betaFeedback

WorksMathematics

Joint quadratic base changes of elliptic surfaces with torsion Z/4Z and Z/2Z x Z/6Z

Abstract

We study elliptic curves over Q(t)\mathbb{Q}(t) obtained from an elliptic surface by imposing two quadratic conditions, each of which produces a new point. By the Kani-Rosen decomposition, the joint cover is rational only when the branch loci of the two conditions meet; for two quadratic conditions this means that they share a root. We apply this to two surfaces. For Elkies's rank-four K3 surface with torsion Z/4Z\mathbb{Z}/4\mathbb{Z}, a catalogue of 194194 quadratic conditions contains exactly two such pairs. One recovers the rank-six curve of the authors; the other gives a second, inequivalent curve with torsion Z/4Z\mathbb{Z}/4\mathbb{Z} and rank six. For the universal curve with torsion Z/2Z×Z/6Z\mathbb{Z}/2\mathbb{Z}\times\mathbb{Z}/6\mathbb{Z}, 4242 conditions give exactly 2121 rational joint covers, which yield seven curves of rank two. Two more arise from a tower of conditions. Five of these nine curves are the curves of Dujella, Kazalicki and Peral; the other four are not equivalent to any of them. In both cases we determine the Mordell-Weil groups over Q(t)\mathbb{Q}(t) exactly, and we prove that the curves are pairwise inequivalent under invertible changes of parameter, even over Q‾\overline{\mathbb{Q}}.

Introduction

By the Mordell-Weil theorem, E(Q)≅E(Q)tors×ZrE(\mathbb{Q}) \cong E(\mathbb{Q})_{\mathrm{tors}} \times\mathbb{Z}^{r} for an elliptic curve E/QE/\mathbb{Q}, and by Mazur’s theorem [20] the torsion subgroup is one of fifteen groups. For each of them one asks for curves of large rank, over Q\mathbb{Q} and over Q(t)\mathbb{Q}(t); the current records are collected in Dujella’s tables [3, 4].

A standard way to raise the rank over Q(t)\mathbb{Q}(t) is quadratic base change. Suppose a surface E→Pt1E \to\mathbb{P}^{1}_{t} acquires a new point over Q(t)(Q)\mathbb{Q}(t)(\sqrt{Q}) for a quadratic polynomial QQ. If the new point has infinite order, the rank over Q(t)(Q)\mathbb{Q}(t)(\sqrt{Q}) is at least one larger than over Q(t)\mathbb{Q}(t). If moreover the conic w2=Q(t)w^{2}=Q(t) has a rational point, then Q(t)(Q)\mathbb{Q}(t)(\sqrt{Q}) is again a rational function field. Imposing two such conditions Q1,Q2Q_{1},Q_{2} raises the rank by at least two over Q(t)(Q1,Q2)\mathbb{Q}(t)(\sqrt{Q_{1}},\sqrt{Q_{2}}). To obtain a curve over a rational function field again, the curve with this function field must be rational.

Section 2 develops this systematically:

  • a genus criterion for joint covers (Lemma 2.1);

  • the character decomposition of the rank;

  • exact ranks through injective specialization, via the criteria of Gusić and Tadić [11];

  • two invariants of the jj-map that detect inequivalence of families under invertible changes of parameter.

We then apply it to two classical situations.

Torsion Z/4Z\mathbb{Z}/4\mathbb{Z}

Torsion Z/4Z\mathbb{Z}/4\mathbb{Z}. Curves of positive rank with this torsion were constructed by Kihara [15] and by Khoshnam and Moody [14]; see also [5]. Elkies [10] gave a K3 surface with torsion Z/4Z\mathbb{Z}/4\mathbb{Z} and rank four over Q(t)\mathbb{Q}(t), and in [8] we made a number of its quadratic sections explicit. In [9] we combined two quadratic sections of it to obtain a curve of rank six, the current record for this torsion group over Q(t)\mathbb{Q}(t) to our knowledge [4], and proved by an injective specialization that its rank is exactly six.

Theorem A. Let E1E_1, E2E_2 be the curves over Q(q)\mathbb{Q}(q) of § 3.

  1. Ei(Q(q))≅Z/4Z×Z6E_i(\mathbb{Q}(q)) \cong\mathbb{Z}/4\mathbb{Z} \times\mathbb{Z}^{6} for i=1,2i = 1, 2.

  2. The specializations of E1E_1 at q=11/7q = 11/7 and of E2E_2 at q=−2q = -2 are injective, and these curves over Q\mathbb{Q} have rank six.

  3. E1E_1 and E2E_2 are not equivalent (Definition 2.3), even over C\mathbb{C}.

  4. In a catalogue of 194 quadratic conditions on Elkies’s surface, exactly two pairs have rational joint covers: the pair used for E1E_1 (which is the curve of [9]) and the pair used for E2E_2.

For E1E_1, parts (1) and (2) were proved in [9]. We include the short proof because it applies verbatim to E2E_2, which is new.

Torsion Z/2Z×Z/6Z\mathbb{Z}/2\mathbb{Z} \times\mathbb{Z}/6\mathbb{Z}

Here the universal curve has rank zero. Curves of rank one over Q(t)\mathbb{Q}(t) were found by Kulesz [17], Lecacheux [18], Dujella [2] and Woo [25], among others. Rank two was first reached by Dujella and Peral [7]. Dujella, Kazalicki and Peral [6] described five curves of rank exactly two and observed that the examples of MacLeod [19] are isomorphic to some of them.

Theorem B. Let U1,…,U9U_1, \ldots, U_9 be the curves over Q(t)\mathbb{Q}(t) of Section 4.

  1. Ui(Q(t))≅Z/2Z×Z/6Z×Z2U_i(\mathbb{Q}(t)) \cong\mathbb{Z}/2\mathbb{Z} \times\mathbb{Z}/6\mathbb{Z} \times\mathbb{Z}^{2} for every ii.

  2. No two of them are equivalent (Definition 2.3), even over Q‾\overline{\mathbb{Q}}.

  3. U1U_1, U2U_2, U5U_5, U8U_8, U6U_6 are the five curves of [6 §8], and U1U_1, U5U_5, U8U_8 are the three curves of [7]. The curves U3U_3, U4U_4, U7U_7, U9U_9 are not equivalent to any of the curves of [7] or [6], nor therefore, by [6 §1.3], to the examples of MacLeod [19].

  4. In a catalogue of 42 quadratic conditions on the universal curve, exactly 21 pairs have rational joint covers, and they give exactly the classes U1,…,U7U_1, \ldots, U_7.

The paper ends with two remarks on the Z/4Z\mathbb{Z}/4\mathbb{Z} case (§ 3.6). First, an exhaustive search shows that neither E1E_1 nor E2E_2 has a further quadratic section whose xx-coordinate is a constant times a divisor of the last Weierstrass coefficient (Proposition 3.3). Second, we exhibit a specialization of E2E_2 whose rank is seven, provided the 2-primary part of its Tate-Shafarevich group is finite.

The method

Quadratic sections

Let E:y2=x3+a(t)x2+b(t)xE : y^2 = x^3 + a(t)x^2 + b(t)x be an elliptic curve over K=Q(t)K = \mathbb{Q}(t), and let Q∈Q[t]Q \in\mathbb{Q}[t] be squarefree and not a constant times a square. A quadratic section attached to QQ is an xQ∈Kx_Q \in K with

xQ3+axQ2+bxQ∈Q⋅K×2.x_Q^3 + a x_Q^2 + b x_Q \in Q \cdot K^{\times2}.

It gives a point PQP_Q of EE over K(Q)K(\sqrt{Q}) that the nontrivial automorphism sends to −PQ-P_Q, or equivalently a point of the quadratic twist E(Q)E^{(Q)} over KK. In particular, if PQP_Q has infinite order it is independent of E(K)E(K).

The genus of a joint cover

For quadratic sections Q1,…,QmQ_1, \ldots, Q_m let CQ1,…,QmC_{Q_1,\ldots,Q_m} be the curve with function field K(Q1,…,Qm)K(\sqrt{Q_1},\ldots,\sqrt{Q_m}). For a squarefree polynomial QQ of degree 1 or 2, let S(Q)⊂P1(Q‾)S(Q) \subset\mathbb{P}^1(\overline{\mathbb{Q}}) be the branch locus of w2=Q(t)w^2 = Q(t):

S(Q)={{the two roots of Q},deg⁡Q=2,{the root of Q,∞},deg⁡Q=1.S(Q) = \begin{cases} \{\text{the two roots of } Q\}, & \deg Q = 2,\\ \{\text{the root of } Q, \infty\}, & \deg Q = 1. \end{cases}

Lemma 2.1. Let Q1,…,Qm∈Q[t]Q_1,\ldots,Q_m \in\mathbb{Q}[t] be squarefree of degree 1 or 2, with classes independent in Q(t)×/Q(t)×2\mathbb{Q}(t)^\times/\mathbb{Q}(t)^{\times2}. Then C=CQ1,…,QmC=C_{Q_1,\ldots,Q_m} is geometrically integral of genus

g(C)=2m−2∣S(Q1)∪⋯∪S(Qm)∣−(2m−1).g(C)=2^{m-2}\lvert S(Q_1)\cup\cdots\cup S(Q_m)\rvert-(2^m-1).

In particular:

  1. for m=2m=2, g(C)=0g(C)=0 if and only if S(Q1)∩S(Q2)≠∅S(Q_1)\cap S(Q_2)\ne\varnothing; for two quadratics this means a common root;

  2. for m=3m=3, g(C)=2∣S(Q1)∪S(Q2)∪S(Q3)∣−7g(C)=2\lvert S(Q_1)\cup S(Q_2)\cup S(Q_3)\rvert-7.

