Numerical algorithm for finding balanced metrics (Q854540)

Let \((X, L)\) be a compact complex manifold with a very ample line bundle. Let \(h\) be a Hermitian metric on \(L\), and \(\omega _h\) the corresponding Kähler form. Using the metric \(h\) it is possible to define an inner product on the \((N+1)\)-dimensional vector space \(E=H^0(X, L)\). Let \(\{S_0,\dots ,S_N\}\) be an orthonormal basis with respect to this inner product. The metric form \(\omega _h\) is called \textit{balanced} if the Bergman kernel \(\sum_{k=0}^N| S_k(p)| _h^2\) is constant on \(X\). The existence of balanced metrics is equivalent to the existence of a specific projective embedding of \(X\). In a final remark in the work ``Scalar curvature and projective embeddings, II'' (arXiv:math.DG/0407534), \textit{S. K. Donaldson} obtained a numerical algorithm for finding balanced metrics. In the present paper, the author carries out this procedure explicitly. He proves that (Theorem 1.2) if a balanced metric exists, then this can be obtained as a limit point of a sequence of certain maps defined on the sets \(\mathcal K\) and \(M\), of metrics \(h\) on \(L\) with \(\omega _h\) a positive \((1, 1)\)-form on \(X\), and Hermitian metrics on \(E\), respectively.