MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Suppose $X\subset \mathbb{P}^n$ is a smooth hypersurface defined over $\mathbb{Q}$. For a "generic" prime $p$, what can be said about the set of hyperplanes $H$ in $\mathbb{P}^n(\mathbb{F}_p)$ for which $H \cap X$ is smooth over $\mathbb{F}_p$? For $p$ fixed and $X$ varying, by contrast, the situation can be arbitrarily bad: in fact, every hyperplane section of

${\sum_{i=1}^{n+1}X_i X_{i+n+1}^p=0} \subset \mathbb{P}^{2n+1}$ over $\mathbb{F}_p$ is singular.

share|cite|improve this question
up vote 9 down vote accepted

Spread out $X$ over some $R=\mathbf{Z}[1/n]$ to a hypersurface $\mathcal{X} \subseteq \mathbf{P}^n_R$ that is smooth and projective over $R$. The standard proof of the Bertini smoothness theorem (as given in Hartshorne, Algebraic geometry, for instance) works over $R$: there is a Zariski dense open subscheme $U$ of the dual projective space $\mathbf{P}^n_R$ such that for $p \nmid n$, the hyperplanes $H$ in $\mathbf{P}^n_{\mathbf{F}_p}$ such that $H \cap \mathcal{X}_p$ is smooth are exactly those corresponding to $\mathbf{F}_p$-points of $U$. The complement of $U$ has at most $O(p^{n-1})$ points over $\mathbf{F}_p$ as $p \to \infty$, but $\#\mathbf{P}^n(\mathbf{F}_p)= p^n+p^{n-1}+\cdots+1$, so when $p$ is large enough, most hyperplanes over $\mathbf{F}_p$ will intersect the fiber $\mathcal{X}_p$ in something smooth.

share|cite|improve this answer
Thanks, Bjorn! Are there situations where the compliment of U has $O(p^{n-i})$ points for $i>1$? – David Hansen Feb 28 '10 at 4:20
Yes, for instance if X itself is a hyperplane. But for most X, one cannot improve the exponent in O(p^{n-1}). It depends on the dimension of the generic fiber of Z:=P^n-U, since all but finitely many fibers above primes will have the same dimension. Usually this dimension will be n-1. – Bjorn Poonen Feb 28 '10 at 5:17

By Robinson's theorem in model theory, the reduction $X_p$ is smooth over $\bar{F_p}$ for almost all $p$. By Bertini's theorem, $X_p \cap H$ is thus smooth for H ranging over a dense subset of the hyperplane sections possible in $P^n(\bar{F_p})$.

I think the statement of nonsingularity can be stated within first-order logic. Nonsingularity is local, so assume $X$ affine. Let the ideal of $X$ be generated by $f_1, \dots, f_{n-d}$. Then nonsingularity means that each point where $f_1, \dots, f_{n-d}$ vanish, a certain Jacobian determinant is nonzero.

share|cite|improve this answer
Everything you wrote is correct, and in particular your remark about the good H's forming a dense subset of the dual P^n is key. The only reason I added an answer at all was to word it to say something about F_p-points and not just F_p-bar points, and to make it uniform as p varies. – Bjorn Poonen Feb 28 '10 at 5:51
Good to know! I wasn't all that sure. – Akhil Mathew Feb 28 '10 at 13:15

You mey be interested in the paper: "Bertini Theorems over Finite Fields"(2002) Bjorn Poonen.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.