Positive topological quantum field theories (Q887027)

A great mathematical obstacle in quantum field theory is to make precise sense of state sums (or path integrals), which calculate expected values of observables on an infinite-dimensional space of fields. This can be done successfully in quantum mechanics, where the space of fields is a space of paths and the relevant measure for integration is the Wiener measure; however, higher dimensional analogues of the Wiener measure have proven hard to construct. The paper under review explores a purely algebraic remedy to this problem, namely allowing state sums to take values in a complete semiring. A \textit{complete} semiring is a semiring endowed with a summation law (indexed by sets of arbitrary cardinality) compatible, in the expected way, with the semiring operations. One example is the set \(\mathbb R_+^\infty = [0, +\infty) \cup \{ \infty \}\) with the usual algebraic operations and summation law. A second example is the tropical semiring structure on the set \(\mathbb R^\infty_+\), where summation is given by the infimum and the product is given by addition of real numbers. By the so-called ``Eilenberg swindle'', one sees that the presence of a summation law precludes the existence of additive inverses, hence the need to only allow semirings. One important example of such kind of algebraic structure is the convolution monoid \(Q = Q_S(\mathbf C)\) of a strict monoidal category \(\mathbf C\); it admits a summation law if so does the coefficient semiring \(S\), and there are two products on \(Q_S(\mathbf C)\) making it into a complete semiring: one is determined by composition of morphisms in \(\mathbf C\), and the other by the tensor product. These are denoted \(Q^c\) and \(Q^m\) respectively. Following the ideas of Atiyah and others, the author defines a positive topological field theory (positive TFT) to be an assignment of a \textit{state module} \(Z(M)\) to each closed \(n\)-manifold \(M\) and a \textit{state sum} \(Z_W \in Z(\partial W)\) to each \((n+1)\)-dimensional bordism \(W\). \(Z(M)\) is required to be a semialgebra over both \(Q^c\) and \(Q^m\), and these two different algebraic structures play a role in the axioms imposed on \(Z(M)\) and \(Z(W)\) concerning gluing of bordisms and disjoint union of bordisms, respectively. Then, a quantization procedure is given: starting from a system of fields and a system of action functionals taking values in \(\mathbf C\), a positive TFT is constructed. The (often nonrigourous) integrals needed to calculate state sums are replaced by summations in \(Q_S(\mathbf C)\). The remainder of the paper mostly deals with examples and applications. This includes the sketch of a field theory that detects exotic structures on spheres. The definitions and constructions in this paper are very general and allow many variations and refinements (in particular, a positive TFT is not defined as a monoidal functor), which explains in part its considerable length.