Generalized \(L^2\) extension theorem and a conjecture of Ohsawa (Q2376602)

A solution of a conjecture by \textit{T. Ohsawa} [Nagoya Math. J. 161, 1--21 (2001; Zbl 0986.32002); erratum ibid. 163, 229 (2001)] is claimed. The article is rather a sketch and it is claimed that the full details are to appear elsewhere. The conjecture of Ohsawa is that the following inequality holds \[ \frac{\pi^{k}}{k!}\kappa_{M}(x)\geq\kappa_{M/S}(x) \] for any \(x\in S\), where \(\kappa_{M}\) is the Bergman kernel on the diagonal of some (rather very general) Bergman space on a manifold \(M\), \(\kappa_{M/S}\) is the Bergman kernel of the Bergman space defined by means of the restrictions of sections of the canonical bundle on \(M\) to a closed complex submanifold \(S\) of codimension \(k\) with appropriately chosen \(L^2\) norm (there are some restrictions). As such, Ohsawa's conjecture can be seen as the multi-dimensional analog of Suita's conjecture concerning the relationship between the Bergman kernel and logarithmic capacity in dimension \(1\), which was recently solved. As in the case of Suita's conjecture, the optimality of the constant \[ \frac{\pi^{k}}{k!} \] is what matters in Ohsawa's conjecture. The authors focus on an extension theorem of Ohsawa-Takegoshi type from which Ohsawa's conjecture should follow as a direct application. While ideologically the statement of the theorem is clear (sections from the restriction of a bundle on \(M\) to \(S\) have extensions to the whole bundle with norm controlled by the norm of the restriction in a uniform way), the technical details of the assumptions are too complicated to be presented here.