del Pezzo surfaces with one bad prime over cyclotomic ℤ ℓ \mathbb{Z}_{\ell} -extensions Thanks: Nowroozi is supported by the EPSRC studentship. Maryam Nowroozi Abstract . Let K K be a number field and S S a finite set of primes of K K . Scholl proved that there are only finitely many K K -isomorphism classes of del Pezzo surfaces of any degree 1 ≤ d ≤ 9 1\leq d\leq 9 over K K with good reduction away from S S . Let instead K K be the cyclotomic ℤ 5 \mathbb{Z}_{5} -extension of ℚ \mathbb{Q} . In this paper, we show, for d = 3 d=3 , 4 4 , that there are infinitely many ℚ ¯ \overline{\mathbb{Q}} isomorphism classes of del Pezzo surfaces, defined over K K , with good reduction away from the unique prime above 5 5 . 2020 Mathematics Subject Classification 11G35 (14G05). 1. Introduction Let \ell be a rational prime and r a positive integer. Write \mathbb{Q}_{r,\ell} for the unique degree \ell^{r} totally real subfield of \cup_{n=1}^{\infty}\mathbb{Q}(\mu_{n}) , where \mu_{n} denotes the set of \ell^{n} -th roots of 1 . We let \mathbb{Q}_{\infty,\ell}=\cup_{r}\mathbb{Q}_{r,\ell} ; this is called the \mathbb{Z}_{\ell} -cyclotomic extension of \mathbb{Q} , and \mathbb{Q}_{r,\ell} is called the r -th layer of \mathbb{Q}_{\infty,\ell} . Now let K be a number field, and write K_{\infty,\ell}=K\cdot\mathbb{Q}_{\infty,\ell} and K_{r,\ell}=K\cdot\mathbb{Q}_{r,\ell} . To ease notation we shall sometimes write K_{\infty} for K_{\infty,\ell} . Many theorems and conjectures regarding the arithmetic of curves and varieties over number fields have analogues over K_{\infty} . We mention some examples. (1) The Mordell–Weil theorem asserts that for an abelian variety A over a number field K , the Mordell–Weil group A(K) is finitely generated. A celebrated conjecture of Mazur [ 4 ] asserts that A(K_{\infty}) is finitely generated for an abelian variety A over K_{\infty} . (2) Let X be a curve of genus \geq 2 over a number field K . Faltings’ theorem (previously known as the Mordell conjecture) [ 2 ] asserts that X(K) is finite. A conjecture of Parshin and Zarhin [ 10 , page 91] asserts that X(K_{\infty}) is finite, for a curve X of genus \geq 2 over K_{\infty} . (3) A theorem of Zarhin [ 11 , Corollary 4.2] , asserts that the Tate homomorphism conjecture (also a theorem of Faltings [ 2 ] over number fields) continues to hold over K_{\infty} . The Shafarevich conjecture for abelian varieties asserts that for a number field K , a finite set of primes S of K , and a given dimension d\geq 1 , there are only finitely many principally polarized abelian varieties A defined over K of dimension d and good reduction away from S . The Shafarevich conjecture was proved by Faltings [ 2 ] in the same paper in which he proved the Tate homomorphism conjecture and the Mordell conjecture. In a recent paper [ 8 ] , Siksek and Visser show that the Shafarevich conjecture for abelian varieties does not hold over K_{\infty} . For example, they construct infinitely many principally polarized abelian