A functorial LMO invariant for Lagrangian cobordisms (Q929206)

The LMO invariant is defined for all closed oriented 3-manifolds, but is powerful only for homology \(3\)-spheres where it is a universal finite type invariant and a universal quantum invariant. Quantum invariants extend to functors (TQFT's) from compact oriented \(3\)-manifolds with boundary, so their universal object should do so as well. The first such extension of the LMO invariant to a TQFT was carried out by \textit{J. Murakami} and \textit{T. Ohtsuki} [Commun. Math. Phys. 188, No. 3, 501--520 (1997; Zbl 0938.57022)], but their solution has a complicated anomaly. Ten years later, \textit{D. Cheptea} and \textit{T. Le} [Commun. Math. Phys. 272, No.~3, 601--634 (2007; Zbl 1145.81049)] succeeded in constructing an anomaly-free TQFT for the LMO invariant. This important paper offers a new extension of the LMO invariant to \(3\)-manifolds with boundary, which in the reviewer's opinion seems the most natural. This extension is a functor, which the authors call the LMO functor. Its domain is chosen to be the category of Lagrangian \(q\)-cobordisms between once-punctured surfaces. This is designed to be the largest category of cobordisms in which the property of `capping-off to an integral homology \(3\)-sphere' is preserved under composition, and it includes the categories of integral homology cubes and homology cylinders and sub-categories. The LMO functor is a universal finite-type invariant for Lagrangian cobordisms, and it induces Cheptea-Le's TQFT for first Betti number zero by restricting to a certain natural sub-category and `capping off'. A certain `tree-reduction' of this functor gives a powerful invariant of the Torelli group, which deserves further study. Surely this invariant will lead to a deeper understanding of the Torelli group. A natural question to ask is what the relationship is between the LMO functor and the Reshetikhin-Turaev TQFT's.