Proof. The cover C→P1C\to\mathbb{P}^1 is Galois with group (Z/2)m(\mathbb{Z}/2)^m, and CC is geometrically integral by the independence assumption. By the Kani-Rosen decomposition [13], J(C)J(C) is isogenous to the product of the Jacobians of the quotients by the subgroups of index two. These are the double covers CI:w2=∏i∈IQiC_I:w^2=\prod_{i\in I}Q_i for ∅≠I⊆{1,…,m}\varnothing\ne I\subseteq\{1,\ldots,m\}. The branch locus of CIC_I is the symmetric difference of the sets S(Qi)S(Q_i), i∈Ii\in I, so g(CI)=12∣△i∈IS(Qi)∣−1g(C_I)=\frac{1}{2}\lvert\triangle_{i\in I}S(Q_i)\rvert-1. A point lying in at least one S(Qi)S(Q_i) belongs to the symmetric difference for exactly 2m−12^{m-1} of the sets II, namely those meeting the set of ii with that point in S(Qi)S(Q_i) in an odd number of elements. Summing over II gives the formula. For m=2m=2 it reads g=∣S(Q1)∪S(Q2)∣−3g=\lvert S(Q_1)\cup S(Q_2)\rvert-3, and the union has 3 or 4 elements. □

Two irreducible quadratics with a common root are proportional. So for families of pairwise non-proportional conditions, rational joint covers come only from split quadratics with a common root, or from conditions involving linear polynomials.

The character decomposition

Let L=K(Q1,Q2)L=K(\sqrt{Q_1},\sqrt{Q_2}) with Q1,Q2Q_1,Q_2 independent in K×/K×2K^\times/K^{\times2}, and put Q3=Q1Q2Q_3=Q_1Q_2. The isotypic decomposition of E(L)⊗QE(L)\otimes\mathbb{Q} under Gal⁡(L/K)≅(Z/2)2\operatorname{Gal}(L/K)\cong(\mathbb{Z}/2)^2 gives

rank⁡E(L)=rank⁡E(K)+rank⁡E(Q1)(K)+rank⁡E(Q2)(K)+rank⁡E(Q3)(K).(1)\operatorname{rank}E(L)=\operatorname{rank}E(K)+\operatorname{rank}E^{(Q_1)}(K)+\operatorname{rank}E^{(Q_2)}(K)+\operatorname{rank}E^{(Q_3)}(K). \tag*{(1)}

If CQ1,Q2C_{Q_1,Q_2} is a conic with a rational point, then L≅Q(s)L\cong\mathbb{Q}(s). Two quadratic sections of infinite order then raise the rank over Q(s)\mathbb{Q}(s) by at least two. When the rank over Q(s)\mathbb{Q}(s) is known exactly, (1) also determines the ranks of the three twists.

Exact rank by injective specialization

We use two criteria of Gusić and Tadić.

Theorem 2.2 ([11 Thms. 1.1 and 1.3]). Let E/Q(t)E/\mathbb{Q}(t) be nonconstant and t0∈Qt_0\in\mathbb{Q}.

  1. If E:y2=(x−e1)(x−e2)(x−e3)E:y^2=(x-e_1)(x-e_2)(x-e_3) with ei∈Z[t]e_i\in\mathbb{Z}[t], and h(t0)h(t_0) is not a rational square for every nonconstant squarefree divisor h∈Z[t]h\in\mathbb{Z}[t] of (e1−e2)(e1−e3)(e_1-e_2)(e_1-e_3), of (e2−e1)(e2−e3)(e_2-e_1)(e_2-e_3), or of (e3−e1)(e3−e2)(e_3-e_1)(e_3-e_2), then the specialization at t0t_0 is injective.

  2. If E:y2=x3+Ax2+BxE:y^2=x^3+Ax^2+Bx with A,B∈Z[t]A,B\in\mathbb{Z}[t] has exactly one nontrivial rational 2-torsion point, and h(t0)h(t_0) is not a rational square for every nonconstant squarefree divisor h∈Z[t]h\in\mathbb{Z}[t] of BB or of A2−4BA^2-4B, then the specialization at t0t_0 is injective.

In all our applications we tested every squarefree product of the nonconstant irreducible factors, multiplied by every signed squarefree divisor of the integer contents. This is a superset of the divisors required. An injective specialization gives rank⁡E(Q(t))≤rank⁡Et0(Q)\operatorname{rank}E(\mathbb{Q}(t))\leq\operatorname{rank}E_{t_0}(\mathbb{Q}), and the torsion of E(Q(t))E(\mathbb{Q}(t)) injects into that of the fibre.

Invariants of the jj-map

Definition 2.3. Two elliptic curves E,E′E,E' over k(t)k(t), for a field k⊇Qk\supseteq\mathbb{Q}, are equivalent over kk if there is an invertible Möbius transformation μ\mu of the tt-line, defined over kk, such that EE and μ∗E′\mu^*E' are isomorphic over k(t)k(t).

Let φ1,φ2:P1→P1\varphi_{1},\varphi_{2}:\mathbb{P}^{1}\to\mathbb{P}^{1} be base changes of an elliptic surface with jj-invariant jj. If φ1∗E\varphi_{1}^{*}E and μ∗φ2∗E\mu^{*}\varphi_{2}^{*}E are isomorphic over C(t)\mathbb{C}(t) for some Möbius transformation μ\mu, then j∘φ1=j∘φ2∘μj\circ\varphi_{1}=j\circ\varphi_{2}\circ\mu. Quadratic twists are included, since they do not change jj. We use two consequences:

  • (i) j∘φ1j\circ\varphi_{1} and j∘φ2j\circ\varphi_{2} have the same critical values;

  • (ii) for each mm, μ\mu carries the poles of order mm of j∘φ1j\circ\varphi_{1} to those of j∘φ2j\circ\varphi_{2}.

For (ii), let F(T,Z)F(T,Z) be a binary form of degree 12 whose roots are these poles. With the transvectants [22]

(F,G)r=∑i=0r(−1)i(ri)∂rF∂Tr−i∂Zi∂rG∂Ti∂Zr−i,(F,G)_{r}=\sum_{i=0}^{r}(-1)^{i}\binom{r}{i}\frac{\partial^{r}F}{\partial T^{r-i}\partial Z^{i}}\frac{\partial^{r}G}{\partial T^{i}\partial Z^{r-i}},

put I2=(F,F)12I_{2}=(F,F)_{12}, H=(F,F)10H=(F,F)_{10}, I4=(H,H)4I_{4}=(H,H)_{4} and, whenever I2≠0I_{2}\ne0,

ι(F)=I4I22(2)\iota(F)=\frac{I_{4}}{I_{2}^{2}} \tag*{(2)}

This quantity is unchanged by scaling FF and by GL⁡2(C)\operatorname{GL}_{2}(\mathbb{C}) acting on (T,Z)(T,Z), so equivalent base changes have equal values of ι\iota. In all cases below, I2≠0I_{2}\ne0.

Torsion Z/4Z\mathbb{Z}/4\mathbb{Z}: two rank-six curves

Elkies's surface

Put

a(t)=(8t−1)(32t+7),g(t)=(t+1)(15t−8)(31t−7),a(t)=(8t-1)(32t+7),\qquad g(t)=(t+1)(15t-8)(31t-7),
A(t)=a2−64g=65536t4−17472t3−10176t2+18672t−3535,B(t)=1024g2,A(t)=a^{2}-64g=65536t^{4}-17472t^{3}-10176t^{2}+18672t-3535,\qquad B(t)=1024g^{2},

and, following [9], write Elkies’s surface [10] as

E:y2=x3+A(t)x2+B(t)x.(3)E:\quad y^{2}=x^{3}+A(t)x^{2}+B(t)x. \tag*{(3)}

Since B=(32g)2B=(32g)^{2} and A+64g=a2A+64g=a^{2}, the point T=(32g,32ag)T=(32g,32ag) satisfies 2T=(0,0)2T=(0,0) and has order four. The discriminant is

Δ=224(t+1)4(8t−1)2(15t−8)4(31t−7)4(32t+7)2(512t−113)(128t3−64t2−48t+63).\Delta=2^{24}(t+1)^{4}(8t-1)^{2}(15t-8)^{4}(31t-7)^{4}(32t+7)^{2}(512t-113)(128t^{3}-64t^{2}-48t+63).

The surface has four fibres of type I4I_{4} (at t=−1t=-1, 8/158/15, 7/317/31, ∞\infty), two of type I2I_{2} and four of type I1I_{1}. It is K3, and the Shioda-Tate formula [24] bounds the geometric rank by 20−2−4⋅3−2=420-2-4\cdot3-2=4. The sections

x(P1)=−361(t+1)(31t−7),x(P2)=−4(t+1)(15t−8)(16t−7)2,x(P_{1})=-361(t+1)(31t-7),\qquad x(P_{2})=-4(t+1)(15t-8)(16t-7)^{2},
x(P3)=−16(t+1)(15t−8)(8t+7)2,x(P4)=4(15t−8)(31t−7)(16t+1)2x(P_{3})=-16(t+1)(15t-8)(8t+7)^{2},\qquad x(P_{4})=4(15t-8)(31t-7)(16t+1)^{2}

are independent [9, 10], so E(Q(t))≅Z/4Z×Z4E(\mathbb{Q}(t))\cong\mathbb{Z}/4\mathbb{Z}\times\mathbb{Z}^{4}.

