Fay's trisecant identity¶

A three-term relation among curve-theta products at four-point Abel shifts, weighted by prime forms on the same Riemann surface.

Version: v1 · 2026-10-17 · History
Domain-specific #: 14152
Domain group: Formal Sciences
Origin domain: Mathematics
Subdomains: Algebraic Geometry, Theta Functions → Mathematics
Aliases: Fay identity, Fay trisecant formula

Core Idea¶

Fay's trisecant identity is a three-term equality for the Riemann theta function of a compact Riemann surface. Choose four points on the curve. Abel integrals between them translate the theta argument, while prime forms evaluated at pairs of the same points weight products of two theta values. The two weighted products on one side equal a third on the other. The identity is parameterized: the curve, the points and the theta argument can vary while the arrangement of shifts and factors remains the same.^[1]^[2]

One precise convention is that of the NIST Digital Library of Mathematical Functions. Write \(A_{ij}=\int_{P_i}^{P_j}\boldsymbol\omega\), \(E_{ij}=E(P_i,P_j)\), and \(\theta\) for the theta function of the curve's period matrix. With the integration paths chosen compatibly, its equation 21.7.10 is^[1]

\[ E_{32}E_{14}\,\theta(\mathbf z+A_{13})\theta(\mathbf z+A_{24}) +E_{31}E_{42}\,\theta(\mathbf z+A_{23})\theta(\mathbf z+A_{14}) =E_{12}E_{34}\,\theta(\mathbf z)\theta(\mathbf z+A_{13}+A_{24}). \]

This displayed arrangement fixes one source's point order and signs; rearranged versions must be converted as whole equations. “Trisecant” names a geometric reading through lines meeting a Jacobian's Kummer variety.^[3] That reading is a consequence and diagnostic use of the theta relation, not a substitute definition of its operands.

Structural Signature¶

Four roles identify an instance:

Curve and theta carrier. A compact Riemann surface supplies normalized holomorphic differentials, its period matrix and the associated Jacobian theta function. The same typed curve data must underlie the other roles.^[1]
Four points and compatible Abel shifts. Four variable points on that curve determine the integration paths and translations of the theta argument. Arbitrary, unrelated shifts do not fill this role.^[1]^[2]
Prime-form pair weights. The curve's prime form evaluated on specified point pairs supplies the coefficients. In a genus-one rewriting these may be absorbed into Jacobi-theta or Weierstrass factors, but an unexplained deletion changes the equation.^[1]^[4]
Three-term bilinear closure. Each term contains a product of two theta values with its pair weights. The first two products add to the third under one consistent convention; a single theta value or another addition theorem is a different relation.^[1]^[2]

The integration convention is part of stating these four roles correctly, especially the Abel shifts. It is not a fifth independent mathematical mechanism. Klein, Korotkin and Shramchenko explicitly restrict their integration contours after writing their version of the formula.^[2]

What It Is Not¶

The identity is not simply the Jacobian variety. A Jacobian is the abelian variety attached to a curve; the identity is a relation among functions and factors constructed from the curve. It is also not every theta-function addition rule. The four curve points, Abel shifts, prime-form factors and three-term pairing distinguish this named rule from a formula that merely contains theta symbols.^[1]

Nor is a Kummer trisecant the same object as the equation. Fay's formula implies trisecants for a curve Jacobian. The converse characterization that Krichever proves concerns an indecomposable principally polarized abelian variety under its trisecant condition.^[3] The geometric theorem does not authorize applying the curve formula to every abelian variety or identifying a line in projective space with the written theta identity.

Klein, Korotkin and Shramchenko describe deriving finite-gap solutions of KP-type integrable systems from Fay's identity.^[2] A KP-type equation without the four curve-point operands is not, merely by resemblance, another literal instance of the formula displayed above. The connection needs its stated substitutions or limiting argument rather than a shared three-term shape.

