Splitting Field of an Elliptic Surface

• Splitting fields are the minimal number field extensions that allow all Mordell–Weil group generators of an elliptic surface’s generic fiber to be defined over the extended function field.
• Explicit computations use fundamental polynomials and symbolic tools (Maple, PARI/GP) to determine the splitting fields and structure of elliptic surfaces.
• The Galois groups acting on the splitting fields mirror the lattice symmetries of the Mordell–Weil group, revealing deep connections to modular and Kummer constructions.

The splitting field of an elliptic surface is a central concept in the arithmetic geometry of elliptic fibrations, encapsulating the minimal number field extension required to realize all Mordell–Weil group generators of the generic fiber over the function field. For elliptic $K3$ surfaces and the broader family of elliptic surfaces with generic fiber of the form $y^2 = x^3 + t^m + 1$, the splitting field governs the field of definition of sections and the action of the absolute Galois group on the Mordell–Weil lattice structure. Recent research has yielded explicit computations, diverse methods, and structural theorems for splitting fields in various families, especially those built via modular or Kummer constructions.

1. Definition and Structural Framework

Let $k \subset \mathbb{C}$ be a number field, and let $\mathcal{E}/k(t)$ denote the generic fiber of an elliptic surface $\pi\colon \mathcal{E} \to \mathbb{P}_k^1$. For any field extension $K/k$, $\mathcal{E}(K(t))$ is the group of $K(t)$-rational points—i.e., $K$-rational sections of $\pi$—which is finitely generated (by the Silverman–Tate theorem). The splitting field $\mathcal{K}/k$ of $\mathcal{E}$ is the minimal finite extension for which

$$\mathcal{E}(\mathbb{C}(t)) = \mathcal{E}(\mathcal{K}(t)),$$

i.e., the Mordell–Weil group over $\mathbb{C}(t)$ descends to $\mathcal{K}(t)$. Equivalently, the splitting field is the fixed field of the kernel of the Galois representation

$$\rho:\operatorname{Gal}(\mathbb{C}/k)\to \operatorname{Aut}_{\mathbb{Z}}\left(\mathcal{E}(\mathbb{C}(t))/\text{tors}\right).$$

The Galois group $$\operatorname{Gal}(\mathcal{K}/k) \cong \operatorname{im}\rho$$ is finite and reflects the symmetries of the Mordell–Weil group, acting through isometries on the canonical height pairing lattice (Salami et al., 2022, Salami et al., 18 Dec 2025).

2. Determination of Splitting Fields: Computational Approach

Modern computations of splitting fields for explicit families, notably for $y^2 = x^3 + t^m + 1$, proceed via the construction of a fundamental polynomial $\Phi_m(u)$ associated with "minimal" (height-minimal norm) sections. This polynomial encodes the obstruction to rationality of sections and is often determined by:

• Forming ansatz polynomials $x(t)$, $y(t)$ for section coordinates.
• Solving the Weierstrass equation and analyzing the resulting polynomial system.
• Identifying the minimal polynomial whose splitting field over $\mathbb{Q}$ defines all section fields of definition.

This process, accelerated by symbolic computation packages (e.g., Maple's PolynomialIdeals and PARI/GP's polcompositum/polredbest), allows the explicit construction of the splitting field $K_m$ as the Galois closure of $\mathbb{Q}$ in which $\Phi_m$ splits. For higher-rank surfaces (e.g., with $m=5,6,9,12,360$), the degree of $K_m$ grows rapidly, and the defining polynomials reach degrees in the dozens, hundreds, or thousands (Salami, 31 Dec 2025, Salami et al., 18 Dec 2025).

3. Explicit Results for Specific Families

Shioda’s Family $y^2 = x^3 + t^m + 1$

For $1 \le m \le 12$, each surface $E_m$ has a splitting field $K_m$ uniquely determined by the factorization of $\Phi_m$, with Galois groups and generators detailed as follows (Salami et al., 18 Dec 2025):

$m$	Rank $r_m$	Lattice $T_m^*$	$[K_m:\mathbb{Q}]$	Structure of $K_m$
2	2	$A_2^*$	2	$\mathbb{Q}(\zeta_3)$
3	4	$D_4^*$	6	$\mathbb{Q}(\zeta_3,\, 2^{1/3})$
4	6	$E_6^*$	16	$\mathbb{Q}(\zeta_{12},\, \alpha_1)$, $\alpha_1^4+24\alpha_1^2-48=0$
5	8	$E_8^*$	120	$\mathbb{Q}(\zeta_{30},\,(60v_1)^{1/30})$, $v_1$ as in (Salami et al., 18 Dec 2025)
6	8	$E_8^*$	12	$\mathbb{Q}(\zeta_{12},\, 2^{1/3})$
8	6	$E_6^*[2]$	16	$K_8 = K_4$
9	10	see (Salami et al., 18 Dec 2025)	54	Splitting field of a degree-240 polynomial; reduced degree 54
10	10	$E_8^*[2]$	120	$K_{10} = K_5$
12	16	$E_8^*[3]+\cdots$	96	Minimal degree-96 field; minimal polynomial in (Salami et al., 18 Dec 2025)

In every case, a basis of $r_m$ independent points in $E_m(K_m(t))$ is constructed, whose canonical heights form a lattice isomorphic to $T_m^*$.