surfaces defined over \mathbb{Q}_{\infty,11} with good reduction away from 2 , 11 , that are pairwise non-isomorphic over \overline{\mathbb{Q}} . In view of the above theorems, conjectures and counterexamples, it is interesting to test whether Shafarevich-type conjectures for other families of varieties continue to hold over K_{\infty} . In 1985 Scholl proved the anaologue of the Shafarevich conjecture for del Pezzo surfaces. More precisely, Scholl [ 6 ] proved that, for any 1\leq d\leq 9 , and for any number field K and for S a finite set of primes of K , then there are only finitely many K -isomorphism classes of del Pezzo surfaces of degree d over K with good reduction away from S . In this paper, we construct infinite families of del Pezzo surfaces over \mathbb{Q}_{\infty,5} with good reduction away from 5 . Theorem 1 . Let r\geq 1 . Then there is a del Pezzo surface X_{r} of degree 3 defined over \mathbb{Q}_{r,5} with good reduction away from the unique prime above 5 . Moreover, within the family \{X_{r}\;:\;r\geq 1\} , there is an infinite subfamily whose members are pairwise non-isomorphic over \overline{\mathbb{Q}} . Theorem 2 . Let r\geq 1 . Then there is a del Pezzo surface Y_{r} of degree 4 defined over \mathbb{Q}_{r,5} with good reduction away from the unique prime above 5 . Moreover, within the family \{Y_{r}\;:\;r\geq 1\} , there is an infinite subfamily whose members are pairwise non-isomorphic over \overline{\mathbb{Q}} . We quickly sketch the construction of the surfaces X_{r} and Y_{r} . Let \zeta=\zeta_{5^{r+1}}=\exp(2\pi i/5^{r+1}) . The extension \mathbb{Q}(\zeta)/\mathbb{Q} is cyclic with Galois group \Gal(\mathbb{Q}(\zeta)/\mathbb{Q})\;\cong\;\left(\mathbb{Z}/5^{r+1}\mathbb{Z}\right)^{\times}\;\cong\;\mathbb{Z}/4\mathbb{Z}\times\mathbb{Z}/5^{r}\mathbb{Z}. Let \tau:\mathbb{Q}(\zeta)\rightarrow\mathbb{Q}(\zeta) be the field automorphism given by \tau(\zeta)=\zeta^{\delta} where \delta\in(\mathbb{Z}/5^{r+1}\mathbb{Z})^{\times} is the element satisfying \delta^{2}\equiv-1\pmod{5^{r+1}},\qquad\delta\equiv 2\pmod{5}; such \delta exists and is unique by Hensel’s lemma. We note that \tau is an automorphism of order 4 . Let H=\langle\tau\rangle be the subgroup of \Gal(\mathbb{Q}(\zeta)/\mathbb{Q}) generated by \tau , and let \mathbb{Q}_{r,5} by the fixed field of H which has degree 5^{r} and Galois group isomorphic to \mathbb{Z}/5^{r}\mathbb{Z} . We note that \mathbb{Q}_{\infty,5}=\bigcup_{r=1}^{\infty}\mathbb{Q}_{r,5} Since 5 is totally ramified in \mathbb{Q}(\zeta) , it is totally ramified in \mathbb{Q}_{r,5} . Let P=(\zeta^{2}:\zeta:1)\in\mathbb{P}^{2}(\mathbb{Q}(\zeta)) . We let X_{r} be the blow-up of \mathbb{P}^{2} in the six points (1) \{\sigma(P):\sigma\in H\}\cup\{(1:0:0),(0:1:0)\}. Let Y_{r} be the blow-up of \mathbb{P}^{2} in the five points (2) \{\sigma(P):\sigma\in H\}\cup\{(1:0:0)\}. Since these sets of points are both stable under the action of H , the surfaces X_{r} and Y_{r} are defined over \mathbb{Q}_{r,5} . Lemma 3 . The