Frey Representations in Diophantine Analysis

• Frey representations are engineered from Diophantine solutions, characterized by controlled ramification and conductor properties rather than the solution size.
• They manifest classically as mod-p representations on Frey elliptic curves and in higher-dimensional settings as constituents from Jacobians or GL₂-type abelian varieties.
• Recent frameworks extend these concepts to hypergeometric motives and function-field analogues, enhancing modular methods and level lowering techniques in Diophantine analysis.

Frey representations are Galois or monodromy representations engineered from a putative Diophantine solution so that their ramification, reduction type, and conductor are controlled primarily by the signature and coefficients of the equation, rather than by the size of the solution. In the classical setting they are the mod-$p$ representations on the $p$-torsion of a Frey elliptic curve; in Darmon’s framework they are residual finite-field-valued representations over $K(t)$ with prescribed three-point projective inertia; and in more recent work they also appear as $2$-dimensional constituents cut out from Jacobians or abelian varieties of $\mathrm{GL}_2$-type, as well as in function-field analogues of Frey–Mazur rigidity (Madriaga et al., 2024, Billerey et al., 2016, Azon, 19 Mar 2025).

1. Definitions and range of usage

The literature represented here uses the term in more than one sense. In the narrowest and historically most familiar usage, a Frey representation is the residual mod-$p$ Galois representation

$$\bar\rho_{E,p}: \operatorname{Gal}(\overline K/K)\to \operatorname{GL}_2(\mathbf F_p)$$

attached to a Frey elliptic curve $E$. In Darmon’s generalized Fermat program, the term is more structural: a Frey representation is a continuous representation

$$\rho_t:\operatorname{Gal}(\overline{K(t)}/K(t))\to \operatorname{GL}_2(\mathbb F)$$

whose geometric projectivization is unramified outside $\{0,1,\infty\}$ and whose inertia images at those points have prescribed orders $p$0 (Madriaga et al., 2024). In higher-dimensional variants, the relevant representation is the $p$1-dimensional $p$2-adic or residual constituent cut out by real multiplication from the Tate module of a Frey abelian variety or hyperelliptic Jacobian (Billerey et al., 2016, Azon, 19 Mar 2025).

Context	Geometric source	Representation used
Classical modular method	Frey elliptic curve	$p$3 on $p$4
Darmon’s generalized Fermat program	Family over $p$5	Residual $p$6 with three-point ramification
Higher-dimensional Frey objects	$p$7-type abelian variety or Jacobian	$p$8-dimensional $p$9 or $K(t)$0
Function-field analogues	$K(t)$1-torsion local system or monodromy	$K(t)$2-representation with Frey–Mazur-type rigidity

This terminological breadth matters. A Frey representation is not intrinsically tied to elliptic curves over $K(t)$3, and it is not always a global $K(t)$4-adic representation in characteristic zero. In Darmon’s sense it is fundamentally residual and geometric over a one-variable function field; in the hyperelliptic and motive-based refinements, it is often realized as the reduction of a compatible system (Madriaga et al., 2024).

2. Classical elliptic-curve constructions

The basic classical pattern begins with a putative nontrivial solution of a Fermat-type equation and attaches an elliptic curve whose discriminant contains a large $K(t)$5-power while the conductor remains comparatively small. In the standard Fermat setting, a classical model is

$K(t)$6

and in the generalized $K(t)$7 setting the same object appears, after a suitable specialization of the Legendre family, as

$K(t)$8

for a solution of $K(t)$9 (Madriaga et al., 2024).

The decisive arithmetic feature is the mismatch between discriminant and conductor. The discriminant carries a large $2$0-power contribution from the putative solution, while local reduction is often only multiplicative at the relevant odd primes, so the conductor remembers mainly the squarefree support of the coefficients and variables. This is what makes Ribet-style level lowering applicable in the first place. In the totally real setting, irreducibility of the residual mod-$2$1 representation is a nontrivial input; Freitas and Siksek formulate this through isogeny signatures $2$2, twisted norms

$2$3

and explicit resultant conditions involving Frobenius polynomials (Freitas et al., 2013).

