A pattern with exceptions
The preprint's clearest result is a split in the way eight Platonic cases are classified. The authors report that eight cases are modular for non-congruence subgroups, while the cases carrying the fibre label IV are outliers. In plain language, modularity here is a classification drawn from the subgroup pattern of monodromy matrices, which record how solutions behave when they are analytically continued along paths that avoid singularities.
This is a result in pure mathematics built around three reflection groups: T for the tetrahedron, O for the octahedron and I for the icosahedron. From them, the paper constructs rational elliptic surfaces with four reduced singular fibres and examines their algebraic-geometric, Picard-Fuchs, sequence, Laurent-polynomial and arithmetic properties.
From symmetry to surfaces
The construction starts with invariants, quantities that remain unchanged under the group symmetries. An invariant hypersurface written as 0 = S^2 - F(P,Q,R) is treated as an elliptically fibred threefold over the P-Q plane. Specializing the base variables then yields elliptic curves, giving the family its surface-by-surface form.
The singular fibres retain the Platonic pattern. In the tetrahedral case, the finite fibres are labelled I2/I3/I3 and the fibre at infinity is IV. The octahedral case has I2/I3/I4 with III at infinity, while the icosahedral case has I2/I3/I5 with II at infinity. These labels summarize the distinct fibre configurations reported for the three families.
Related surfaces are connected by isogenies, the degree-labelled algebraic links used in the paper. One reported link has degree 3 and joins 233IV to 116IV. Another has degree 2 and joins 234III to 126III. The links add a second layer of structure to the family, alongside the fibre classifications.
Equations become sequences
To study the resulting periods, the authors determine special Heun-type Picard-Fuchs operators and derive a two-term recursion for their coefficients. That recursion produces Apéry-like integer sequences. After critical scaling, the paper reports eight differential operators and associated period sequences.
One part of the study tests whether the integer sequences can be represented by Laurent polynomials, expressions that allow both positive and negative powers of variables. The genus-1 search uses a Newton-polygon ansatz, varies integer coefficients over 16 reflexive polygons, and compares the constant terms of polynomial powers with the sequence coefficients. This makes the search a direct match between algebraic expressions and the reported periods.
The search also reveals a boundary. The preprint states that O(2) and I(2) have no genus-1 Laurent-polynomial representation, while genus-3 and genus-10 representations were found for those cases and for T(3), T(2) and O(4). The authors present this as a construction result, so the broader question of genus-1 representations for O(2) and I(2) remains open.
Where the arithmetic resists
The p-adic calculations expose a separate obstacle. For the critically scaled Platonic operators, the diagonal U(0) ansatz fails to produce the p-adic expansion and cannot evaluate U([t0]) modulo p^N when N is 2 or larger. A more general ansatz is therefore required.
For a multiplicative fibre of type I_n at the origin, the paper gives the logarithmic term x_p = (p - 1)log_p(c)/n. Here c is the TPF constant and n is the fibre width. The formula is the paper's explicit link between the fibre data and the logarithmic part of the expansion.
The monodromy calculations come with an explicit caveat. They use numerical analytic continuation along paths avoiding singularities, SAGE's ORE-algebra routines and certified precision. The entries are then identified as putative exact matrices, a qualification that should stay in view when reading the subgroup classifications.
On the numerical side, some Apéry-like limits are identified as logarithmic terms, but the O(2) and I(2) constants remain unidentified despite high-precision computation. The paper therefore leaves part of the sequence story unresolved.
An unfinished map
Taken together, the results describe a structured but unfinished mathematical family: eight reported cases fit a non-congruence modular pattern, IV cases sit outside it, simple diagonal Frobenius calculations fail, and two Apéry-like limits remain unidentified. The supplied result is therefore a partial classification rather than a completed account.
The document is an arXiv version-1 preprint dated 26 Aug 2026. The authors acknowledge support from the Deutsche Forschungsgemeinschaft, or DFG, under Project-ID 444845124 and TRR 326.
Paper data and sources
Original title: The platonic elliptic surfaces
Authors: Nutsa Gegelia, Duco van Straten
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text