six points in ( 1 ) are in general position. Moreover, for any prime ideal \mathfrak{q}\subset\mathcal{O}_{\mathbb{Q}_{r,5}} with \mathfrak{q}\nmid 5 , the reduction of the six points in ( 1 ) modulo \mathfrak{q} are in general position. Lemma 3 is proved in Section 3 . Since the six points in ( 1 ) are in general position, the surface X_{r} is a del Pezzo surface of degree 9-6=3 . Moreover, for a prime \mathfrak{q}\nmid 5 , as the six points remain in general position modulo \mathfrak{q} , it follows [ 6 , Section 3] that X_{r} has good reduction modulo \mathfrak{q} . To complete the proof of Theorem 1 we need to show that among the X_{r} there are infinitely many that are pairwise non-isomorphic over \overline{\mathbb{Q}} . We do this in Section 4 by using the Clebsch–Salmon invariants of the degree 3 del Pezzo surfaces X_{r} . The five points in ( 2 ) are a subset of the six points in ( 1 ) and it follows that they are in general position, and their reduction modulo any \mathfrak{q}\nmid 5 are in general position. We conclude immediately that the Y_{r} are degree 9-5=4 del Pezzo surfaces with good reduction away from the unique prime above 5 . In Section 4 we again use invariants to show that among the Y_{r} there are infinitely many that are pairwise non-isomorphic over \overline{\mathbb{Q}} . All the computations are done in Magma [ 1 ] and SageMath [ 9 ] . You can find the code here: https://github.com/maryam-nowroozi/del-pezzo-surfaces 2. background In this section, we review some of the definitions we will need in further sections. Let k be a field. Definition 3 . Let 1\leq r\leq 8 be an integer and let P_{1},\ldots,P_{r} be r distinct points in \mathbb{P}^{2}_{k} . We say that P_{1},\ldots,P_{r} are in general position if the following hold: (1) no three of the points lie on a line; (2) no six of the points lie on a conic; (3) no eight of the points lie on a singular cubic, with one of these eight points at the singular point. Definition 4 . Let k be a field. A surface X over k is a del Pezzo surface if its anticanonical divisor -K_{X} is ample. Note that if -K_{X} is very ample, then it determines an embedding of X into \mathbb{P}^{n} for some n , under which -K_{X} corresponds to a hyperplane section H . The degree of this embedding is H^{2} , which explains why the integer (-K_{X})^{2} is called the degree of a del Pezzo surface X . Example 4 . Let X=Q_{1}\cap Q_{2} be the intersection of two non-singular quadric hypersurfaces in \mathbb{P}^{4}_{k} . Then X is a Del Pezzo surface of degree 4 (see [ 5 ] , Example 4.5). Example 5 . Let X be a cubic surface over k , i.e. a non-singular projective variety of dimension 2 and degree 3 , then X is a Del Pezzo surface of degree 3 (see [ 5 ] , Example 4.6). Proposition 6 . Let k be a field. If X is a blow-up of \mathbb{P}^{2}_{k} in r\leq 8 k -rational points in general position, then X is a del Pezzo surface of degree 9-r over k . Proof. See [ 5 ] , Proposition 4.34. ∎ Remark 7 . If k is an algebraically closed field, then a del Pezzo surface X is either isomorphic to \mathbb{P}^{1}\times\mathbb{P}^{1} or X arises as a blowing up of \mathbb{P}^{2}_{k} in r\leq 8 points in general position and K_{X}^{2}=9-r (see [ 5 ] , Theorem 4.22). In general, this does not hold over an arbitrary field. 