The catalogue

Appendix A lists the split members of a catalogue of 194 rationally soluble quadratic sections F1,…,F194F_{1},\ldots,F_{194} of (3). There are 50 split quadratics, 143 irreducible quadratics and one linear condition, F166F_{166}.

Proposition 3.1. All 194 polynomials are squarefree, and no two are proportional. The only pairs with a common root are {F1,F2}\{F_{1},F_{2}\} (root t=4/7t=4/7) and {F79,F124}\{F_{79},F_{124}\} (root t=134/111t=134/111). Hence these are the only pairs with a joint cover of genus zero, and no triple has a joint cover of genus at most one.

Proof. A common root can only occur among the 51 split or linear conditions, and comparing their roots gives the two pairs. The only linear condition, F166:w2=18t+7F_{166}:w^{2}=18t+7, has branch locus {−7/18,∞}\{-7/18,\infty\}, and no other condition vanishes at −7/18-7/18. So by Lemma 2.1(1) these are the only genus-zero pairs. In any triple at most one of the three pairs has intersecting branch loci, because the two pairs above are disjoint. Hence the union of the three branch loci has at least 5 points, and Lemma 2.1(2) gives genus at least 3.

The two curves

The pair F1,F2F_1,F_2. Here

Q1=−(7t−4)(17t+4),Q2=(7t−4)(5551t−1324),Q_1=-(7t-4)(17t+4),\qquad Q_2=(7t-4)(5551t-1324),

with

x1=64(t+1)2Q1(4t+1)2,x2=576(15t−8)2Q249(28t−39)2.x_1=\frac{64(t+1)^2Q_1}{(4t+1)^2},\qquad x_2=\frac{576(15t-8)^2Q_2}{49(28t-39)^2}.

The substitution t=4(U2−49)/(7(U2+119))t=4(U^2-49)/(7(U^2+119)) solves the first condition. The second becomes the conic 11U2+V2=186311U^2+V^2=1863, which has the point (13,2)(13,2). With

K(q)=72q4−26q3−271q2+286q+8712,K(q)=72q^4-26q^3-271q^2+286q+8712,

the line V=2+q(U−13)V=2+q(U-13) gives

t1(q)=4(3q2−2q−110)(10q2−2q−33)7K(q)(4)t_1(q)=\frac{4(3q^2-2q-110)(10q^2-2q-33)}{7K(q)} \tag*{(4)}
w1=24(13q2−4q−143)(q2+11)K(q),w2=168(11−143q−q2)(q2+11)K(q).w_1=\frac{24(13q^2-4q-143)(q^2+11)}{K(q)},\qquad w_2=\frac{168(11-143q-q^2)(q^2+11)}{K(q)}.

Since 4K(q)=(13q2−4q−143)2+119(q2+11)2>04K(q)=(13q^2-4q-143)^2+119(q^2+11)^2>0, t1t_1 is defined for all rational qq. This is the curve of [9] with r=7qr=7q.

The pair F79,F124F_{79},F_{124}. Here

Q79=(111t−134)(1599t−1526),Q124=−(111t−134)(5217t+1694),Q_{79}=(111t-134)(1599t-1526),\qquad Q_{124}=-(111t-134)(5217t+1694),

with

x79=28224(31t−7)2Q79961(124t−273)2,x124=64g2Q124(111t−41)2(444t2−131t−119)2.x_{79}=\frac{28224(31t-7)^2Q_{79}}{961(124t-273)^2},\qquad x_{124}=\frac{64g^2Q_{124}}{(111t-41)^2(444t^2-131t-119)^2}.

With r=−37w79/(111t−134)r=-37w_{79}/(111t-134) and s=−w124/(111t−134)s=-w_{124}/(111t-134), eliminating tt gives the conic 27r2+1870s2=44457727r^2+1870s^2=444577, and conversely

t=134s2−1694111s2+5217,w79=23976r111s2+5217,w124=887112s111s2+5217.t=\frac{134s^2-1694}{111s^2+5217},\qquad w_{79}=\frac{23976r}{111s^2+5217},\qquad w_{124}=\frac{887112s}{111s^2+5217}.

The conic has the point (−69,13)(-69,13). The line s=13+q(r+69)s=13+q(r+69) gives

r(q)=1863−48620q−129030q227+1870q2,s(q)=351+3726q−24310q227+1870q2.r(q)=\frac{1863-48620q-129030q^2}{27+1870q^2},\qquad s(q)=\frac{351+3726q-24310q^2}{27+1870q^2}.

With R=27+1870q2R=27+1870q^2 and S=351+3726q−24310q2S=351+3726q-24310q^2 this gives

t2(q)=134S2−1694R2111S2+5217R2(5)t_2(q)=\frac{134S^2-1694R^2}{111S^2+5217R^2} \tag*{(5)}

Sections. The new points have x=N/D2x=N/D^2 and y=Hw/D3y=Hw/D^3, where N(N2+AND2+BD4)=H2QN(N^2+AND^2+BD^4)=H^2Q:

N, DN,\ DHH
F1F_164(t+1)2Q1, 4t+164(t+1)^2Q_1,\ 4t+164(t+1)2(8t−1)h164(t+1)^2(8t-1)h_1
F2F_2576(15t−8)2Q2, 7(28t−39)576(15t-8)^2Q_2,\ 7(28t-39)192(8t−1)(15t−8)2h2192(8t-1)(15t-8)^2h_2
F79F_{79}28224(31t−7)2Q79, 31(124t−273)28224(31t-7)^2Q_{79},\ 31(124t-273)1344(8t−1)(31t−7)2h791344(8t-1)(31t-7)^2h_{79}
F124F_{124}64g2Q124, (111t−41)(444t2−131t−119)64g^2Q_{124},\ (111t-41)(444t^2-131t-119)64(t+1)2(8t−1)(15t−8)2(31t−7)2h12464(t+1)^2(8t-1)(15t-8)^2(31t-7)^2h_{124}

Table 3.

where

h1=3448t3−580t2−749t−100,h_1 = 3448t^3 - 580t^2 - 749t - 100,
h2=3756536t3−6792380t2+5347839t−1050120,h_2 = 3756536t^3 - 6792380t^2 + 5347839t - 1050120,
h79=1093902456t3−3605542420t2+5214872859t−2814810390,h_{79} = 1093902456t^3 - 3605542420t^2 + 5214872859t - 2814810390,
h124=185997816t5−919787292t4+486783063t3+128110240t2−79350285t−17513090.h_{124} = 185997816t^5 - 919787292t^4 + 486783063t^3 + 128110240t^2 - 79350285t - 17513090.

Write ti=Ni/Dit_i = N_i/D_i with coprime integral polynomials:

N1=4(3q2−2q−110)(10q2−2q−33),N_1 = 4(3q^2 - 2q - 110)(10q^2 - 2q - 33),
D1=7K(q),D_1 = 7K(q),
N2=2(339199300q4−112385130q3−2766301q2+1622673q+70713),N_2 = 2(339199300q^4 - 112385130q^3 - 2766301q^2 + 1622673q + 70713),
D2=111(6993800q4−1677390q3+14477q2+24219q+1458).D_2 = 111(6993800q^4 - 1677390q^3 + 14477q^2 + 24219q + 1458).

With ai=(8Ni−Di)(32Ni+7Di)a_i = (8N_i - D_i)(32N_i + 7D_i) and gi=(Ni+Di)(15Ni−8Di)(31Ni−7Di)g_i = (N_i + D_i)(15N_i - 8D_i)(31N_i - 7D_i), the curve Ei=Eti(q)E_i = E_{t_i(q)} is isomorphic, via X=Di4xX = D_i^4x, to

Ei: Y2=X3+AiX2+BiX,Ai=ai2−64giDi,Bi=1024giDi2.(6)E_i:\ Y^2 = X^3 + A_iX^2 + B_iX,\qquad A_i = a_i^2 - 64g_iD_i,\qquad B_i = 1024g_iD_i^2. \tag*{(6)}

Exact rank and torsion

Injectivity. In (6), Ai2−4Bi=ai2(ai2−128giDi)A_i^2 - 4B_i = a_i^2(a_i^2 - 128g_iD_i) is not a square, so Theorem 2.2(2) applies. B1B_1 and A12−4B1A_1^2 - 4B_1 each have 9 distinct nonconstant irreducible factors; B2B_2 has 6 and A22−4B2A_2^2 - 4B_2 has 9. We tested 8176 and 2044 values for E1E_1 at q=11/7q = 11/7 (t=10676/136283t = 10676/136283), and 1008 and 1022 values for E2E_2 at q=−2q = -2 (t=2104010001/2318620688t = 2104010001/2318620688). None is a square, so both specializations are injective.

Lower bound. At a good prime pp with 3∣Mp=#Ei,q0(Fp)3 \mid M_p = \#E_{i,q_0}(\mathbb{F}_p), the map P↦[Mp/3]PˉP \mapsto[M_p/3]\bar{P} is a homomorphism to E(Fp)[3]E(\mathbb{F}_p)[3] that kills 3E(Q)3E(\mathbb{Q}) and the torsion. At the primes used, the images of P1,…,P4,PQ,PQ′P_1,\ldots,P_4,P_Q,P_{Q'} lie in one cyclic subgroup, so each prime gives a row over F3\mathbb{F}_3. For E1E_1 we use p=17,19,41,47,61,79p = 17,19,41,47,61,79, and for E2E_2 we use p=7,17,23,29,41,53p = 7,17,23,29,41,53:

(111211011002110001011011012022112221),(112101111110122211110211001120102121).\begin{pmatrix} 1 & 1 & 1 & 2 & 1 & 1 \\ 0 & 1 & 1 & 0 & 0 & 2 \\ 1 & 1 & 0 & 0 & 0 & 1 \\ 0 & 1 & 1 & 0 & 1 & 1 \\ 0 & 1 & 2 & 0 & 2 & 2 \\ 1 & 1 & 2 & 2 & 2 & 1 \end{pmatrix}, \qquad \begin{pmatrix} 1 & 1 & 2 & 1 & 0 & 1 \\ 1 & 1 & 1 & 1 & 1 & 0 \\ 1 & 2 & 2 & 2 & 1 & 1 \\ 1 & 1 & 0 & 2 & 1 & 1 \\ 0 & 0 & 1 & 1 & 2 & 0 \\ 1 & 0 & 2 & 1 & 2 & 1 \end{pmatrix}.

