On the theta divisor of SU(r; 1)

From MaRDI portal
Publication:4545206





Let \(C\) be a compact Riemann surface of genus \(g\geq 2\), and denote by \(\text{SU}(r,d)\) the moduli space of semi-stable vector bundles over \(C\) of rank \(r\) and with fixed determinant bundle \(M\in\text{Pic}^d(C)\). It is well-known that \(\text{SU} (r,d)\) is an irreducible projective variety of dimension \((r^2-1)\cdot (g-1)\) and with infinite cyclic Picard group. The ample generator \(L\) is called the generalized theta divisor in \(\text{SU}(r,d)\). The associated theta map \(\varphi_L: \text{SU}(r,d) \to|L|^*\) is known to be an embedding if \(r=2\), whereas the case of \(r\geq 3\) is still much less understood. This paper under review turns to the special case of \(r\geq 3\), \(d=1\) and \(M={\mathcal O}_C(p)\) where \(p\in C\) is any fixed point. As to this step towards the general theory of \(\text{SU} (r,d)\), the authors prove the following two theorems: (1) Under the above-mentioned assumptions, one has that \(\varphi_L\) is indeed a morphism (i.e., the linear system \(|L|\) on \(\text{SU}(r,1)\) is base-point free), and the equality \(\deg(\varphi_L)=1\) holds. (2) For any curve \(C\) of genus \(g\geq 2\), the theta map \(\varphi_L\) is generically one-to-one on \(\text{SU} (r,1)\). This implies, among other things, that \(\text{SU} (r,1)\) is a rational variety. Thus the authors obtain an alternative proof of this crucial result already discovered by \textit{P. E. Newstead} in 1975 [Math. Ann. 215, 251-268 (1975; Zbl 0288.14003)].











This page was built for publication: On the theta divisor of SU(r; 1)

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4545206)