3. Proof of Lemma 3 The goal of this section is to prove Lemma 3 . Let k be a field and let u , v\in k^{\times} . Let T_{u,v}\;=\;\{(u^{2}:u:1),~(u^{-2}:u^{-1},1),~(v^{2}:v:1),~(v^{-2}:v^{-1}:1),~(1:0:0),~(0:1:0)\}. We note that the set of six points in ( 1 ) is simply T_{\zeta,\zeta^{\delta}} . Lemma 8 . Let f(u,v)=(u^{4}-1)(v^{4}-1)(u^{2}-v^{2})(u^{2}v^{2}-1). The set of six points T_{u,v} is in general position if and only if f(u,v)\neq 0 . Proof. Let M_{u,v} be the 6\times 6 determinant with rows (x^{2},y^{2},z^{2},xy,xz,yz) with (x,y,z) running through the elements of T_{u,v} . We see that the six points of T_{u,v} lie on a conic if and if this determinant is 0 . A quick Magma computation shows that M_{u,v}=(u+1)(u-1)(v+1)(v-1)(u-v)^{2}(uv-1)^{2}/u^{3}v^{3}. Moreover, three points (x_{i}:y_{i}:z_{i}) , with i=1,2,3 , lie on a line if and only if the 3\times 3 determinant with rows (x_{i},y_{i},z_{i}) is zero. We computed this determinant for the \binom{6}{3}=20 possible choices of triples of points from among six points of T_{u,v} . All these determinants have numerators with factors among u\pm 1 , u^{2}+1 , v\pm 1 , v^{2}+1 , u\pm v , uv\pm 1 . The lemma follows. ∎ Proof of Lemma 3 Let r\geq 1 and \zeta=\exp(2\pi i/5^{r+1}) . Write u=\zeta , v=\zeta^{\delta} . Then the six points in ( 1 ) form the set T_{u,v} . We consider the elements u^{4}-1,v^{4}-1,u^{2}-v^{2},u^{2}v^{2}-1. By Lemma 8 , the six points are in general position provided these quantities are non-zero. All of these, up to multiplication by a root of unity, have the form \zeta^{\mu}-1 , where \mu is among 4 , 4\delta , 2\delta-2 , 2\delta+2 . Since \delta\equiv 2\pmod{5} , we see that \mu\equiv 4 , 3 , 2 , 1\pmod{5} respectively. Thus in all cases \zeta^{\mu} is a primitive 5^{r+1} -th root of unity, and it follows that the six points in ( 1 ) are in general position. Next we consider the six points modulo a prime ideal \mathfrak{p} of \mathcal{O}_{\mathbb{Q}_{r,5}} . We note that, for the above four values of \mu , that the element \zeta^{\mu}-1 generates the unique prime ideal of \mathcal{O}_{\mathbb{Q}(\zeta)} above 5 . Let \mathfrak{p} be a prime ideal of \mathcal{O}_{\mathbb{Q}_{r,5}} satisfying \mathfrak{p}\nmid 5 . Let k=\mathcal{O}_{\mathbb{Q}_{r,5}}/\mathfrak{p} . Then the reduction of \zeta^{\mu}-1 in \overline{k} is non-zero. Thus the reduction of six points in ( 1 ) in \mathbb{P}^{2}(\overline{k}) are in general position. This completes the proof. 