Warped convolutions, Rieffel deformations and the construction of quantum field theories (Q538410): Difference between revisions

In [``Warped convolutions: a novel tool in the construction of quantum field theories'', in: E. Seiler (ed.) et al., Quantum field theory and beyond. Essays in honor of Wolfhart Zimmermann. Proceedings of the symposium in honor of Wolfhart Zimmermann's 80th birthday, Ringberg Castle, Tegernsee, Germany, 2008. Hackensack, NJ: World Scientific. 107--121 (2008; Zbl 1206.81072)] (hereafter referred to as [1]), \textit{D. Buchholz} and \textit{S. J. Summers} introduced the notion of warped convolution and applied it to quantum field theory. In this paper, a detailed definition of warped convolution and its applications to quantum field theories together with some correction of announcements of [1] are given. Warped convolution is closely related to the Rieffel product [\textit{M. A. Rieffel}, ``Deformation quantization for actions of \(\mathbb{R}^ d\)'', Mem. Am. Math. Soc. 506 (1993; Zbl 0798.46053)] (hereafter referred to as [2]). In \S2, after exposing the Rieffel product, warped convolution is defined. Let \(({\mathcal A},\mathbb{R}^n)\) be a \(C^*\)-dynamical system, where \({\mathcal A}\) is a \(C^*\)-algebra on a Hilbert space \({\mathcal H}\). The automorphism action \(\alpha\) of the group \(\mathbb{R}^n\) is implemented by \(\alpha_x(A)= U(x)AU(x)^{-1}\), \(U\) is a unitary representation of \(\mathbb{R}^n\). Let \({\mathbf A}, {\mathbf B}\) be uniformly continuous bounded \(C^{n+1}\)-class functions from \(\mathbb{R}^n\) to \({\mathcal B}({\mathcal H})\), the algebra of all bounded operators on \({\mathcal H}\), and let \(f\in{\mathcal S}(\mathbb{R}^n\times\mathbb{R}^n\)) with \(f(0,0)= 1\). Then it is shown the limit of the Bochner integral \[ \lim_{\varepsilon\to 0} (2\pi)^{-n} \int\int dx\,dy\,f(\varepsilon x,\varepsilon y) e^{-ixy}{\mathbf A}(x){\mathbf B}(y)={\mathbf A}\times{\mathbf B} \] is well-defined (Lemma 2.4). \({\mathbf A}\times{\mathbf B}\) is called the Rieffel product. Let \(E\) be the spectral resolution of \(U\), \(B\) a bounded Borel set in \(\mathbb{R}^n\), and \(F\) a finite-dimensional projection, then it is shown that \[ \int_B \alpha_{Qz}({\mathbf A})F\,dE(x)={\mathbf A}_{\alpha_Q}\times FU_B,\;\int_B dE(x) F_{\alpha_{Qx}}({\mathbf A})= U_B F\times{\mathbf A}_{\alpha_Q}. \] By using these equalities, \(\int\alpha_{Qx}({\mathbf A})\,dE(x)\Phi\) and \(\int dE(x)\alpha_{Qx}({\mathbf A})\Phi\) are defined. For a \(C^\infty\)-class \(A\), the warped convolution \(A_Q\), \(Q\) a real skew symmetric matrix, is defined by \[ A_Q= \int dE(x)\alpha_{Qx}(A)= \int \alpha_{Qx}(A)\, dE(x) \] (Def.\ 2.3). Then it is shown that \[ \|\pi_Q(A)\|\leq\| A\|_Q, \] where \(\pi_Q(A)= A_Q\) and \(\|\cdot\|_Q\) is the \(C^*\)-norm of \((C^\infty,\times_Q)\) ([2], Chap.\ 4). Hence contrary to an assertion in [1], warped convolution is defined for any bounded operator on \({\mathcal H}\) (Thm.\ 2.6). As a representation, \(\pi_Q\) is faithful and irreducible (Prop.\ 2.7) and the homomorphism \[ \pi^Q:{\mathcal A}^\infty\to{\mathcal B}({\mathcal H}),\quad \pi^Q(A)= \pi(A)_Q, \] where \((\pi,{\mathcal H})\) is an \(\alpha\)-covariant representation of \({\mathcal A}\), extends continuously to an \(\alpha\)-covariant representation of the \(C^*\)-algebra \(({\mathcal A}^Q, \times^Q)\). \(\pi^Q\) is faithful if and only if \(\alpha\) is faithful (Thm.\ 2.8). In \S3, warped convolutions are used to construct deformations of Borchers triples [\textit{H. J. Borchers}, ``The CPT-theorem in two-dimensional theories of local observables'', Commun. Math. Phys. 143, No.~2, 315--332 (1992; Zbl 0751.46045)]. A~Borchers triple relative to \({\mathcal W}= \{(x_0,x_1,\dots, x_{n-1})\in \mathbb{R}^n: x_1\geq |x_0|\}\) is a triple \(({\mathcal R}, U,\Omega)\), where \({\mathcal R}\subset{\mathcal B}({\mathcal H})\) is a von Neumann algebra, \(U\) a unitary representation of \(\mathbb{R}^n\) in \({\mathcal H}\) whose spectrum is contained in the closed light cone \(V_+\) and satisfies \(\alpha_x({\mathcal R})\subset{\mathcal R}\), \(x\in{\mathcal W}\), and \(\Omega\) is a unit vector invariant under the action of \(U\) and is cyclic and separating for \({\mathcal R}\) (Def.\ 3.1). Tomita-Takesaki theory is applicable to the pair \((U,\Omega)\). The associated modular operator and involution are denoted by \(\Delta\) and \(J\). Borchers triples can be used as a building block for the construction of a quantum field theory in space-time dimension two and any Borchers triple defines a relativistic quantum field theory in two dimensions. To construct deformations of Borchers triples, \(Q\) needs to satisfy \(QV_+\subset{\mathcal W}\). Then, denoting the double commutator of \(\{A_Q: A\in{\mathcal R}\cap{\mathcal C}^\infty\}\) by \({\mathcal R}_Q\), it is shown that \(({\mathcal R}_Q, U,\Omega)\) is a Borchers triple and its associated modular operator and involution are \(\Delta\) and \(J\) (Thm.\ 3.3, Thm.\ 3.5). To apply warped convolutions to relativistic quantum field theory in more than two space-time dimensions, Borchers triples are replaced to causal Borchers triples in \S4, the last section. A~causal Borchers triple \(({\mathcal R}, U,\Omega)\) relative to \({\mathcal W}\) is defined similar to a Borchers triple, but \(U\) is a unitary representation of the Poincaré group (Def.\ 4.1). If the underlying algebra cannot be enlarged without coming into conflict with causality, then the triple is said to be maximally causal (Def.\ 4.2). Warping procedures of causal Borchers triples are similar to those done in \S3, and \(({\mathcal R}_Q, U,\Omega)\) is again a causal Borchers triple and it is maximal if the original triple is maximal (Thm.\ 4.2, Prop.\ 4.4). In general, such deformations are nontrivial, which is shown in the Appendix. The authors also remark that in more than two space-time dimensions the warped algebra can, in general, not be interpreted in terms of some underlying point fields.