Both are invertible modulo 3. Rational 3-torsion is excluded since #E1,q0(F29)=40\#E_{1,q_0}(\mathbb{F}_{29}) = 40 and #E2,q0(F43)=52\#E_{2,q_0}(\mathbb{F}_{43}) = 52 are prime to 3. So the six points are independent on the fibres, and hence over Q(q)\mathbb{Q}(q).

Upper bound. For C:y2=x3+αx2+βxC : y^2 = x^3 + \alpha x^2 + \beta x and C′:y2=x3−2αx2+(α2−4β)xC' : y^2 = x^3 - 2\alpha x^2 + (\alpha^2 - 4\beta)x we have

rank⁡C(Q)=dim⁡κ(C(Q))+dim⁡κ′(C′(Q))−2,\operatorname{rank} C(\mathbb{Q}) = \dim\kappa(C(\mathbb{Q})) + \dim\kappa'(C'(\mathbb{Q})) - 2,

where κ,κ′\kappa,\kappa' are the Kummer maps of the two-isogenies. The local images on CC and C′C' are exact annihilators under the Hilbert pairing. So explicit local points on one side cut down the possible global classes on the other. For both fibres this leaves 16 classes on each side, giving rank at most 4+4−2=64 + 4 - 2 = 6. The same bound is obtained independently by the two-isogeny descent of ellrank in PARI/GP [23].

Torsion. The reductions of E1,11/7E_{1,11/7} at p=17,43p = 17,43 have 24 and 52 points, and those of E2,−2E_{2,-2} at p=7,43p = 7,43 have 12 and 52. Hence the fibre torsion is Z/4Z\mathbb{Z}/4\mathbb{Z}, and so is that of Ei(Q(q))E_i(\mathbb{Q}(q)). This proves parts (1) and (2) of Theorem A.

Corollary 3.2. For (Q,Q′)=(Q1,Q2)(Q,Q') = (Q_1,Q_2) or (Q79,Q124)(Q_{79},Q_{124}), the twists E(Q)E^{(Q)} and E(Q′)E^{(Q')} of (3) have rank one over Q(t)\mathbb{Q}(t), and E(QQ′)E^{(QQ')} has rank zero.

Proof. Apply (1) with rank E(Q(t))=4E(\mathbb{Q}(t)) = 4 and total rank 6. □

Inequivalence

Write

jE=FG,F=(a4−128a2q+1024q2)3,G=4096q4a2(a2−128q).j_E = \frac{F}{G}, \qquad F = \left(a^4 - 128a^2q + 1024q^2\right)^3, \qquad G = 4096q^4a^2(a^2 - 128q).

with deg⁡F=24\deg F = 24, and let c=jE(4/7)c = j_E(4/7). Since

t1(q)−47=−24(q2+11)2K(q),t_1(q) - \frac{4}{7} = -\frac{24(q^2 + 11)^2}{K(q)},

t1t_1 ramifies over 4/74/7, so cc is a critical value of jE∘t1j_E \circ t_1. The map t2t_2 is branched only over 134/111134/111, 1526/15991526/1599 and −1694/5217-1694/5217. Modulo 101 we find jE(4/7)≡98j_E(4/7) \equiv98, while jE(134/111)≡52j_E(134/111) \equiv52, jE(1526/1599)≡38j_E(1526/1599) \equiv38 and jE(−1694/5217)≡51j_E(-1694/5217) \equiv51. Moreover, F−98G∈F101[t]F - 98G \in\mathbb{F}_{101}[t] has degree 24 and is coprime to its derivative. Hence cc is neither the image of a branch point of t2t_2 nor a critical value of jEj_E, so it is not a critical value of jE∘t2j_E \circ t_2. By § 2.5, this proves part (3) of Theorem A.

Further remarks

No further quadratic sections of the simplest type.

Proposition 3.3. Let y2=x3+ax2+bxy^2 = x^3 + ax^2 + bx be either (6) for i=1,2i = 1,2 or its two-isogenous curve y2=x3−2Aix2+(Ai2−4Bi)xy^2 = x^3 - 2A_i x^2 + (A_i^2 - 4B_i)x. Let mm be any product of the irreducible factors of bb in Z[q]\mathbb{Z}[q], with exponents at most their multiplicities, and let c∈Q×c \in\mathbb{Q}^{\times}. If x=cmx = cm and x3+ax2+bxx^3 + ax^2 + bx has squarefree part of degree at most four, then this squarefree part is constant; that is, the point is defined over a constant field extension of Q(q)\mathbb{Q}(q), or xx is a 2-torsion point. In particular neither E1E_1 nor E2E_2 has a further quadratic section of this form, nor one whose joint cover has genus one.

Sketch of proof. Write Pc=c2m2+cam+bP_c = c^2m^2 + cam + b, so that x3+ax2+bx=cmPcx^3 + ax^2 + bx = cmP_c. The proof has three steps.

Modular sieve. Let pp be a prime at which the factors of bb keep their degrees and stay distinct and squarefree. If the squarefree part over Q\mathbb{Q} has degree at most 4, so does that of the reduction modulo pp of the primitive part. If c≡0c \equiv0 or c≡∞c \equiv\infty modulo pp, this reduction is mbmb, respectively m3m^3, up to constants. So a divisor mm is excluded for all c∈Q×c \in\mathbb{Q}^{\times} as soon as some such pp exists at which all p+1p + 1 residues of cc in P1(Fp)\mathbb{P}^1(\mathbb{F}_p) fail. Over the four curves this excludes 17890 of the 32076 divisors.

Exact step. For each of the remaining 14186 divisors, let G=gcd⁡(m2,am,b)G = \gcd(m^2,am,b) and Pc=GPc′P_c = GP'_c. For cc outside the rational roots of the leading coefficient of Pc′P'_c, of Res⁡q(Pc′,f)\operatorname{Res}_q(P'_c,f) for the irreducible factors ff of mGmG, and of the polynomials attached to the factors of Res⁡c(Pc′,∂qPc′)\operatorname{Res}_c(P'_c,\partial_qP'_c), the polynomial Pc′P'_c is squarefree, coprime to mGmG and of full degree. The squarefree part then has degree at least 16 in every case. Among these values, every one for which the squarefree part has degree at most four gives a constant squarefree part, or a 2-torsion point.

Pencil cases. For the two divisors m=ai2−128giDim = a_i^2 - 128g_iD_i and m=ai2m = a_i^2 (up to constants) on each isogenous side, the polynomial Pc′P'_c factors as (c−c0)(cU+V)(c-c_0)(cU+V). These were treated in the same way with Res⁡c\operatorname{Res}_c replaced by UV′−VU′UV'-VU'; the only special value is c=c0c=c_0, for which xx is a 2-torsion point. □

By Proposition 3.1, the catalogue also contains no triple leading to a genus-one cover.

A conditional rank-seven specialization. At q=3/94q = 3/94, i.e. t=252401754905/264613965267t = 252401754905/264613965267, the fibre of E2E_2 is isomorphic to y2=x3+a2x2+a4xy^2 = x^3 + a_2x^2 + a_4x with a2=216235610937917987314078524977768347645903988121745a_2 = 216235610937917987314078524977768347645903988121745, a4=1904441638805661511492199946744527042653541957019129936327806604179289574729399252763025788264054784a_4 = 1904441638805661511492199946744527042653541957019129936327806604179289574729399252763025788264054784.

For this curve:

  • the six sections remain independent;

  • the two-isogeny descent gives rank at most 7;

  • the 2-Selmer group has dimension 8 (computed with ell2cover);

  • the root number is −1-1.

All of these were computed with PARI/GP [23]. As a check, the same computation at the fibre q=−2q = -2, whose rank is proved to be six, gives a 2-Selmer group of dimension 7. Since E(Q)[2∞]≅Z/4ZE(\mathbb{Q})[2^\infty] \cong\mathbb{Z}/4\mathbb{Z}, the 2∞2^\infty-Selmer corank ss satisfies 6≤s≤76 \le s \le7. By the 2-parity theorem [21, 1], ss is odd, so s=7s = 7. Hence the rank is seven unless \Sha(E/\mathbb{Q})[2^\infty] is infinite. We have not found a seventh point.

Torsion Z/2Z×Z/6Z\mathbb{Z}/2\mathbb{Z} \times\mathbb{Z}/6\mathbb{Z}: nine curves of rank two

The universal curve

Let

Ek:y2=x(x−(k−1)3(k+3))(x−(k+1)3(k−3))=x3−2(k4−6k2−3)x2+(k2−1)3(k2−9)x(7)E_k : y^2 = x\left(x-(k-1)^3(k+3)\right)\left(x-(k+1)^3(k-3)\right) = x^3 - 2(k^4-6k^2-3)x^2 + (k^2-1)^3(k^2-9)x \tag*{(7)}

[16]. Besides the points of order two, it has the point T3=((k2−1)2,4(k2−1)2)T_3 = \left((k^2-1)^2, 4(k^2-1)^2\right) of order three.

