalisability of a number of models. But after a while a severe problem rose at higher loop order, the so-called ultraviolet/infrared mixing. A possible way to cure this problem has been found by the authors of the paper under review in their previous work. The paper under review is a generalization of their previous work. The paper concentrates on the general framework of field-theoretical matrix models for a compact operator on a Hilbert space. The action of the group of unitaries on Hilbert space induces an infinite number of Ward identities. The most crucial step which goes far beyond the authors' previous work in the paper under review is to turn this Ward identity into a formula for the second derivative of the partition function. It is then straight forward to algebraically derive the Schwinger-Dyson equation of any correlation function of the matrix model, (and not only the planar 2-point and 4-point functions as in their previous work). Their generalization includes non-planar functions and with several boundary components. The main consequence of their algebraic formulae is that if the 2-point function is renormalisable by wavefunction and a mass renormalization, then the whole quartic matrix model is non-perturbatively renormalisable and has vanishing beta-function. So far this applies to any field-theoretical matrix model with cut-off. The authors' solution for the 2-point function and their algebraic solutions for higher correlation functions can be seen as summation of infinitely many renormalized Feynman graphs. The perturbative expansion of their exact results coincides with the Feynman graph computation.