Scope of Application¶

The literal identity applies to compact Riemann surfaces and their curve-derived theta and prime-form data. It spans unlike genera, not only one named curve. DLMF states the relation for genus \(g\); Klein, Korotkin and Shramchenko use its full prime-form expression on higher-genus hyperelliptic curves. The curve points and theta argument are variables, not a fixed historical choice made by Fay.^[1]^[2]

Genus one has an elliptic realization. Boutillier, Cimasoni and de Tilière call their Jacobi-theta equation a genus-one Fay or Weierstrass identity; their Corollary 10 gives a four-point version. Its factors are normalized into a different-looking expression, so the match is by sourced specialization and rewriting, not by visually equating each printed coefficient with the general prime form. A derived telescoping identity helps them construct functions in the kernel of an elliptic Kasteleyn operator for a dimer model.^[4]

In higher-genus hyperelliptic geometry, the full four-point formula helps derive theta-functional results for the Ernst equation, together with degenerate Fay identities and Rauch variational formulas.^[2] In algebraic geometry, its Kummer-trisecant implication helps study the Schottky question of recognizing Jacobians.^[3] These are uses of one curve identity; neither gives a license to drop the curve, prime form or theorem hypotheses.

Clarity¶

Read each \(A_{ij}\) as an integral from one of the chosen points to another of the curve's vector of holomorphic differentials. It is a theta-argument shift, not a coordinate difference between arbitrary points in an unrelated space. Read \(E_{ij}\) as a factor from the same curve. These roles explain why the coefficients and shifts vary together: changing the curve changes its period matrix, theta function, Abel map and prime form as one package.^[1]

The equation has three terms, even though four curve points supply its parameters. “Four-point” counts inputs, while “three-term” counts the weighted theta products. The two statements are compatible. An expression with two terms on the left and one on the right is not a four-term identity. Point-order and sign changes can produce equivalent versions, but only if every affected Abel integral and prime-form factor is transformed consistently.^[1]^[2]

Manages Complexity¶

The four-role check separates difficult but different questions. First ask whether the proposed theta data actually come from a compact curve. Then match the four points and their integration convention. Next check the prime-form coefficients or a justified genus-one rewriting. Finally compare the three paired theta products. The elliptic-dimer paper's Kasteleyn-kernel use is an application of a derived identity.^[4] Fay's Kummer-trisecant implication is a geometric consequence; Krichever proves the bounded converse characterization.^[3] Neither can repair a failure to match the formula's operands.

This order is useful because a Kummer line, a Jacobi-theta calculation and an integrable-system formula may all be described using the word “Fay.” The name is informative only after the equation's operands have been typed. For example, the elliptic dimer paper uses a telescoping consequence in its operator calculation; treating that operational corollary as if it were the unmodified general four-point formula would conceal a real derivation step.^[4]

Abstract Reasoning¶

The invariant is a constrained relation among evaluations, not one numerical value. It says that for each admissible curve and four-point configuration, certain paired theta evaluations and prime-form coefficients cannot vary independently: the third weighted product is fixed by the first two. That organization survives a change from a one-dimensional elliptic theta argument to a multidimensional theta argument on a higher-genus curve, although the computational expressions change.^[1]^[4]^[2]

The geometric reading is an inference from the identity under Jacobian data. Krichever recounts how Fay's formula produces Kummer trisecants when the period matrix comes from a smooth curve, then proves a converse characterization under its stated conditions. The inference is strong precisely because the theta relation carries curve-specific structure; replacing its premise with “some three-term equation” would destroy the reasoning.^[3]

Knowledge Transfer¶

Within curve and theta geometry, transfer means carrying the test, not copying a formula's typography. A proposed higher-genus case can be checked against the four roles, and an elliptic form can be recognized after an explicit normalization. The DLMF and Klein displays label their four points differently. DLMF requires each integration path to be identical across the differential components, while Klein and colleagues specify contours avoiding the canonical basis cycles.^[1]^[2] Copying one display's coefficients into the other's shifts without conversion is a transcription error, not a new instance.