(7) is Hadano’s model [12] with a=1a = 1, b=(k2−1)/4b = (k^2-1)/4, x=16Xx = 16X. It is isomorphic over Q(v)\mathbb{Q}(v) to the model y2=x3+A26(v)x2+B26(v)xy^2 = x^3 + A_{26}(v)x^2 + B_{26}(v)x of [7, 6] with k=(3v−5)/(v+1)k = (3v-5)/(v+1).

The discriminant is 212k2(k2−1)6(k2−9)22^{12}k^2(k^2-1)^6(k^2-9)^2, with fibres I6I_6 at k=±1,∞k = \pm1,\infty and I2I_2 at k=0,±3k = 0,\pm3. The trivial lattice of this K3 surface has rank 2+15+3=202+15+3=20, so rank Ek(Q(k))=0E_k(\mathbb{Q}(k)) = 0. The jj-invariant is

j(k)=(k2+3)3(k6−15k4+75k2+3)3k2(k−3)2(k−1)6(k+1)6(k+3)2.j(k) = \frac{(k^2+3)^3(k^6-15k^4+75k^2+3)^3}{k^2(k-3)^2(k-1)^6(k+1)^6(k+3)^2}.

The group GG of Möbius maps k↦±kk \mapsto\pm k, ±(k+3)/(k−1)\pm(k+3)/(k-1), ±(k−3)/(k+1)\pm(k-3)/(k+1) has order six. For each σ∈G\sigma\in G, Eσ(k)≅EkE_{\sigma(k)} \cong E_k over Q(k)\mathbb{Q}(k); for example, for u=(k+3)/(k−1)u = (k+3)/(k-1),

x=(k−1)416xu+(k+1)3(k−3),y=(k−1)664yu.x = \frac{(k-1)^4}{16}x_u + (k+1)^3(k-3), \qquad y = \frac{(k-1)^6}{64}y_u.

The catalogue and the joint covers

Appendix B lists 42 quadratic conditions C1,…,C42C_1,\ldots,C_{42}, one in each orbit of k↦−kk \mapsto-k. Each has a rational parametrization and a point of infinite order. We write Ci+C_i^+ for Qi(k)Q_i(k) and Ci−C_i^- for Qi(−k)Q_i(-k).

Proposition 4.1. Fourteen of the conditions are even, namely CiC_i for

i∈{3,4,5,6,7,10,11,12,13,18,28,30,33,38},i \in\{3,4,5,6,7,10,11,12,13,18,28,30,33,38\},

and for these Ci+=Ci−C_i^+ = C_i^-. After identifying these repetitions, the signed catalogue contains 70 distinct conditions, no two of them proportional. Among unordered pairs of distinct conditions, 84 have a common root. These form 42 classes under simultaneous reflection k↦−kk \mapsto-k, and exactly 21 of the classes have rationally soluble joint covers. These 21 form seven orbits of three under GG (Table 1).

Proof. For Q1=c1(k−r)(k−a)Q_1=c_1(k-r)(k-a) and Q2=c2(k−r)(k−b)Q_2=c_2(k-r)(k-b), put s=1/(k−r)s=1/(k-r); the joint cover becomes the conic

r−bc1u2−r−ac2w2=a−b.\frac{r-b}{c_1}u^2-\frac{r-a}{c_2}w^2=a-b.

We decided its solubility with qfsolve [23]. Two pairs are equivalent when some σ∈G\sigma\in G maps the subgroup ⟨Q1,Q2⟩⊂Q(k)×/Q(k)×2\langle Q_1,Q_2\rangle\subset\mathbb{Q}(k)^\times/\mathbb{Q}(k)^{\times2} onto the other subgroup. Since Eσ(k)≅EkE_\sigma(k)\cong E_k, equivalent covers give equivalent curves. □

ClassPairs
U1U_1C4+/C20+C_4^+/C_{20}^+, C9+/C13+C_9^+/C_{13}^+, C9+/C2−C_9^+/C_2^-
U2U_2C4+/C22+C_4^+/C_{22}^+, C9+/C21+C_9^+/C_{21}^+, C9+/C24−C_9^+/C_{24}^-
U3U_3C4+/C21+C_4^+/C_{21}^+, C19/C22+C_{19}/C_{22}^+, C19+/C24−C_{19}^+/C_{24}^-
U4U_4C8+/C41+C_8^+/C_{41}^+, C8+/C42+C_8^+/C_{42}^+, C12+/C40+C_{12}^+/C_{40}^+
U5U_5C13+/C24+C_{13}^+/C_{24}^+, C20+/C21+C_{20}^+/C_{21}^+, C20+/C22+C_{20}^+/C_{22}^+
U6U_6C14+/C15+C_{14}^+/C_{15}^+, C15+/C17+C_{15}^+/C_{17}^+, C16+/C18+C_{16}^+/C_{18}^+
U7U_7C34+/C35+C_{34}^+/C_{35}^+, C35+/C37+C_{35}^+/C_{37}^+, C36+/C38+C_{36}^+/C_{38}^+

Table 1 (PDF p. 9). The 21 rational joint covers, grouped into classes.

The nine curves

For U1,…,U7U_1,\ldots,U_7 we take the first pair of each class; Table 2 gives the degree-four map k=ki(t)k=k_i(t). UiU_i is the pullback Eki(t)E_{k_i(t)}, with the two sections whose xx-coordinates are in Appendix § 6.

Conditionski(t)k_i(t)
U1 C4+,C20+U_1\ C_4^+, C_{20}^+−t4+24t3−152t2+336t−196t4−12t3+62t2−168t+196\frac{-t^4+24t^3-152t^2+336t-196}{t^4-12t^3+62t^2-168t+196}
U2 C4+,C22+U_2\ C_4^+, C_{22}^+t4+8t3−56t2+48t+36t4−4t3+10t2−24t+36\frac{t^4+8t^3-56t^2+48t+36}{t^4-4t^3+10t^2-24t+36}
U3 C4+,C21+U_3\ C_4^+, C_{21}^+5(t4+36)(t2−6t+6)(t2+6t+6)\frac{5(t^4+36)}{(t^2-6t+6)(t^2+6t+6)}
U4 C8+,C41+U_4\ C_8^+, C_{41}^+3(t4−2t2+4)2(t2−t+2)(t2+t+2)\frac{3(t^4-2t^2+4)}{2(t^2-t+2)(t^2+t+2)}
U5 C13+,C24+U_5\ C_{13}^+, C_{24}^+5t4+8t3−2t2−8t+53(t−1)(t+1)(t+3)(3t−1)\frac{5t^4+8t^3-2t^2-8t+5}{3(t-1)(t+1)(t+3)(3t-1)}
U6 C14+,C15+U_6\ C_{14}^+, C_{15}^+−(3t2−7)(7t2−3)(3t2−5t+3)(3t2+5t+3)\frac{-(3t^2-7)(7t^2-3)}{(3t^2-5t+3)(3t^2+5t+3)}
U7 C34+,C35+U_7\ C_{34}^+, C_{35}^+−3(3t4+7t2+27)(t2+7)(7t2+9)\frac{-3(3t^4+7t^2+27)}{(t^2+7)(7t^2+9)}

Table 2. Representatives U1,…,U7U_1,\ldots,U_7.

The condition C29−:v2=9k2−22k+49C_{29}^-:v^2=9k^2-22k+49 has

x=(k−1)2(k2−2k+13)2(k−3)2,k=t2+12t−13t2−9.x=\frac{(k-1)^2(k^2-2k+13)^2}{(k-3)^2},\qquad k=\frac{t^2+12t-13}{t^2-9}.

With d=t2−9d = t^2 - 9 and q0=t2+6t−11q_0 = t^2 + 6t - 11, the pullback is isomorphic, via x29=d4x/16x_{29} = d^4x/16 and y29=d6y/64y_{29} = d^6y/64, to

C29:y2=x3+a29x2+b29x,C_{29}: y^2 = x^3 + a_{29}x^2 + b_{29}x,

where

a29=t8+12t7−40t6−1188t5−1134t4+13860t3−13360t2−5772t+9157,a_{29} = t^8 + 12t^7 - 40t^6 - 1188t^5 - 1134t^4 + 13860t^3 - 13360t^2 - 5772t + 9157,
b29=−16(t−7)(t−2)(t+1)(t+5)(3t−1)3q03.b_{29} = -16(t - 7)(t - 2)(t + 1)(t + 5)(3t - 1)^3q_0^3.

It has the section with

x=4(t+5)(3t−1)3q0,y=4(t−1)(t+3)(t+5)(3t−1)3(t2+2t+17)q0.x = 4(t + 5)(3t - 1)^3q_0,\qquad y = 4(t - 1)(t + 3)(t + 5)(3t - 1)^3(t^2 + 2t + 17)q_0.

On C29C_{29} we impose two further conditions:

D1:w2=4t2+19t−29,x=(t+1)(t+5)(3t−1)3q04;D_1: w^2 = 4t^2 + 19t - 29,\qquad x = \frac{(t + 1)(t + 5)(3t - 1)^3q_0}{4};
D2:w2=3(t−3),x=−64(t−7)(t−2)(9t3−5t2−5t+73)2(t2−5t+22)2.D_2: w^2 = 3(t - 3),\qquad x = -\frac{64(t - 7)(t - 2)(9t^3 - 5t^2 - 5t + 73)^2}{(t^2 - 5t + 22)^2}.