4. Isomorphism of del Pezzo surfaces 4.1. Degree 3 The goal of this section is to prove that there exists an infinite subfamily of \{X_{r}:r\geq 1\} whose members are pairwise non-isomorphic. We will confirm this by using the Clebsch-Salmon invariants for cubic surfaces. We now recall the definition of Clebsch-Salmon invariants. Let X be a cubic surface given by a system of equations of the form \displaystyle a_{0}x_{0}^{3}+a_{1}x_{1}^{3}+a_{2}x_{2}^{3}+a_{3}x_{3}^{3}+a_{4}x_{4}^{3}=0 \displaystyle x_{0}+x_{1}+x_{2}+x_{3}+x_{4}=0 which is also called its pentahedral form . The coefficients a_{0},a_{1},a_{2},a_{3},a_{4} are called pentahedral coefficients of the surface. Definition 5 (Clebsch-Salmon invariants) . Let \sigma_{1},\ldots,\sigma_{5} be the elementary symmetric functions. We have the following invariants of degrees 8,16,24,32, and 40, respectively \displaystyle I_{8}=\sigma_{4}^{2}-4\sigma_{3}\sigma_{5}, \displaystyle I_{16}=\sigma_{1}\sigma_{5}^{3}, \displaystyle I_{24}=\sigma_{4}\sigma_{5}^{4}, \displaystyle I_{32}=\sigma_{2}\sigma_{5}^{6}, \displaystyle I_{40}=\sigma_{5}^{8}. Let X_{u,v} be the blow-up of \mathbb{P}^{2} in T_{u,v} and let [I_{8}(X_{u,v}):I_{16}(X_{u,v}):I_{24}(X_{u,v}):I_{32}(X_{u,v}):I_{40}(X_{u,v})]\in\mathbb{P}^{8,16,24,32,40} be the invariants of X_{u,v} . The following lemma proves that there are at most finitely many del Pezzo surfaces isomorphic over \overline{\mathbb{Q}} to X_{u,v} . Lemma 9 . Let (u,v)\in\overline{\mathbb{Q}} , f(u,v)\neq 0 . Then, there are at most finitely many pairs (r,s)\in\overline{\mathbb{Q}} , f(r,s)\neq 0 such that X_{r,s} is isomorphic over \overline{\mathbb{Q}} to X_{u,v} Proof. Suppose X_{r,s}\cong_{\overline{\mathbb{Q}}}X_{u,v} . Then \frac{I_{16}(X_{u,v})}{I_{8}^{2}(X_{u,v})}=\frac{I_{16}(X_{r,s})}{I_{8}^{2}(X_{r,s})}\ \ \text{and}\ \ \frac{I_{24}(X_{u,v})}{I_{8}^{3}(X_{u,v})}=\frac{I_{24}(X_{r,s})}{I_{8}^{3}(X_{r,s})} Let \displaystyle F(r,s)=I_{16}(X_{u,v})-\alpha I_{8}^{2}(X_{u,v})\in\mathbb{Q}(\alpha)[r,s] \displaystyle G(r,s)=I_{24}(X_{u,v})-\beta I_{8}^{3}(X_{u,v})\in\mathbb{Q}(\beta)[r,s] Then, F(r,s)=G(r,s)=0 . Let h_{1}(r)=\operatorname{Res}_{s}(F(r,s),G(r,s)) and h_{2}(s)=\operatorname{Res}_{r}(F(r,s),G(r,s)) . Then h_{1}(r)=h_{2}(s)=0 . So, the number of possibilities for the pair (r,s) is at most \deg(h_{1})\cdot\deg(h_{2}) . A quick Magma computation shows that \deg(h_{1})\cdot\deg(h_{2})=518400 . ∎ This proves that within the family \{X_{r}\;:\;r\geq 1\} , there is an infinite subfamily whose members are pairwise non-isomorphic over \overline{\mathbb{Q}} . 4.2. degree 4 Recall that a Del Pezzo surface of degree 4 is a complete intersection of two quadrics in \mathbb{P}^{4} . Let X\subset\mathbb{P}^{4} be a smooth surface defined by the intersection of two quadrics over an algebraically closed field k of characteristic different from 2. We know that X is charactrized upto isomorphism by the degeneracy locus of the pencil of quadrics containing X , i.e., by the form h(t,w)=\det(tA_{F}+wA_{G}), where A_{F} and A_{G} are symmetric 5\times 5 matrices whose associated quadric forms define X (see [ 3 ] for more details). The two-dimensional space of binary quintic forms with non-vanishing discriminant up to