Higher genus symplectic invariants and sigma models coupled with gravity (Q1385171)

In a previous paper [\textit{Y. Ruan} and \textit{G. Tian}, J. Differ. Geom. 42, No. 2, 259-367 (1995; Zbl 0860.58005)], the authors provided one possible rigorous mathematical framework for the physicists' theory of quantum cohomology rings. Basically, they developed a mathematical foundation for properly defining a quantum cohomology ring on a given semi-positive symplectic manifold. This incorporated the existence of so-called higher-genus symplectic invariants not coupled with gravity (topological sigma models), which were shown to obey a certain composition law (i.e., an associativity law with respect to the multiplication in the quantum cohomology ring) as it was required (and predicted) by physicists. The present work is a direct continuation of that foregoing paper. The starting point is the observation, first made by E. Witten, that topological gravity should be reflected by certain relations, again suggested by the physical interpretation of the theory, between intersection numbers in the moduli space of algebraic curves of given genus \(g\) and invariants associated with the Korteweg-de Vries hierarchy. This stipulated interrelation has recently been clarified by \textit{M. Kontsevich} [Commun. Math. Phys. 147, No. 1, 1-23 (1992; Zbl 0756.35081)] and, in the sequel, by several other authors. However, these geometric-numerical phenomena become much more interesting (and theoretically intriguing) if the topological sigma models are assumed to be coupled with topological gravity. It was again \textit{E. Witten} who proposed an approach to tackling topological sigma models coupled with gravity [J. Differ. Geom., Suppl. 1, 243-310 (1991; Zbl 0808.32023)]. Witten's approach culminated in a very crucial conjecture about the basic features of his new coupled model. The purpose of the authors' present paper is to establish a rigorous mathematical foundation for the theory of topological sigma models coupled with topological gravity over semi-positive symplectic manifolds, very much so along the path they had prepared in their previous work cited above. To this end, they first associate, to each semi-positive symplectic manifold \(V\), a topological sigma model with gravity, together with suitable Gromov-Witten invariants. Using the basic techniques developed in their foregoing paper, the existence of the new Gromov-Witten invariants is established by the authors in a rather general, unified way. Moreover, the Gromov-Witten invariants are shown to be symplectic invariants, which is surprising even from the physical viewpoint, and they can factually be taken as correlation functions for the theory, just as required in the axiomatic setup of topological field theory. The next fundamental result then states that the new Gromov-Witten invariants satisfy the required conservation law that governs the behaviour of the invariants under degenerations of stable curves. Again, the proof of this crucial fact is based on the authors' general techniques developed in their previous paper, and, being of this sort, it appears to be rather general. Finally, E. Witten's arguments that led him to his aforementioned conjecture on topological sigma models with gravity (``Witten conjecture'') are then translated in terms of the author's new Gromov-Witten invariants, and this leads to a variant of the so-called ``generalized Witten conjecture'' for semi-symplectic ground-manifolds. The subtle, in parts very technical, yet always deep-going analysis transpires a wealth of particular results that are related to a (hopefully) affirmative answer to Witten's conjecture in the general case, and this makes the paper under review another milestone in the development of general quantum cohomology theory.