They are parametrized by

t=T1(u)=5u2−8u+964(u−2)(u+2),t=T2(u)=3(u2+1).t = T_1(u) = \frac{5u^2 - 8u + 96}{4(u - 2)(u + 2)},\qquad t = T_2(u) = 3(u^2 + 1).

We put U8=C29(T1(u))U_8 = C_{29}(T_1(u)) and U9=C29(T2(u))U_9 = C_{29}(T_2(u)).

Two further conditions give nothing new.

  • D3:w2=−2(t2+8t−17)D_3: w^2 = -2(t^2 + 8t - 17) is parametrized by t=T3(u)=−(3u2+136u+170)/(11(u2+2))t = T_3(u) = -(3u^2 + 136u + 170)/(11(u^2 + 2)), and

k29(T3(16−18u5u+20))=σ(k29(T1(u))),σ(k)=k+3k−1.k_{29}\left(T_3\left(\frac{16 - 18u}{5u + 20}\right)\right) = \sigma\left(k_{29}(T_1(u))\right),\qquad\sigma(k) = \frac{k + 3}{k - 1}.
  • D4:w2=−6(t+3)(3t−1)D_4: w^2 = -6(t + 3)(3t - 1) is parametrized by t=T4(u)=(6−3u2)/(u2+18)t = T_4(u) = (6 - 3u^2)/(u^2 + 18), and k29(T4(4/u))=k29(T2(u))k_{29}(T_4(4/u)) = k_{29}(T_2(u)).

Here k29(t)=(t2+12t−13)/(t2−9)k_{29}(t) = (t^2 + 12t - 13)/(t^2 - 9). These are identities of rational functions of degree four, and σ∈G\sigma\in G, so D3D_3 gives U8U_8 and D4D_4 gives U9U_9. The map T2T_2 ramifies over t=3,∞t = 3, \infty, which lie over the I6I_6 fibres k=∞,1k = \infty, 1, so U9U_9 has two fibres of type I12I_{12}.

Exact rank and torsion

For k=N/Dk = N/D the curve has the integral model

y2=x(x−(N−D)3(N+3D))(x−(N+D)3(N−3D))(8)y^2 = x\bigl(x - (N - D)^3(N + 3D)\bigr)\bigl(x - (N + D)^3(N - 3D)\bigr) \tag*{(8)}

to which Theorem 2.2(1) applies. For each UiU_i we found the value t0t_0 below at which the criterion holds, testing all three products. At this t0t_0 the two sections are independent and ellrank [23] gives rank exactly two. The fibre torsion is Z/2Z×Z/6Z\mathbb{Z}/2\mathbb{Z} \times\mathbb{Z}/6\mathbb{Z}.

U1U_1U2U_2U3U_3U4U_4U5U_5U6U_6U7U_7U8U_8U9U_9
t0t_0−2/5-2/5−5/7-5/7−1/6-1/6−3/8-3/86/56/5−12/7-12/7−1/4-1/4−3/5-3/5−3/5-3/5

Table 4.

This proves part (1) of Theorem B. We do not claim that the given sections generate the Mordell-Weil group; by injectivity, that question reduces to the fibres above, as in [6].

Corollary 4.2. Let Q,Q′Q, Q' be the conditions defining one of U1,…,U7U_1,\ldots,U_7. Over Q(k)\mathbb{Q}(k), Ek(Q)E_k^{(Q)} and Ek(Q′)E_k^{(Q')} have rank one, and Ek(QQ′)E_k^{(QQ')} has rank zero. C29C_{29} has rank one over Q(t)\mathbb{Q}(t), and so do its twists by 4t2+19t−294t^2 + 19t - 29 and 3(t−3)3(t - 3).

Proof. For U1,…,U7U_1,\ldots,U_7, apply (1) with rank Ek(Q(k))=0E_k(\mathbb{Q}(k))=0 and total rank 2. For U8,U9U_8,U_9, apply the same argument to the quadratic extension Q(u)/Q(t)\mathbb{Q}(u)/\mathbb{Q}(t). □\square

Inequivalence

For U1,…,U8U_1,\ldots,U_8 the map k(t)k(t) is unramified over k=±1,∞k=\pm1,\infty. So the order-six poles of j∘kj\circ k are the twelve roots of F=D(N2−D2)F=D(N^2-D^2), with N,DN,D homogenized. The invariant (2) takes the values

Classι\iota
U1U_13290567866923750/12504895030321213290567866923750/1250489503032121
U2U_25728687791846/25504505221695728687791846/2550450522169
U3U_3213643125/49441568213643125/49441568
U4U_4201327525/72673568201327525/72673568
U5U_57026563975136/17860264349297026563975136/1786026434929
U6U_62810533160401893654/11455239356433273612810533160401893654/1145523935643327361
U7U_7251576064840491543371254/193174570213501578374761251576064840491543371254/193174570213501578374761
U8U_8704606945786056950/881192578069527889704606945786056950/881192578069527889

Table 5.

These eight values are distinct. The pole divisor of j∘kj\circ k for U9U_9 consists of two poles of order 12, eight of order 6 and twelve of order 2; for U1,…,U8U_1,\ldots,U_8 it consists of twelve poles of order 6 and twelve of order 2. By §2.5(ii), this proves part (2) of Theorem B. Together with Proposition 4.1, it also proves part (4).

Comparison with earlier constructions

The three curves of [7] are obtained as follows.

  • The first two come from v=3/(5−w2)v=3/(5-w^2) followed by 2w2−7=□2w^2-7=\square, respectively 5w2−4=□5w^2-4=\square. Under k=(3v−5)/(v+1)k=(3v-5)/(v+1) these are the pairs C79/C2020C^9_7/C^{20}_{20} (class U1U_1) and C2020/C21−21C^{20}_{20}/C^{-21}_{21}, the reflection of C2020/C21+21C^{20}_{20}/C^{+21}_{21} (class U5U_5).

  • The third, built from Hadano’s model, is the curve of [6 §8.4], and hence equivalent to U8U_8 by the table below. Directly, with kDP(u)k_{DP}(u) the corresponding map to the kk-line, kDP(−(9u+12)/(2u+16))=σ(k29(T1(u)))k_{DP}(-(9u+12)/(2u+16))=\sigma(k_{29}(T_1(u))) with σ(k)=−(k−3)/(k+1)\sigma(k)=-(k-3)/(k+1).

For the five curves of [6 §8], with uu their parameter:

[6]ClassChange of parameterσ∈G\sigma\in G
§8.1U1U_1t=ut=u−(k−3)/(k+1)-(k-3)/(k+1)
§8.2U2U_2t=ut=u−(k−3)/(k+1)-(k-3)/(k+1)
§8.3U5U_5t=(u−6)/(u−18)t=(u-6)/(u-18)−(k+3)/(k−1)-(k+3)/(k-1)
§8.4U8U_8uU8=4(7u+24)/(19u+18)u_{U_8}=4(7u+24)/(19u+18)(k+3)/(k−1)(k+3)/(k-1)
§8.5U6U_6t=(u−1)/(u+1)t=(u-1)/(u+1)kk

Table 1 (PDF p. 11).

In each line, write v(u)v(u) for the parametrization of [6] and kUik_{U_i} for our map. Then (3v(u)−5)/(v(u)+1)=σ(kUi(t(u)))(3v(u)-5)/(v(u)+1)=\sigma(k_{U_i}(t(u))) is an identity of rational functions of degree four. Since E(3v−5)/(v+1)E_{(3v-5)/(v+1)} is isomorphic over Q(v)\mathbb{Q}(v) to the model of [6], and Eσ(k)≅EkE_{\sigma(k)}\cong E_k over Q(k)\mathbb{Q}(k), the curves are isomorphic over Q(u)\mathbb{Q}(u), not merely twists of each other. By [6 §1.3], MacLeod’s examples [19] are isomorphic to the first three of these curves. Hence U3,U4,U7,U9U_3,U_4,U_7,U_9 are not equivalent to any of these curves, which proves part (3) of Theorem B.

Appendix A. Split quadratic sections of Elkies’s surface

The table lists the 50 split conditions and the linear condition F166F_{166} of the Z/4Z\mathbb{Z}/4\mathbb{Z} catalogue, in its original numbering.

