First order deformations of schemes with normal crossing singularities (Q644987)

The paper under review studies deformations of schemes with normal crossing singularities, a kind of singularities appearing naturally in many different problems in algebraic geometry. The well-known Mumford's semi-stable reduction theorem states that, after a finite base change and a birational modification, any flat projective morphism \(f: \mathcal X \rightarrow C\) from a variety \(\mathcal X\) to a curve \(C\) can be brought to \(f': \mathcal X' \rightarrow C'\), where \(\mathcal X'\) is smooth and the special fibers are simple normal crossing varieties. In the compactification of the moduli space of varieties of general type, stable varieties appear in the boundary. Recall that a stable variety is a proper reduced scheme \(X\) such that \(X\) has only semi-log-canonical singularities and \(\omega_X^{[k]}\) is locally free and ample for some \(k>0\). The simplest class of non-normal semi-log-canonical singularities are the normal crossing singularities. In the Minimal Model Program, normal crossing singularities also play an important role. One of the outcomes of the Minimal Model Program starting with a smooth \(n\)-dimensional projective variety \(X\) is a Mori fiber space \(f: Y \rightarrow Z\), where \(f\) is a projective morphism, \(Y\) is a \(\mathbb Q\)-factorial terminal projective variety such that \(-K_Y\) is \(f\)-ample, and \(Z\) is normal with \(\dim Z \leq n-1\). If \(Z\) is a curve, then \(Y_z=f^{-1} (z)\) for \(z \in Z\) is a Fano variety of dimension \(n-1\) and \(Y\) is a \(\mathbb Q\)-Gorenstein smoothing of \(Y_z\). In general \(Y_z\) has non-isolated singularities and may not even be normal. It is difficult to describe the singularities of the special fibers but normal crossing singularities naturally occur and are the simplest possible non-normal singularities. Motivated by the above different problems, the author investigates the deformation spaces of varieties with normal crossing singularities. Let \(X\) be a reduced scheme with normal crossing singularities defined over a field \(k\) and let \(T^1 (X)\) denote the sheaf of first order deformations of \(X\). Then \(T^1 (X)\) is an invertible sheaf supported on the singular locus \(D\) of \(X\). In general, \(D\) is not smooth, and may not even be Cohen-Macaulay. So the author works in a suitable log resolution of \((X, D)\) and obtains explicit formulas for \(T^1 (X)\) in this setting. The paper under review is well written. In section \(1\) the author explains the motivation to study deformations of schemes with normal crossing singularities and states the main theorem. In section \(2\) the author recalls some technical results for future use. He proves the first part of the main theorem in section \(3\), where the case that \(X\) has only double point singularities is treated. The general case that \(X\) has higher multiplicity singularities is done in section \(4\). The log resolution \((X', D')\) of \((X, D)\) is obtained by successively blowing ups the singular locus of highest multiplicity of \(X\). This works in all dimensions but is not unique. In dimension at most three, the author obtains a unique log resolution \((\tilde{X}, \tilde{D})\) by running an explicit minimal model program on \((X', D')\) and proves the second part of the main theorem. At the end of the paper, the author constructs a Fano \(3\)-fold with normal crossing singularities and show that it is not smoothable.