A few computations about the real cycle class map in low dimensions (Q6917971)

Let \(X\) be a smooth real algebraic variety of dimension \(d\). The Borel-Haefliger real cycle class map \(\mathfrak{c}_\mathrm{BH}\colon\mathrm{CH}^k(X)/2\to H^k(X(\mathbb{R});\mathbb{Z}/2)\) admits an integral real lift \(\mathfrak{c}_\mathbb{R}\colon H^k(X,I^k)\to H^k(X(\mathbb{R});\mathbb{Z})\), and the surjectivity of this lift can be regarded as one of the real counterparts of the Hodge conjecture. The map is known to be surjective for \(k=d\), and the author attempts a few first computations for \(k=d-1\). By a simple example (a typical elliptic curve with two ovals) it is shown that \(\mathfrak{c}_\mathbb{R}^{d-1}\) may fail to be surjective even for curves. For surfaces, the results obtained are as follows: if \(X\) is geometrically ruled or Enriques, then \(\mathfrak{c}_\mathbb{R}^1\) is surjective; if \(X\) is a \(K3\)-surface, then \(\mathfrak{c}_\mathbb{R}^1\) is surjective if and only if so is \(\mathfrak{c}_\mathrm{BH}^1\); if \(X\) is an abelian variety, then \(\mathfrak{c}_\mathbb{R}^1\) is surjective if and only if \(X(\mathbb{R})\) is connected.