LabelQi(t)Q_i(t)
F1F_1−(7t−4)(17t+4)-(7t-4)(17t+4)
F2F_2(7t−4)(5551t−1324)(7t-4)(5551t-1324)
F47F_{47}(50209t+18034)(553554225t+1143732242)(50209t+18034)(553554225t+1143732242)
F48F_{48}(87t−56)(346262175t−569042072)(87t-56)(346262175t-569042072)
F49F_{49}(2431t−382)(26801775t−28655582)(2431t-382)(26801775t-28655582)
F50F_{50}(95t−22)(855t−623)(95t-22)(855t-623)
F51F_{51}(928t−1241)(14848t−12881)(928t-1241)(14848t-12881)
F52F_{52}(3289t+967)(3289t+4792)(3289t+967)(3289t+4792)
F53F_{53}(399t−128)(9975t+1543)(399t-128)(9975t+1543)
F54F_{54}(104t−113)(6656t−1793)(104t-113)(6656t-1793)
F55F_{55}(1408t+13049)(90112t+22961)(1408t+13049)(90112t+22961)
F56F_{56}(5880t+5041)(7680t+4081)(5880t+5041)(7680t+4081)
F57F_{57}(552t−553)(317952t−372673)(552t-553)(317952t-372673)
F58F_{58}(11968t−3017)(765952t−274673)(11968t-3017)(765952t-274673)
F59F_{59}1793(4352t−3019)(69632t−314843)1793(4352t-3019)(69632t-314843)
F67F_{67}(3615t+238)(5135t−1138)(3615t+238)(5135t-1138)
F71F_{71}(7089t+2942)(19185t+15038)(7089t+2942)(19185t+15038)
F75F_{75}(6041t+5573)(44849t+12349)(6041t+5573)(44849t+12349)
F79F_{79}(111t−134)(1599t−1526)(111t-134)(1599t-1526)
F95F_{95}(4t−421)(4t−1)(4t-421)(4t-1)
F109F_{109}(2279t−2563)(13039t−13483)(2279t-2563)(13039t-13483)
F111F_{111}(113152t−32369)(12864512t+2563111)(113152t-32369)(12864512t+2563111)
F112F_{112}−(409088t+89489)(1024095232t−953498393)-(409088t+89489)(1024095232t-953498393)
F116F_{116}(69719t−26293)(3242879t−568621)(69719t-26293)(3242879t-568621)
F122F_{122}(t+1)(1537t+889)(t+1)(1537t+889)
F124F_{124}−(111t−134)(5217t+1694)-(111t-134)(5217t+1694)
F125F_{125}(2644575t−748727)(140839575t−17847503)(2644575t-748727)(140839575t-17847503)
F129F_{129}−(181185t+40187)(757335t−187763)-(181185t+40187)(757335t-187763)
F131F_{131}−(407745t+89747)(4286295t−881483)-(407745t+89747)(4286295t-881483)
F132F_{132}−(266305t+1385482)(927095t−4164322)-(266305t+1385482)(927095t-4164322)
F141F_{141}2(1432t−1943)(18431t−17119)2(1432t-1943)(18431t-17119)
F142F_{142}(8t+1175)(392t+167)(8t+1175)(392t+167)
F143F_{143}−(5888t+2513)(13919968t−71114057)-(5888t+2513)(13919968t-71114057)
F144F_{144}(5888t−1961)(1849568t−559721)(5888t-1961)(1849568t-559721)
F145F_{145}2(24t−7)(57615t−20288)2(24t-7)(57615t-20288)
F146F_{146}−(9728t+2753)(558837120t−385598929)-(9728t+2753)(558837120t-385598929)
F147F_{147}(9728t−1841)(2881920t−783889)(9728t-1841)(2881920t-783889)
F149F_{149}2(2072t−359)(595999t−356953)2(2072t-359)(595999t-356953)
F150F_{150}(7688t−2921)(13448t−2681)(7688t-2921)(13448t-2681)
F151F_{151}(36608t+17033)(1032450848t+816636473)(36608t+17033)(1032450848t+816636473)
F152F_{152}(36608t−13601)(500326112t−387820889)(36608t-13601)(500326112t-387820889)
F153F_{153}2(9t+77)(680t+239)2(9t+77)(680t+239)
F154F_{154}2(49t+22)(872t+791)2(49t+22)(872t+791)
F155F_{155}−2(25t+7)(5592t−3199)-2(25t+7)(5592t-3199)
F166F_{166}18t+718t+7
F189F_{189}(356168t+322679)(1417928t+945959)(356168t+322679)(1417928t+945959)
F190F_{190}(6083072t+5817847)(11727872t+7052647)(6083072t+5817847)(11727872t+7052647)

Table 1 (PDF p. 12).

LabelQi(t)Q_i(t)
F191F_{191}(763466888t−133116361)(1167217928t−92109241)(763466888t-133116361)(1167217928t-92109241)
F192F_{192}(1169538248t+535858319)(3207311048t+281136719)(1169538248t+535858319)(3207311048t+281136719)
F193F_{193}(7120209613952t−1606644901673)(7202855130752t−1588566194873)(7120209613952t-1606644901673)(7202855130752t-1588566194873)
F194F_{194}(533496016772168t+414269992621679)(563331048336968t+410540613676079)(533496016772168t+414269992621679)(563331048336968t+410540613676079)

Table 6.

Appendix B. Quadratic conditions on the universal Z/2Z×Z/6Z\mathbb{Z}/2\mathbb{Z} \times\mathbb{Z}/6\mathbb{Z} curve

For each ii the conic v2=Qi(k)v^2=Q_i(k) has a rational point, and xi(k)x_i(k) is the xx-coordinate of a point of infinite order on EkE_k over Q(k)(Qi)\mathbb{Q}(k)(\sqrt{Q_i}).

Qi(k)Q_i(k)xi(k)x_i(k)
C1C_13(k2−4k+7)3(k^2-4k+7)(k−3)(k+1)2(k+3)3\frac{(k-3)(k+1)^2(k+3)}{3}
C2C_2k2−4k−1k^2-4k-1(k−3)(k+1)2(k+3)(k-3)(k+1)^2(k+3)
C3C_33(3k2+1)3(3k^2+1)(k+1)3(k+3)(k+1)^3(k+3)
C4C_4−3(k−2)(k+2)-3(k-2)(k+2)−(k−3)(k−1)(k+1)(k+3)3-\frac{(k-3)(k-1)(k+1)(k+3)}{3}
C5C_55−k25-k^2−(k−3)(k−1)(k+1)2-(k-3)(k-1)(k+1)^2
C6C_6k2+3k^2+3(k−3)(k−1)(k+1)2(k-3)(k-1)(k+1)^2
C7C_73(k−5)(k+5)3(k-5)(k+5)27(k−3)(k+3)27(k-3)(k+3)
C8C_82(k−9)(2k−3)2(k-9)(2k-3)(k−1)(k+1)2(2k−3)2\frac{(k-1)(k+1)^2(2k-3)}{2}
C9C_93(k+5)(3k−1)3(k+5)(3k-1)(k−3)(k−1)2(k+5)(k-3)(k-1)^2(k+5)
C10C_{10}3(3k2−2)3(3k^2-2)k2(3k2−2)3\frac{k^2(3k^2-2)}{3}
C11C_{11}5k2+45k^2+4(k−1)2(k+1)2(5k2+4)k2\frac{(k-1)^2(k+1)^2(5k^2+4)}{k^2}
C12C_{12}(5k−3)(5k+3)(5k-3)(5k+3)(k−3)(k−1)2(k+1)2(k+3)k2\frac{(k-3)(k-1)^2(k+1)^2(k+3)}{k^2}
C13C_{13}(3k−1)(3k+1)(3k-1)(3k+1)(k−1)3(k+1)3k2\frac{(k-1)^3(k+1)^3}{k^2}
C14C_{14}−3(3k+7)(11k−16)-3(3k+7)(11k-16)(k+1)(3k−5)2(3k+7)27\frac{(k+1)(3k-5)^2(3k+7)}{27}
C15C_{15}3(k+4)(3k+7)3(k+4)(3k+7)(k+1)(k+9)2(3k+7)3\frac{(k+1)(k+9)^2(3k+7)}{3}
C16C_{16}6(5k−49)(5k+1)6(5k-49)(5k+1)27(k−3)(k+1)2(k+3)32\frac{27(k-3)(k+1)^2(k+3)}{32}
C17C_{17}3(k+4)(27k−17)3(k+4)(27k-17)(k+1)3(27k−17)27\frac{(k+1)^3(27k-17)}{27}
C18C_{18}6(5k−1)(5k+1)6(5k-1)(5k+1)3(k−3)(k−1)2(k+1)2(k+3)128k2\frac{3(k-3)(k-1)^2(k+1)^2(k+3)}{128k^2}
C19C_{19}−6(k−2)(3k−1)-6(k-2)(3k-1)(k+1)3(3k−7)2(3k−1)27(k−1)2\frac{(k+1)^3(3k-7)^2(3k-1)}{27(k-1)^2}
C20C_{20}2(k−5)(k−2)2(k-5)(k-2)(k−5)(k−1)3(k+1)2(k−3)2\frac{(k-5)(k-1)^3(k+1)^2}{(k-3)^2}
C21C_{21}3(k−5)(3k−5)3(k-5)(3k-5)(k−3)(k−1)3(k+1)2(k−2)2\frac{(k-3)(k-1)^3(k+1)^2}{(k-2)^2}
C22C_{22}6(k−7)(k−2)6(k-7)(k-2)3(k−2)(k−1)2(k+1)2\frac{3(k-2)(k-1)^2(k+1)}{2}
C23C_{23}3(3k2−16k+25)3(3k^2-16k+25)(k−7)2(k−3)(k−1)(k-7)^2(k-3)(k-1)
C24C_{24}6(2k−1)(3k+1)6(2k-1)(3k+1)(k+1)2(k+3)(2k−1)2\frac{(k+1)^2(k+3)(2k-1)}{2}
C25C_{25}(k−9)(5k+3)(k-9)(5k+3)(k−1)3(5k+3)4\frac{(k-1)^3(5k+3)}{4}
C26C_{26}9k2−12k−1219k^2-12k-121(k−9)2(k−3)(k+3)(k-9)^2(k-3)(k+3)

Table 7.

