Explicit estimates for the number of rational points of singular complete intersections over a finite field (Q495273)

Assume \(\mathbb{F}_{q}\) is a finite field of \(q\) elements and \(V \subset \mathbb{P}^{n}\) is a complete intersection of dimension \(r\) and of multidegree \(\mathbf{d} = (d_{1} , \cdots , d_{n-r})\) defined over \(\mathbb{F}_{q}\). Also assume the singular locus of \(V\) is of dimension at most \(s\) where \(s \leq r-2\). Let \(\delta = \prod d_{i}\) and \(D = \sum (d_{i}-1)\). The main result of the paper under review is the following estimation of the number of \(\mathbb{F}_{q}\) rational points of \(V\). Theorem 4.5. If \(q>2(s+1)D^{r-s-1}(D+r-s)\delta\), \[ ||V(\mathbb{F}_{q}) - p_{r}| \leq (b'_{r-s-1}(n-s-1,\mathbf{d}) + 2\sqrt{\delta} +1) q ^{r+s+1}. \] Here \(p_{r} = |\mathbb{P} ^{r}(\mathbb{F}_{q})| = (q^{r+1}-1)/(q-1)\) and \(b'_{s}(m,\mathbf{d})\) is the \(s\)-th Betti number of a non-singular complete intersection of dimension \(s\) and of multidegree \(\mathbf{d}\) in \(\mathbb{P} ^{m}\). One advantage of this result to the previously known result of \textit{S. R. Ghorpade} and \textit{G. Lachaud} [Mosc. Math. J. 2, No. 3, 589--631 (2002; Zbl 1101.14017)], is that this formula does not contain an exponential term of the dimension of the ambient projective space \(n\). The key step of the proof is a suitable version of Bertini theorem. (Theorem 3.5)