“…Given a Lagrangian fibration , there is a natural affine structure on , for which the discriminant locus agrees with the discriminant locus of [19, 20, 55]. Moreover, this affine structure is integral if and only if the symplectic form on represents an integral cohomology class by [52, Remark 5.10] — see also [32, Remark 1.2]. Conversely, fixing an affine manifold with singularities, describing the obstructions to constructing a torus fibration over it is a technically challenging problem, and is discussed further in Appendix B.…”