Proof of Wahl's Conjecture on Surjectivity of the Gaussian Map for Flag Varieties (Q5285562)

Let \(L\) be a line bundle on a projective variety \(X\). The Gauss-Wahl map is the homomorphism \(\Phi_ L:\bigwedge^ 2\Gamma(X,L)\to\Gamma(X,L^{\otimes 2}\otimes\Omega^ 1_ X)\) defined by the formula \(\Phi(s\wedge t)=sdt-tds\). This map was introduced by \textit{J. Wahl} [Duke Math. J. 55, 843-871 (1987; Zbl 0644.14001)] who found the relationship of the cokernel of this map to the deformation theory. Since then many interesting results on the Gauss-Wahl map were obtained by various people, however most of them are applied only to the case of curves. -- In this paper the surjectivity of the Gauss-Wahl map is proven for the case of flag varieties. In fact, a more general result, conjectured by Wahl, is proven here. First, instead of considering one line bundle \(L\) one may consider two line bundles \(L\) and \(M\) and define the Gauss-Wahl map as a map from the kernel \(R_ 1(L,M)\) of the natural multiplication map \(\Gamma(X,L)\otimes\Gamma(X,L)\to\Gamma(X,L\otimes M)\) to the space \(\Gamma(X,L\otimes M\otimes\Omega^ 1_ X)\). Interpreting \(\Omega^ 1_ X\) as the sheaf \(I/I^ 2\), where \(I\) is the ideal sheaf of the diagonal of \(X\times X\), we obtain that \[ R_ 1(L,M)=\Gamma(X\times X,I\otimes L\boxtimes M). \] Then we can define, more generally, the spaces \(R_ i(L,M)=\Gamma(X\times X,I^ i\otimes L\boxtimes M)\) which fit in the exact sequences \[ 0\to R_{i+1}(L,M)\to R_ i(L,M)\to\Gamma(X,L\otimes M\otimes S^ i(\Omega^ 1_ X)). \] The last map can be viewed as a generalization of the Gauss-Wahl map. To prove its surjectivity it suffices to show that the cohomology groups \(H^ 1(X\times X,R_{i+1}(L,M))\) vanish. The main result of this paper is that \(H^ 1(X\times X,R_ 2(L,M))=0\) in the case when \(X=G/P\) is a projective homogeneous space of a semi- simple complex Lie group \(G\), \(L\) and \(M\) are ample homogeneous line bundles. The proof uses extensively the theory of representations of semisimple groups. -- It is expected that its generalization to higher \(R_ i(L,M)\) will provide new results on the decomposition of tensor product of representations.