On the Calabi-Yau compactifications of toric Landau-Ginzburg models for Fano complete intersections (Q1646345)

The theory of Landau-Ginzburg (LG) models entered mathematics as being the mirror objects to Fano varieties. These are non-compact pencils of Calabi-Yau varieties whose singularity theory is supposed to reflect the symplectic geometry of the Fano. The most successful strand of this theory is for a complete intersection \(X\) in a toric variety \(\mathbb{T}\), with mirror \(f: \check{X} \to \mathbb{C}\). In these cases the toric action can be used to write many invariants of \(X\) as localisations of invariants on \(\mathbb{T}\). This was exploited originally by \textit{A. Givental} [Prog. Math. 160, 141--175 (1998; Zbl 0936.14031)] to give explicit descriptions of mirror symmetry for such Fano varieties, although without mention of LG models. These techniques have been applied by Coates-Iritani-Jiang to prove the crepant transformation conjecture in [\textit{T. Coates} et al., Adv. Math. 329, 1002--1087 (2018; Zbl 1394.14036)]. An ongoing project by Coates-Corti-Kasprzyk and others aims to classify all Fano varieties by instead classifying all possible LG models and giving a reconstruction technique. In terms of LG models these conditions mean that \(\check{X}\) may be taken to be just the \(n\)-dimensional torus and \(f\) a choice of Laurent series on it. Then any invariants one wants can be read off purely in terms of the arithmetic of \(f\). One can see a recent preprint of Coates-Kasprzyk-Prince on ``Laurent inversion'' for an inverse to this process. Whilst the definition of an LG model only asks for a non-compact pencil the possible compactifications are also of interest. In particular one is interested in those with smooth total space and whose anti-canonical class is the class of a fibre. Such compactifications are known as log Calabi-Yau compactifications and similar degenerations arise in many other places in mirror symmetry. See for instance works by Gross-Siebert for applications of toric degenerations to constructions of mirror families. In this paper the author considers possible compactifications of the mirror to a complete intersection in projective space. These are combinatorial the easiest models to consider and so amenable to explicitly calculating all the relevant data. In particular the author shows that such compactifications exist and describes combinatorially the divisor at infinity. The introductory section in particular is well written and sets out what one expects to see in the general theory of LG models.