Real algebraic hypersurfaces in complex projective varieties (Q1345930)

Let \(M\) be a compact affine nonsingular real algebraic variety of dimension \(m\). Denote by \(H^{\text{alg}}_{m-1} (M, \mathbb{Z}/2)\) the subgroup of the homology group \(H_{m-1} (M, \mathbb{Z}/2)\) which consists of all homology classes represented by Zariski closed real algebraic hypersurfaces of \(M\). Let \(H^ 1_{\text{alg}} (M,\mathbb{Z}/2)\) be the image of \(H^{\text{alg}}_{m-1} (M, \mathbb{Z}/2)\) under the Poincaré duality isomorphism \(H_{m-1} \to H^ 1\). The group \(H^ 1_{\text{alg}} (M,\mathbb{Z}/2)\) is an important invariant with numerous interesting applications. In the paper under review one investigates the group \(H^ 1_{\text{alg}} (M, \mathbb{Z}/2)\) where \(M\) is the underlying real algebraic variety of a complex projective nonsingular variety. Every complex projective variety \(V\) can be, in the obvious way, considered as a real algebraic variety and, as such, will be denoted by \(V_ \mathbb{R}\). Of course, \(\dim V_ \mathbb{R} = 2 \dim V\). Moreover, \(V_ \mathbb{R}\) is an affine real algebraic variety and if \(V\) is nonsingular, then \(V_ \mathbb{R}\) is also nonsingular. Define \(d(V) = \dim_{\mathbb{Z}/2} H^ 1_{\text{alg}} (V_ \mathbb{R}, \mathbb{Z}/2)\). In many cases, \(d(V)\) is strictly less than \(\dim_{\mathbb{Z}/2} H^ 1(V_ \mathbb{R}, \mathbb{Z}/2)\), which makes \(d(V)\) an interesting invariant. The paper contains the computation (or at least estimates) of \(d(V)\) for a large class of complex projective nonsingular varieties \(V\). Here are samples of results. Let \(\text{Alb} (V)\) be the Albanese variety of \(V\) and \(\text{End(Alb} (V))\) its ring of endomorphisms. If \(\dim V = 1\), \(\text{Alb}(V)\) is just the jacobian \(\text{Jac} (V)\) of \(V\). Theorem 1. One has \(d(V) \leq \text{rank End(Alb} (V))\). Let us recall that for a generic abelian variety \(A\) (resp. generic complex projective nonsingular curve \(V)\), one has \(\text{End} (A) \approx \mathbb{Z}\) (resp. End (Jac\((V)) \approx \mathbb{Z})\). It follows that for a generic abelian variety \(A\) (resp. generic projective curve \(V)\), one has \(d(A) \leq 1\) (resp. \(d(V) \leq 1)\). Theorem 2. Let \(V\) be a complex projective irreducible nonsingular curve of positive genus, defined over \(\mathbb{R}\), and let \(V(\mathbb{R})\) be its real part. If \(V_ \mathbb{R} \backslash V (\mathbb{R})\) is disconnected or \(V(\mathbb{R}) = \emptyset\), then \(d(V) \leq \text{rank End(Jac}(VI))-1\). Corollary 3. For \(V\) as in theorem 2, if \(\text{End(Jac} (V)) \cong \mathbb{Z}\), then \(d(V) = 1\) if \(V_ \mathbb{R} \backslash V (\mathbb{R})\) is connected and \(V(\mathbb{R}) \neq \emptyset\), \(d(V) = 0\) otherwise. Theorem 4. For every complex elliptic curve \(E\) without complex multiplication, one has \(d(E)\leq 1\). If, moreover, the \(j\)- invariant \(j(E)\) of \(E\) is in \(\mathbb{R}\), then \(d(E) = 0\) if \(j(E) > 1728\), \(d(E)=1\) if \(j(E) < 1728\). Proposition 5. For \(k = 0, 1\) or 2 let \(B_ k = \{j \in \mathbb{C} \mid j = j(E)\), \(E\) is a complex elliptic curve, \(d(E) = k\}\). Then each of these sets is dense in \(\mathbb{C}\), the sets \(B_ 0\) and \(B_ 1\) are uncountable, and \(B_ 2\) is countable. Theorem 6. Let \(V\) be a complex projective irreducible nonsingular curve and let \(\text{Jac} (V)\) be its jacobian variety. Then \(d(V) = d (\text{Jac} (V))\).