Calabi flow and projective embeddings (Q987987): Difference between revisions

Given a complex submanifold \(X\subset\mathbb{CP}^N\), the author defines a flow to deform the given projective embedding through a family of projectively equivalent embeddings, whose fixed points are balanced embeddings. Note that in general a given embedding is projectively equivalent to a balanced one if and only if the Chow point of the embedding is GIT polystable. This flow, called the balacing flow, is simply the gradient flow of the \(SU(N+1)\)-moment map acting on the Chow variety. If now \((X,L)\) is a compact Kähler manifold polarized by an ample line bundle, then we can use the sections of \(L^k\) for high powers \(k\) to embed \(X\) into higher and higher projective spaces, and get a sequence of balancing flows. If we pull back the Fubini-Study metric by these embeddings, we get flows of Kähler metrics \(\omega_k(t)\) cohomologous to \(c_1(L)\). The main question that the author addresses is the asymptotic behaviour of these metrics as \(k\) goes to infinity. Take now any Kähler metric \(\omega\) in \(c_1(L)\) and evolve it into a family \(\omega(t)\) (which exists on some time interval \([0,T]\)) that solves the Calabi flow \[ \frac{\partial\omega(t)}{\partial t}=i\partial\overline{\partial}R(\omega(t)). \] For any \(k\) large take a basis of sections of \(L^k\) which is orthonormal with respect to the \(L^2\) inner product defined by \(\omega\), get a projective embedding of \(X\) and pull back the Fubini-Study metric, to get Kähler metrics \(\omega_k\). Now run the (suitably sped-up) balacing flow starting at \(\omega_k\) and get metrics \(\omega_k(t)\). The main theorem of the paper says that, in this case, if \((X,L)\) has discrete automorphisms, then for any \(0\leq t\leq T\) the metrics \(\omega_k(t)\) converge smoothly to \(\omega(t)\) as \(k\) goes to infinity. This paper is highly influenced by the seminal work of \textit{S. K. Donaldson} [J. Differ. Geom. 59, No. 3, 479--522 (2001; Zbl 1052.32017)], where the ``static'' version of this problem is solved: constant scalar curvature Kähler metrics (with discrete automorphisms) are limits of balanced metrics. The main technical ingredients of the proof are similar to Donaldson's, except that a more refined analysis of the Bergman kernel is needed. This is provided by the appendix written by K. Liu and X. Ma.