K3 Surfaces $y^2 = x^3 + t^n + t^{-n}$

For $1 \leq n \leq 6$, Salami–Zargar analyze the family

$\mathcal{E}_n: y^2 = x^3 + t^n + t^{-n},$

computing the explicit splitting fields $\mathcal{K}_n$ and generators via pullback from a rational elliptic surface and subsequent resolution of a “fundamental polynomial.” The structure is parallel to that of the pure-$t^m$ family, with degrees and radicals increasing rapidly with $n$ (Salami et al., 2022).

Extremely High-Rank Examples

For the surface $Y^2 = X^3 + t^{360} + 1$, the splitting field $\mathcal{K}$ has degree $1728 \cdot 5760$ and is explicitly computed as the compositum of two number fields defined by irreducible polynomials $F_{1728}(x)$ and $G_{5760}(x)$ (Salami, 31 Dec 2025).

4. Galois Structure and Arithmetic of Splitting Fields

The Galois group $\operatorname{Gal}(K_m/\mathbb{Q})$ acts by permuting the roots of the fundamental polynomial, and thus the corresponding minimal sections of the Mordell–Weil group. Key properties:

• For $m=2$, the Galois group is $C_2$ (cyclic of order 2), acting on the two nontrivial $2$-torsion points.
• For $m=3$, it is $S_3$, permuting the cube roots of $2$ and the third roots of unity.
• For $m=4$, the Galois group is of order $16$ and constructed from the automorphisms of $\zeta_{12}$ and $\alpha_1$.
• For $m=5,6,9,12$, the structure is increasingly complex (e.g., order 120 for $m=5$), often realized as subgroups of the Weyl group $W(E_8)$ or products of cyclotomic and Kummer extensions.
• For high-rank surfaces (e.g., $m=360$), the Galois group is a finite solvable extension built from cyclic and abelian subgroups, but the precise internal composition is complicated and not always decomposed explicitly (Salami, 31 Dec 2025).

The splitting field is always Galois and its ramification is tightly controlled by the arithmetic of $m$ and the cyclotomic units present in the minimal polynomial.

5. Verification and Lattice-Theoretic Properties

Verification that the computed field $\mathcal{K}$ is indeed the splitting field is achieved by:

• Computation of the height-pairing matrix of the constructed sections using Shioda’s canonical formula and comparison with the predicted discriminant of the Mordell–Weil lattice.
• Specialization of the parameter $t$ to rational points $t_0$, confirming the independence of the images via height-pairing and Mordell-Weil group structure computations.
• Explicit symbolic computation (via software) for elimination ideals, resultant calculations, and minimal polynomial factorization.

For high-rank cases, the Mordell–Weil lattice decomposes as an orthogonal sum of lattices corresponding to sub-surfaces (rational or $K3$), and the global height matrix is block-diagonal, maintaining positivity of the global discriminant (Salami, 31 Dec 2025).

6. Splitting Fields in Modular and Double-Cover Constructions

For $K3$ surfaces arising as double covers of extremal rational elliptic surfaces, the splitting field is described as follows (Cantoral-Farfán et al., 2020):

• The field of definition of full Mordell–Weil group generators is typically the compositum $k_{R_i,\tau_i}$ of the field splitting all reducible fibers and torsion, and the field generated by the branch-points of the base change.
• For these families, $[k_{MW}:k] \leq 4$, with all necessary extensions being at most quadratic in each variable, and full splitting fields constructed by explicit Galois theory and lattice decompositions.

This setting highlights the general principle that for many explicit elliptic surfaces, especially those built from modular or Kummer structures, the splitting field is tightly governed by cyclotomic, Kummer, and radical extensions, and can be computed systematically by analysis of the low-height sections and their Galois orbits.

7. Patterns, Complexity, and Ramification

The degree of the splitting field $[K_m:\mathbb{Q}]$ grows rapidly with $m$—for $m=5$, it is $120$; for $m=12$, it reaches $96$; for $m=360$, the degree is in the thousands. The ramification locus of $K_m$ is controlled by primes dividing $m$ and the orders of roots of unity adjoined (e.g., $\zeta_{30}$ or $\zeta_{12}$) (Salami et al., 18 Dec 2025). The presence of base-change and lattice-symmetry phenomena allows for nested structure in these fields: when $m|m'$, $K_m\subset K_{m'}$ and there is a corresponding lattice embedding.

A plausible implication is that similar computational methodologies extend to other large-scale families where the Mordell–Weil group admits a modular, lattice-theoretic, or toroidal description, subject to advances in computational number theory for factoring high-degree polynomials and computing with large cyclotomic/Kummer extensions.

---

References:

• "Generators and splitting fields of certain elliptic K3 surfaces" (Salami et al., 2022)
• "The splitting fields and Generators of Shioda's elliptic surfaces $y^2=x^3 +t^{m} +1$ (I)" (Salami et al., 18 Dec 2025)
• "The splitting field and generators of the elliptic surface $Y^2=X^3 +t^{360} +1$" (Salami, 31 Dec 2025)
• "Fields of definition of elliptic fibrations on covers of certain extremal rational elliptic surfaces" (Cantoral-Farfán et al., 2020)