Exact results for perturbative Chern-Simons theory with complex gauge group (Q1032776)

From the physics point of view, on a 3-dimensional manifold \(M\), the Chern-Simons theory with a compact gauge group \(G\) has been well developed since the 1980s. Moreover the theory has led to important mathematical applications in constructing various topological invariants of \(M\) and knots inside it. The paper under review initiates the study of the Chern-Simons theory with complex gauge group \(G_{\mathbb{C}}\) which is non-compact. It is pointed out that the new physics theory is considerably different from the original theory with compact \(G\). For example, in terms of TQFT, the Hilbert space \(\mathcal{H}_\Sigma\) assigned to a surface \(\Sigma\) is infinite dimensional. For this reason the non-perturbative approach is difficult to adopt hence in this paper the authors concentrate on the computation perturbatively. As before, physically the quantum \(G_{\mathbb{C}}\)-invariant \(Z(M)\) is defined as the path integral of the Chern-Simons action \(S\). Here \(S\) contains two coupling constants \(t,\bar{t}\). To compute \(Z(M)\) perturbatively invloves some holomorphic and anti-holomorphic terms \(Z^{(\rho)}(M;t), Z^{(\rho)}(M;\bar{t})\), where \(\rho\) corresponds to a flat connection on \(M\). Applying the stationary phase approximation to \(Z(M)\), it is sufficient to compute the coefficients \(S^{(\rho)}_n(M)\). Unlike the compact \(G\) case, the authors emphasize that the new \(G_{\mathbb{C}}\)-invariants \(S^{(\rho)}_n(M)\) are not of finite type and typically not rational numbers. The paper proposes four approaches to computing \(Z^{(\rho)}(M)\) or \(S^{(\rho)}_n(M)\) through ``Feynman diagrams'', ``Quantization'' of the moduli space of flat connections over \(\Sigma\)'', ``Analytic continuation'' based on the \(G\) Chern-Simons theory, and ``State sum model''. The last approach has a new feature that one has to use an integral, not a discrete sum, over the states arising from a triangualation of \(M\), since the labeling representations of \(G_{\mathbb{C}}\) form a continuous space.