Symmetric differentials and the fundamental group (Q2016423): Difference between revisions

This is a very interesting paper. The authors solve the representation version of the Esnault problem: Does the infinite fundamental group of a compact projective manifolds imply the existence of a nonzero section of \(S^i \Omega ^1\) for some \(i>0\)? The authors prove the following theorem. Theorem 0.1. Any compact Kähler manifold whose fundamental group has a finite dimension representation with infinite image must have a nonzero symmetric differential (a section of a symmetric power of the cotangent bundle). We notice that \(H^0 (X, \Omega ^1_X )\neq 0\) if and only if \(\pi _1 X/[\pi _1 X, \pi _1 X]\) is infinite. In Section 5, the authors prove that Theorem 0.1 is true if the representation is over a field with a characteristic \(p>0\). Then, one can concentrate on the case of a complex representation. In Section 4, the authors conclude that if the representation is nonrigid, then it is true and \(1\leq i\leq n\), where \(n\) is the complex dimension. This is Theorem 4.1 and it is due to the second author in [Invent. Math. 192, No. 2, 257--286 (2013; Zbl 1269.22006), Theorem 1.6] and [\textit{D. Arapura}, Int. Press Lect. Ser. 3, No. II, 605--624 (2002; Zbl 1048.32006), Proposition 2.4] for the projective case. Then on page 2810, the authors say that one can consider the complex representation being rigid and according to [\textit{C. T. Simpson}, Publ. Math., Inst. Hautes Étud. Sci. 75, 5--95 (1992; Zbl 0814.32003), Corollary 4.2], the semisimplified representation can be made into a complex variation of Hodge structure over the manifold. If the semisimplified representation only have a finite image, then there is a finite covering such that the fundamental group have a representation into \(U\) the strictly upper triangular matrices of some \(\mathrm{GL}(m, {\mathbb C})\), which has an infinite image. That would induce some symmetric differentials. Therefore, one could consider now that the representation is semisimple and hence a complex variation of Hodge structures. Of course, the image of \(\pi _1\) might not be discrete. But, one could assume that the representation is over an algebraic number field since the representation is rigid. Then, it can be regarded as a rational representation. Passing to the charateristic \(p>0\), one can prove Theorem 0.1 if any of them has an infinite image. Now, we consider the case in which all of them has a finite image. Then, the original representation is a representation over \({\mathbb Z}\). This implies that the image is discrete. In the case when the image is discrete, the authors obtain Corollary 3.2 with a period map \(\phi : X \rightarrow Z=D/\Gamma\). There, \(Z\) is a standard manifold with nonpositive bisectional holomophic curvatures and the image \(Y\) of \(\phi\), i.e., \(Y=\phi (X)\), is big. That is, the resolution of \(Y\) is big. That induces the nonzero symmetric differential from the smallest number \(a\) such that there is a positive number \(C\) with \(\sum _{i\leq j} H^0 (Y, S^i \Omega ^1) <Cj^a\) for all \(j>0\) is \(2\dim Y\). See also the author's summary: ``We show that any smooth complex projective variety whose fundamental group has a complex representation with infinite image must have a nonzero symmetric differential (a section of a symmetric power of the cotangent bundle). Along the way, we produce many symmetric differentials on the base of a variation of Hodge structures.''