On the extension of a TCFT to the boundary of the moduli space (Q656992)

Given a Calabi-Yau elliptic space \(A\) (a Frobenius algebra of differential forms with extra structures), \textit{K. Costello} defined a differential form \(\omega_{g,n}\) on the moduli space \(\mathcal{M}_{g,n}\) of compact Riemann surfaces of genus \(g\) with \(n\) marked points via a construction abbreviated as TCFT (Topological Conformal Field Theory) [Geom. Topol. 11, 1539--1579 (2007; Zbl 1139.32006)]. The integral of \(\omega_{g,n}\) over \(\mathcal{M}_{g,n}\) is supposed to represent the worldsheet integral of interacting open strings. The construction involves an operator \(e^{-t\Delta}:A\to A\), where \(\Delta:=[d,d^*]\) is the Laplacian on \(A\), represented by a kernel \(K_t\in A^{\widehat{\otimes}\,2}\) called the heat kernel. Unfortunately, \(\omega_{g,n}\) is not compactly supported and diverges near the boundary of \(\mathcal{M}_{g,n}\) due to a singularity of \(K_t\) at \(t=0\), rendering the integral meaningless. This is known as the ultra-violet divergence in physics. In this paper the author studies a version of Costello's construction for \(A\) being a finite-dimensional differential graded Frobenius algebra. The divergent terms are finite in this case and \(\omega_{g,n}\) can be extended to a compactification of \(\mathcal{M}_{g,n}\). Even so, the author shows that the extension is only possible if the boundary of the compactification is large enough and has the right combinatorial structure. This structure is given by open topological field theory as axiomatized by Atiayh-Segal, and the compactification space is a quotient of the Deligne-Mumford compactification considered by \textit{E. Looijenga} [Prog. Math. 129, 369--400 (1995; Zbl 0862.14017)]. It generalizes the \textit{M. Kontsevich}'s compactification used in his famous proof of Witten's conjectures [Prog. Math. 120, 97--121 (1994; Zbl 0872.57001)]. The author demonstrates that when \(A\) is contractible his construction is in fact equivalent to the ``dual construction'' of Kontsevich. To satisfy the Atiayh-Segal axioms the algebra \(A\) must be finite-dimensional. The author conjectures that the elliptic space of Costello's construction should be replaced by a finite-dimensional algebra of ``discrete differential forms'' on a lattice to regularize the ultra-violet divergencies, and one should then remove them via renormalization. Unfortunately, the existing discrete differential geometry, due to Wilson, based on cochains of a simplicial complex comes with a non-associative cup product, so the author's construction does not apply directly.