Multigraded Fujita approximation (Q635672)

Let \(X\) be an irreducible projective variety of dimension \(d\) over an algebraically closed field \(K\) and \(D\) a big divisor on \(X\). The volume of \(D\), vol\((D)\), is defined as \(d!\) times the limit when \(m\) goes to infinity of the quotient \(h^0(X,{\mathcal O}_X(mD))/m^d\). If \(D\) is ample its volume is the self-intersection number \(D^d\) and when \(D\) is big there exists an approximation result by Fujita showing that vol\((D)\) can be approximated arbitrarily by the self-intersection of an ample divisor on a birational transformation of \(X\). This result has been generalized to graded linear series by \textit{R. Lazarsfeld} and \textit{M. Mustaţǎ} [Ann. Sci. Ec. Norm. Super. (4) 42, No. 5, 783--835 (2009; Zbl 1182.14004)] and in the paper under review to multigraded linear series using some convex bodies (\textit{Okounkov bodies}) associated to linear series. Let \(D_1, \dots, D_r\) be divisors on \(X\) and \(\overline{m}=(m_1,\dots,m_r)\) a \(r\)-uple of natural numbers. Write \(\overline{m}D=\Sigma m_iD_i\) and \(|\overline{m}|=\Sigma m_i\). A \textit{multigraded linear series} \(W_{\bullet}\) consists of vector subspaces \(W_{\overline{m}} \subseteq H^0(X,{\mathcal{O}}_X(\overline{m}D))\) verifying that \(W_{\overline{0}}=K\) and that \(W_{\overline{m}}\cdot W_{\overline{n}}\subseteq W_{\overline{m}+\overline{n}}\). Now consider \(W_{\overline{a},\bullet}\) the single graded linear series associated to \(\overline{a}D\) given by the subspaces \(W_{k\overline{a}}\) and define vol\(_{W_{\bullet}}(\overline{a})=\)vol\((W_{\overline{a},\bullet})\). The multigraded Fujita approximation of this paper states that (under some suitable conditions) for \(D_1, \cdots, D_r\) big, the volume of \(W_{\bullet}\) can be approximated by some finitely generated submultigraded linear series of \(W_{\bullet}\). Actually by \(W_{\overline{m}}^{(p)}=\Sigma\; W_{\overline{m}_1}\cdots W_{\overline{m}_k}\) the sum taken along the \(\overline{m}_i\)'s such that \(|\overline{m}_i|=kp\) and \(\overline{m}_i+\cdots +\overline{m}_k=\overline{m}\).