linear change of variables serves as a moduli space of smooth del Pezzo surface of degree 4. The Del Pezzo surface defined by blowing up the five points \displaystyle P_{1}=(u^{2}:u:1), \displaystyle P_{2}=(u^{-2}:u^{-1}:1), \displaystyle P_{3}=(v^{2}:v:1), \displaystyle P_{4}=(v^{-2}:v^{-1}:1), \displaystyle P_{5}=(1:0:0). is a complete intersection of two quadrics in \mathbb{P}^{4} given below \displaystyle F(x_{0},x_{1},x_{2},x_{3},x_{4})=-x_{1}x_{2}+x_{2}^{2}+x_{0}x_{3}+ax_{2}x_{3}+bx_{3}^{2}+ax_{3}x_{4}+x_{4}^{2}, \displaystyle G(x_{0},x_{1},x_{2},x_{3},x_{4})=-x_{1}x_{3}-ax_{3}^{2}+x_{0}x_{4}+ax_{2}x_{4}. where a=\frac{-u^{2}v-uv^{2}-u-v}{uv} and b=\frac{u^{2}v^{2}+u^{2}+2uv+v^{2}+1}{uv} . The defined matrices A_{F} and A_{G} are as follows: A_{F}=\begin{bmatrix}0&0&0&1/2&0\\ 0&0&-1/2&0&0\\ 0&-1/2&1&a/2&0\\ 1/2&0&a/2&b&a/2\\ 0&0&0&a/2&1\end{bmatrix},\ A_{G}=\begin{bmatrix}0&0&0&0&1/2\\ 0&0&0&-1/2&0\\ 0&0&0&0&a/2\\ 0&-1/2&0&-a&0\\ 1/2&0&a/2&0&0\end{bmatrix} We have h(t,w)=\det(tA_{F}+wA_{G})=\begin{vmatrix}0&0&0&\frac{1}{2}t&\frac{1}{2}w\\ 0&0&-\frac{1}{2}t&-\frac{1}{2}w&0\\ 0&-\frac{1}{2}t&t&\frac{a}{2}t&\frac{a}{2}w\\ \frac{1}{2}t&-\frac{1}{2}w&\frac{a}{2}t&-aw+bt&\frac{a}{2}t\\ \frac{1}{2}w&0&\frac{a}{2}w&\frac{a}{2}t&t\end{vmatrix} So, we get h(t,w)=\frac{1}{16}(t^{5}-at^{4}w+bt^{3}w^{2}-at^{2}w^{3}+tw^{4}) which is a binary quintic. The ring SL_{2} -invariants of binary quintics is generated by four invariants, I_{4},I_{8},I_{12} , and, I_{18} (see [ 7 ] ). Let Y_{u,v} be the blow-up of \mathbb{P}^{2} in five points P_{1},\ldots,P_{5} and let [I_{4}:I_{8}:I_{12}:I_{18}]\in\mathbb{P}^{4,8,12,18} be the invariants of the degeneracy locus of the pencil of quadrics containing Y_{u,v} , defined above. The following lemma proves that there are at most finitely many del Pezzo surfaces isomorphic over \overline{\mathbb{Q}} to Y_{u,v} . Lemma 10 . Let (u,v)\in\overline{\mathbb{Q}} , f(u,v)\neq 0 . Then, there are at most finitely many pairs (r,s)\in\overline{\mathbb{Q}} , f(r,s)\neq 0 such that Y_{r,s} is isomorphic over \overline{\mathbb{Q}} to Y_{u,v} Proof. Suppose Y_{r,s}\cong_{\overline{\mathbb{Q}}}Y_{u,v} . Then \frac{I_{8}(Y_{u,v})}{I_{4}^{2}(Y_{u,v})}=\frac{I_{8}(Y_{r,s})}{I_{4}^{2}(Y_{r,s})}\ \ \text{and}\ \ \frac{I_{12}(Y_{u,v})}{I_{4}^{3}(Y_{u,v})}=\frac{I_{12}(Y_{r,s})}{I_{4}^{3}(Y_{r,s})} Let \displaystyle F(r,s)=I_{8}(Y_{u,v})-\alpha I_{4}^{2}(Y_{u,v})\in\mathbb{Q}(\alpha)[r,s] \displaystyle G(r,s)=I_{12}(Y_{u,v})-\beta I_{4}^{3}(Y_{u,v})\in\mathbb{Q}(\beta)[r,s] Then, F(r,s)=G(r,s)=0 . Let h_{1}(r)=\operatorname{Res}_{s}(F(r,s),G(r,s)) and h_{2}(s)=\operatorname{Res}_{r}(F(r,s),G(r,s)) . Then h_{1}(r)=h_{2}(s)=0 . So, the number of possibilities for the pair (r,s) is at most \deg(h_{1})\cdot\deg(h_{2}) . A quick Magma computation shows that \deg(h_{1})\cdot\deg(h_{2})=518400 . ∎ This proves that within the family \{Y_{r}\;:\;r\geq 1\} , there is an infinite subfamily whose members are pairwise non-isomorphic over \overline{\mathbb{Q}} .