Qi(k)Q_i(k)xi(k)x_i(k)
C27C_{27}3(k2−22k+25)3(k^2-22k+25)(k+1)2(7k−9)248\frac{(k+1)^2(7k-9)^2}{48}
C28C_{28}−(2k−3)(2k+3)-(2k-3)(2k+3)(k2−3)2(k^2-3)^2
C29C_{29}9k2+22k+499k^2+22k+49(k+1)2(k2+2k+13)2(k+3)2\frac{(k+1)^2(k^2+2k+13)^2}{(k+3)^2}
C30C_{30}3(11k2+1)3(11k^2+1)3(k−1)(k+1)(11k2+1)3(k-1)(k+1)(11k^2+1)
C31C_{31}−31k2+68k−1-31k^2+68k-1−(k−3)(k−1)(31k2−68k+1)-(k-3)(k-1)(31k^2-68k+1)
C32C_{32}−25k2+80k−19-25k^2+80k-19−(k−3)(k−1)(7k−1)2-(k-3)(k-1)(7k-1)^2
C33C_{33}3(k−9)(k+9)3(k-9)(k+9)(k−3)(k+3)(k2+63)2243\frac{(k-3)(k+3)(k^2+63)^2}{243}
C34C_{34}−(7k+9)(16k−33)-(7k+9)(16k-33)−(k+3)(5k−9)2(7k+9)(17k−81)2729(37k−81)2-\frac{(k+3)(5k-9)^2(7k+9)(17k-81)^2}{729(37k-81)^2}
C35C_{35}−(4k+3)(7k+9)-(4k+3)(7k+9)−(k+3)(3k+1)2(7k+9)(15k2+33k+16)2(47k2+93k+48)2-\frac{(k+3)(3k+1)^2(7k+9)(15k^2+33k+16)^2}{(47k^2+93k+48)^2}
C36C_{36}2(k+15)(49k−15)2(k+15)(49k-15)729(k−1)3(k+1)3(59k−15)232(275k2−192k+45)2\frac{729(k-1)^3(k+1)^3(59k-15)^2}{32(275k^2-192k+45)^2}
C37C_{37}(4k+3)(17k−81)(4k+3)(17k-81)(k+3)(17k−81)(65k2−177k−144)2729(11k−48)2\frac{(k+3)(17k-81)(65k^2-177k-144)^2}{729(11k-48)^2}
C38C_{38}2(k−15)(k+15)2(k-15)(k+15)(k−1)3(k+1)3(k2+375)2128(3k2+125)2\frac{(k-1)^3(k+1)^3(k^2+375)^2}{128(3k^2+125)^2}
C39C_{39}−6(2k−3)(5k−3)-6(2k-3)(5k-3)−(k−3)(5k−3)(19k2−36k+9)2243(k−1)2-\frac{(k-3)(5k-3)(19k^2-36k+9)^2}{243(k-1)^2}
C40C_{40}−(5k−9)(5k−3)-(5k-9)(5k-3)−(k−3)(k−1)3(11k−21)2(2k−3)2-\frac{(k-3)(k-1)^3(11k-21)^2}{(2k-3)^2}
C41C_{41}−2(2k−3)(7k−3)-2(2k-3)(7k-3)(k+3)(2k−3)(8k2−9k+3)22(8k−3)2\frac{(k+3)(2k-3)(8k^2-9k+3)^2}{2(8k-3)^2}
C42C_{42}2(k−6)(k+9)2(k-6)(k+9)(k−6)(k+1)3(7k−27)22(5k−27)2\frac{(k-6)(k+1)^3(7k-27)^2}{2(5k-27)^2}

Table 8 (PDF p. 14).

Acknowledgments

A.D. acknowledges support from the Croatian Science Foundation under the project no. IP2022-10-5008 (TEBAG), the project “Implementation of cutting-edge research and its application as part of the Scientific Center of Excellence for Quantum and Complex Systems, and Representations of Lie Algebras”, Grant No. PK.1.1.10.0004, co-financed by the European Union through the European Regional Development Fund – Competitiveness and Cohesion Programme 2021–2027, and the European Union: NextGenerationEU through the National Recovery and Resilience Plan 2021–2026, Institutional grant of University of Zagreb Faculty of Science (IK IA 1.1.3. Impact4Math).

AI systems were used to optimise the programs and convert the resulting data into a readable format. The authors are responsible for the final manuscript.

References

[1] T. Dokchitser and V. Dokchitser, On the Birch–Swinnerton-Dyer quotients modulo squares, Ann. of Math. (2) 172 (2010), 567–596.

[2] A. Dujella, On Mordell–Weil groups of elliptic curves induced by Diophantine triples, Glas. Mat. Ser. III 42 (2007), 3–18.

[3] A. Dujella, High rank elliptic curves with prescribed torsion, https://web.math.pmf.unizg.hr/~duje/tors/tors.html.

[4] A. Dujella, Infinite families of elliptic curves with high rank and prescribed torsion, https://web.math.pmf.unizg.hr/~duje/tors/generic.html.

[5] A. Dujella, I. Gusić and P. Tadić, The rank and generators of Kihara’s elliptic curve with torsion Z/4Z\mathbb{Z}/4\mathbb{Z} over Q(t)\mathbb{Q}(t), Proc. Japan Acad. Ser. A Math. Sci. 91 (2015), 105–109.

Department of Mathematics, Faculty of Science, University of Zagreb, Bijenička cesta 30, 10000 Zagreb, Croatia

Email address, A. Dujella: [email protected]

Departamento de Matemáticas, Universidad del País Vasco, Aptdo. 644, 48080 Bilbao, Spain

Email address, J. C. Peral: [email protected]

References

  1. [1]T. Dokchitser and V. Dokchitser, On the Birch–Swinnerton-Dyer quotients modulo squares, Ann. of Math. (2) 172 (2010), 567–596.
  2. [2]A. Dujella, On Mordell–Weil groups of elliptic curves induced by Diophantine triples, Glas. Mat. Ser. III 42 (2007), 3–18.DOI
  3. [3]A. Dujella, High rank elliptic curves with prescribed torsion, https://web.math.pmf.unizg.hr/~duje/tors/tors.html
  4. [4]A. Dujella, Infinite families of elliptic curves with high rank and prescribed torsion, https://web.math.pmf.unizg.hr/~duje/tors/generic.html.
  5. [5]A. Dujella, I. Gusić and P. Tadić, The rank and generators of Kihara’s elliptic curve with torsion ℤ/4ℤ over ℚ(t), Proc. Japan Acad. Ser. A Math. Sci. 91 (2015), 105–109.
  6. [6]A. Dujella, M. Kazalicki and J. C. Peral, Elliptic curves with torsion groups ℤ/8ℤ and ℤ/2ℤ × ℤ/6ℤ, Rev. R. Acad. Cienc. Exactas Fís. Nat. Ser. A Mat. RACSAM 115 (2021), Paper No. 169.
  7. [7]A. Dujella and J. C. Peral, Elliptic curves with torsion group ℤ/8ℤ or ℤ/2ℤ × ℤ/6ℤ, in: Trends in Number Theory, Contemp. Math. 649, Amer. Math. Soc., Providence, RI, 2015, pp. 47–62.
  8. [8]A. Dujella and J. C. Peral, Construction of high rank elliptic curves, J. Geom. Anal. 31 (2021), 6698–6724.DOI
  9. [9]A. Dujella and J. C. Peral, An elliptic curve over ℚ(u) with torsion ℤ/4ℤ and rank 6, Rad Hrvat. Akad. Znan. Umjet. Mat. Znan. 28 (2024), 185–192.DOI
  10. [10]N. D. Elkies, Three lectures on elliptic surfaces and curves of high rank, arXiv:0709.2908 (2007).arxiv.org/abs/0709.2908
  11. [11]I. Gusić and P. Tadić, Injectivity of the specialization homomorphism of elliptic curves, J. Number Theory 148 (2015), 137–152.arxiv.org/abs/1409.7189
  12. [12]T. Hadano, Elliptic curves with a torsion point, Nagoya Math. J. 66 (1977), 99–108.DOI
  13. [13]E. Kani and M. Rosen, Idempotent relations and factors of Jacobians, Math. Ann. 284 (1989), 307–327.DOI
  14. [14]F. Khoshnam and D. Moody, High rank elliptic curves with torsion ℤ/4ℤ induced by Kihara’s elliptic curves, Integers 16 (2016), Paper A70.DOI
  15. [15]S. Kihara, On the rank of elliptic curves with a rational point of order 4, Proc. Japan Acad. Ser. A Math. Sci. 80 (2004), 26–27.DOI
  16. [16]D. S. Kubert, Universal bounds on the torsion of elliptic curves, Proc. London Math. Soc. (3) 33 (1976), 193–237.
  17. [17]J. Kulesz, Families of elliptic curves of high rank with nontrivial torsion group over ℚ, Acta Arith. 108 (2003), 339–356.
  18. [18]O. Lechacheux, Rang de courbes elliptiques, Acta Arith. 109 (2003), 131–142.DOI
  19. [19]A. J. MacLeod, A simple method for high-rank families of elliptic curves with specified torsion, arXiv:1410.1662 (2014).arxiv.org/abs/1410.1662
  20. [20]B. Mazur, Modular curves and the Eisenstein ideal, Publ. Math. IHÉS 47 (1977), 33–186.DOI
  21. [21]P. Monsky, Generalizing the Birch–Stephens theorem. I. Modular curves, Math. Z. 221 (1996), 415–420.
  22. [22]P. J. Oliver, Classical Invariant Theory, London Math. Soc. Student Texts 44, Cambridge Univ. Press, 1999.
  23. [23]The PARI Group, PARI/GP version 2.17.2, Univ. Bordeaux, https://pari.math.u-bordeaux.fr/.
  24. [24]M. Schütt and T. Shioda, Mordell–Weil Lattices, Ergeb. Math. Grenzgeb. (3) 70, Springer, Singapore, 2019.DOI
  25. [25]J. Woo, Arithmetic of elliptic curves and surfaces: descent and quadratic sections, PhD thesis, Harvard University, 2010.

Paper details

vlnc:W_qafq59ux
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

AI systems

View statement
Looking for related papers…
Contents