Classical practice also includes multi-Frey constructions. For equations of signature $2$4, Billerey, Dieulefait, and Freitas attach several distinct Frey elliptic curves to the same primitive solution by exploiting the factorization of $2$5 over totally real subfields of $2$6. Different Frey curves give different conductor exponents at $2$7 and $2$8, and therefore different lowered levels; the combined information can eliminate cases that no single curve can handle (Freitas, 2012).

3. Darmon’s framework and generalized Fermat equations

Darmon’s formulation abstracts the modular-method input away from any one elliptic curve. A Frey representation over $2$9 is required to satisfy two geometric conditions on the restriction to $\mathrm{GL}_2$0: trivial determinant and irreducibility, together with projective ramification only at $\mathrm{GL}_2$1, where the inertia images have orders $\mathrm{GL}_2$2 respectively (Madriaga et al., 2024). A notable feature is that $\mathrm{GL}_2$3 depends only on the exponent triple $\mathrm{GL}_2$4, not on an individual solution; a putative solution enters only through specialization of $\mathrm{GL}_2$5.

Recent work recasts this construction in the language of hypergeometric motives. For suitable hypergeometric parameters $\mathrm{GL}_2$6, one obtains a geometric representation

$\mathrm{GL}_2$7

whose local monodromy at $\mathrm{GL}_2$8 is prescribed by explicit matrices $\mathrm{GL}_2$9. After specialization at $p$0, the resulting compatible family has Frobenius traces given by finite hypergeometric sums and inertia controlled by the valuations of $p$1 and $p$2; in this way hypergeometric motives provide characteristic-zero lifts of Darmon’s residual Frey representations (Madriaga et al., 2024).

The same program now includes genuinely higher-dimensional Frey objects. For the equation $p$3, the relevant Frey object in Billerey–Chen–Dieulefait–Freitas is not an elliptic curve but an abelian variety $p$4 over $p$5, of dimension $p$6, with

$p$7

Because it is of $p$8-type, each prime $p$9 of $$\bar\rho_{E,p}: \operatorname{Gal}(\overline K/K)\to \operatorname{GL}_2(\mathbf F_p)$$0 yields a $$\bar\rho_{E,p}: \operatorname{Gal}(\overline K/K)\to \operatorname{GL}_2(\mathbf F_p)$$1-dimensional representation

$$\bar\rho_{E,p}: \operatorname{Gal}(\overline K/K)\to \operatorname{GL}_2(\mathbf F_p)$$2

which is the Frey representation used in the modular argument (Billerey et al., 2016).

A further unification is provided by effective versions of Darmon’s program. For fixed $$\bar\rho_{E,p}: \operatorname{Gal}(\overline K/K)\to \operatorname{GL}_2(\mathbf F_p)$$3, the hyperelliptic families for signatures $$\bar\rho_{E,p}: \operatorname{Gal}(\overline K/K)\to \operatorname{GL}_2(\mathbf F_p)$$4 and $$\bar\rho_{E,p}: \operatorname{Gal}(\overline K/K)\to \operatorname{GL}_2(\mathbf F_p)$$5 are organized into a common framework over $$\bar\rho_{E,p}: \operatorname{Gal}(\overline K/K)\to \operatorname{GL}_2(\mathbf F_p)$$6, with Jacobians of genus $$\bar\rho_{E,p}: \operatorname{Gal}(\overline K/K)\to \operatorname{GL}_2(\mathbf F_p)$$7 carrying real multiplication by $$\bar\rho_{E,p}: \operatorname{Gal}(\overline K/K)\to \operatorname{GL}_2(\mathbf F_p)$$8. Their $$\bar\rho_{E,p}: \operatorname{Gal}(\overline K/K)\to \operatorname{GL}_2(\mathbf F_p)$$9-adic Tate modules split into $E$0-dimensional pieces, which are the actual Frey representations used for modularity, irreducibility, level lowering, and explicit elimination (Azon, 19 Mar 2025).

