Observing that the Hamiltonian of the renormalisable scalar field theory on 4-dimensional Moyal space A is the square of a Dirac operator D of spectral dimension 8, we complete (A, D) to a compact 8-dimensional spectral triple. We add another Connes-Lott copy and compute the spectral action of the corresponding U(1)-Yang-Mills-Higgs model. We find that in the Higgs potential the square φ 2 of the Higgs field is shifted to φ ⋆ φ + const · X µ ⋆ X µ , where X µ is the covariant coordinate. The classical field equations of our model imply that the vacuum is no longer given by a constant Higgs field, but both the Higgs and gauge fields receive non-constant vacuum expectation values. 1 harald.grosse@univie.ac.at 2 raimar@math.uni-muenster.de1 coordinates was wrong and with it our original conclusions. For a discussion of the vacuum configuration of this type of action we refer to [50].
Introduction
Renormalisable field theories on Moyal spaceRenormalisable field theories on Moyal space are by now in mature state. In the first renormalisation proof [1], the matrix base of the Moyal plane was a central philosophy, because we wanted to avoid convergence subtleties with the oscillating integrals in momentum space. We traded the simple matrix product interaction in for a complicated (but manifestly positive) propagator and used exact renormalisation group equations to estimate the ribbon graphs. The technically most challenging part was a brute-force analysis [2] of all possible contractions of ribbon graphs. The scale analysis led to the existence of an additional marginal coupling in the φ 4 -model, which corresponds to a harmonic oscillator potential for the free field. Later on, we interpreted this term as required by Langmann-Szabo duality [3]. A summary of these ideas can be found in [4].The renormalisation proof was considerably simplified by switching to multi-scale analysis as the renormalisation scheme. The first version still relied on the matrix base [5]. Once the bounds for the sliced propagator being proven (which is tedious), one obtains in an efficient way the power-counting theorem in terms of the topology of the graph. Subsequently, the renormalisation proof was also achieved by multi-scale analysis in position space (which is equivalent to momentum space by Langmann-Szabo duality) [6], showing the equivalence of various renormalisation schemes. Recently, the position space amplitude of an arbitrary orientable graph was expressed as an integral over Symanzik type hyperbolic polynomials [7]. With all inner integrations carried out, this is the most condensed way of writing Feynman graph amplitudes. See also [8] for the more complicated case of "critical" models.Additionally, we noticed that the β-function of the renormalisable noncommutative φ 4model tends to zero at large energy scales. This is opposite to the commutative case and supports the hope that a non-perturbative construction of the model is within reach [9,10]. The one-loop β-function was first computed in [11] (its peculiar feature was noticed ...