4. Local invariants, conductors, and representation-theoretic control

The conductor is the principal local invariant of a Frey representation in the modular method, because it determines the level after modularity and before elimination. For hyperelliptic Frey representations, odd conductor exponents can be computed systematically via cluster pictures. A universal biparametric family

$E$1

contains most known Frey hyperelliptic curves for signatures $E$2, $E$3, and $E$4, and its odd local conductor is computed uniformly. For the associated $E$5-dimensional Jacobian representation one has the exact relation

$E$6

which turns cluster-theoretic conductor computations for the curve into conductor computations for the Frey representation itself (García et al., 27 Mar 2025).

The prime $E$7 is more delicate. A dedicated analysis of hyperelliptic models at residue characteristic $E$8 computes the conductor exponent at $E$9 for several important families. For even-degree Frey representations of signature $$\rho_t:\operatorname{Gal}(\overline{K(t)}/K(t))\to \operatorname{GL}_2(\mathbb F)$$0, the conductor exponent of $$\rho_t:\operatorname{Gal}(\overline{K(t)}/K(t))\to \operatorname{GL}_2(\mathbb F)$$1 at the prime above $$\rho_t:\operatorname{Gal}(\overline{K(t)}/K(t))\to \operatorname{GL}_2(\mathbb F)$$2 is $$\rho_t:\operatorname{Gal}(\overline{K(t)}/K(t))\to \operatorname{GL}_2(\mathbb F)$$3 if $$\rho_t:\operatorname{Gal}(\overline{K(t)}/K(t))\to \operatorname{GL}_2(\mathbb F)$$4 and $$\rho_t:\operatorname{Gal}(\overline{K(t)}/K(t))\to \operatorname{GL}_2(\mathbb F)$$5, $$\rho_t:\operatorname{Gal}(\overline{K(t)}/K(t))\to \operatorname{GL}_2(\mathbb F)$$6 if $$\rho_t:\operatorname{Gal}(\overline{K(t)}/K(t))\to \operatorname{GL}_2(\mathbb F)$$7 and $$\rho_t:\operatorname{Gal}(\overline{K(t)}/K(t))\to \operatorname{GL}_2(\mathbb F)$$8, and $$\rho_t:\operatorname{Gal}(\overline{K(t)}/K(t))\to \operatorname{GL}_2(\mathbb F)$$9 if $\{0,1,\infty\}$0. For the new even-degree $\{0,1,\infty\}$1 family, the same local exponent is likewise determined completely in terms of $\{0,1,\infty\}$2 and $\{0,1,\infty\}$3 (Chen et al., 28 Sep 2025).

Irreducibility is the second indispensable input. Over totally real Galois fields, Freitas–Siksek give a practical criterion in terms of semistability at primes above $\{0,1,\infty\}$4, twisted norms of units, and Frobenius resultants. In higher dimension, Billerey–Chen–Dieulefait–Freitas obtain an irreducibility theorem for residual representations of $\{0,1,\infty\}$5-type abelian varieties over totally real fields by combining semistability, inertial exponents, potential good reduction at an auxiliary prime, and a resultant bound involving the possible Frobenius traces (Freitas et al., 2013, Billerey et al., 2016).

In effective hyperelliptic implementations, the local geometry also determines the inertial type. The RM Jacobians occurring in the common $\{0,1,\infty\}$6/$\{0,1,\infty\}$7 framework have only good, totally toric, or totally unipotent reduction. This trichotomy feeds directly into tame conductor exponents, modularity, and level lowering over $\{0,1,\infty\}$8 (Azon, 19 Mar 2025).

5. Frey–Mazur philosophy and function-field analogues

A different but closely related line of work studies rigidity of mod-$\{0,1,\infty\}$9 representations rather than the direct construction of Frey objects from Diophantine equations. For a non-isotrivial family $p$00 of elliptic curves over a complex quasiprojective curve, the $p$01-torsion local system yields a monodromy representation

$p$02

Bakker and Tsimerman prove that, for bounded gonality of $p$03 and sufficiently large $p$04, this $p$05-torsion local system determines the family up to isogeny. Their reformulation through the moduli surface $p$06 makes the connection to Frey–Mazur explicit: curves of bounded complexity in $p$07 must be Hecke, so coincidences of mod-$p$08 representations come from isogenies (Tsimerman et al., 2014).

A quaternionic analogue replaces elliptic curves by “fake elliptic curves,” namely abelian surfaces with quaternionic multiplication. In that setting, low-genus curves in the moduli surface $p$09 factor through Hecke curves for $p$10 sufficiently large, and the $p$11-torsion local system determines the $p$12-isogeny class (Bakker et al., 2013).

Litt’s function-field results push the philosophy further. For arithmetic $p$13-adic representations of geometric fundamental groups, the set of semisimple arithmetic points has no limit points, and for a fixed geometric irreducible representation $p$14 in characteristic $p$15 there is an explicit constant $p$16 such that

$p$17

This is explicitly described as a weak Frey–Mazur statement: it gives local uniqueness in representation space, not a global classification of geometric objects by residual torsion data (Litt, 2018).

These function-field results are not Frey representations in Darmon’s sense. Their relevance is structural: they articulate the rigidity principle that underlies the modular method, namely that large-$p$18 torsion data should determine the isogeny class except on special loci.

6. Applications, effectiveness, and historiography

Frey representations are now used in a wide spectrum of Diophantine problems. In Lucas sequences, a hypothetical perfect power $p$19 is converted into a generalized Fermat-type equation, and a Frey elliptic curve is attached so that modularity and level lowering reduce the problem to a finite list of newforms and then to explicit bounds on $p$20 and $p$21 (Silliman et al., 2013). In the Lebesgue–Nagell equation, semistable Frey–Hellegouarch curves with rational $p$22-torsion, together with modularity and refined sieving, reduce a vast search space to a finite set of candidate triples $p$23 (Bennett et al., 2021). In signature $p$24, multi-Frey constructions over totally real subfields of $p$25 make it possible to combine several residual representations with different local conductors and lowered levels (Freitas, 2012).

A major recent development is effectiveness. For signatures $p$26 and $p$27, a common hyperelliptic framework now supports modularity, irreducibility, level lowering, and a Magma implementation of the elimination step. In the worked cases $p$28, the resulting Hilbert newform computations solve several infinite families of equations asymptotically (Azon, 19 Mar 2025).

There are also representation-theoretic inputs aimed specifically at controlling residual images. Najman studies when an elliptic curve with rational $p$29-invariant acquires an $p$30-isogeny over $p$31, equivalently when $p$32 becomes reducible. This classification is used in modularity arguments for Frey hyperelliptic Jacobians in signature $p$33 (Najman, 2023).

Several limitations remain explicit in the literature. Hypergeometric-motive constructions provide characteristic-zero lifts and often modularity input, but wild ramification at primes dividing the denominators of the parameters is not understood in full generality, and large residual image statements are still mostly unavailable; as a consequence, full level lowering remains conditional in many general families (Madriaga et al., 2024). The same papers make clear that conductor computations at $p$34 are especially delicate, which explains the continuing emphasis on specialized local analyses (Chen et al., 28 Sep 2025).

A final historiographical caution concerns claims about origins. An arXiv record titled "The origin of the Frey elliptic curve in a too narrow margin" asserts a connection with Diophantus’ “double equations” and an elementary proof of Fermat’s Last Theorem, but the supplied source contains no mathematical text, so no paper-specific historical or mathematical reconstruction is possible from that record (Ossicini, 2016). This leaves the historical genesis of Frey representations, in the strict documentary sense, grounded instead in the standard modular-method tradition and its later generalizations.