To move into an application, name the extra bridge. The elliptic dimer analysis derives a telescoping corollary and applies it to a Kasteleyn operator.^[4] The hyperelliptic Ernst analysis uses Fay degenerations and variational formulas.^[2] Kummer geometry converts the theta identity into incidence of projective lines with a Kummer variety.^[3] Each bridge preserves a traceable role of the identity while adding conditions the identity alone does not state.

Examples¶

Canonical: elliptic specialization¶

Let the carrier be the elliptic torus \(\mathbb C/\Lambda\) in Boutillier, Cimasoni and de Tilière. Its Jacobi theta function satisfies their Proposition 7, equation (6), which they explicitly call Fay's trisecant or Weierstrass identity in genus one. Corollary 10 rewrites it in a four-point form. The normalized theta factors play the role that prime-form-weighted products play in the general expression; the paper does not invite a term-by-term visual substitution of the two displays.^[4]

Mapped back: curve/theta carrier = elliptic torus and Jacobi theta; four-point Abel-shift role = variable torus points and arguments under the paper's period-lattice convention; weights = its normalized genus-one factors; closure = three paired terms summing to zero. The elliptic Kasteleyn construction is a documented use of a derived form, not what makes the identity true.^[4]

Applied: higher-genus hyperelliptic realization¶

Klein, Korotkin and Shramchenko work with hyperelliptic spectral curves of genus \(g\). Their equation (2.4) explicitly displays the curve's prime forms and theta functions at Abel shifts for four points. They state the contour condition immediately after the equation. They then use the formula and its degenerations, together with Rauch formulas, in their Ernst-equation analysis.^[2]

Mapped back: curve/theta carrier = the hyperelliptic curve, period matrix and multidimensional theta; four-point Abel-shift role = its \(a,b,c,d\) and the compatible contours specified after equation (2.4); weights = paired prime forms \(E\); closure = two weighted theta products equal a third. The Ernst potential is an application of these equations with additional analytic work, not a fourth term in Fay's identity.^[2]

Structural Tensions¶

T1 — General formula versus computable specialization. DLMF gives the general four-point prime-form equation; Boutillier and colleagues give a normalized genus-one form.^[1]^[4] The first makes the curve-level invariant explicit, while the second makes elliptic calculations usable. Demanding identical notation would reject a valid specialization; treating their coefficients as interchangeable without a conversion would assert an unsupported equality. Diagnostic: What normalization or algebraic rewrite connects the candidate's displayed terms to the curve-level equation?

Structural–Framed Character¶

DLMF states the full curve equation, the elliptic-dimer paper gives a genus-one rewriting, and Klein and colleagues use the prime-form expression on higher-genus hyperelliptic curves.^[1]^[4]^[2] On that evidence, the identity is strongly structural within a mathematical frame. Evaluative weight: the equality is descriptive, not a value judgment about a curve or application. Human-practice dependence: people choose notation, paths and convenient normalizations, while mathematical validity requires the stated conditions. Institutional origin: mathematical literature stabilizes the name and preferred formulations; it does not create the equality by convention. Vocabulary travel: “three-term relation” and “weighted products” can describe other structures, but Abel integrals and prime forms keep this named identity in curve theory. Import versus recognition: an unfamiliar curve case can be recognized by matching all four roles; an unrelated bilinear equation is only analogous unless a curve construction is supplied. Its character: a parameterized mathematical identity that is structural across curves and genera, yet framed by curve-derived theta, Abel and prime-form data; its generic equality skeleton alone does not make the named identity a Prime.

Structural Core vs. Domain Accent¶

Structural core: variable inputs are mapped to function arguments and weights, then linked by a fixed equality of products. This makes Fay's identity an exact instance of the broader Mathematical Relation: it has typed quantities, declared parameters and a satisfaction condition. The Prime Relation has additional first-class tuple-set and relational-algebra commitments that the identity does not independently require. A distinct future Prime for weighted bilinear evaluation identities would need independently sourced instances in at least two genuinely noncurve settings that preserve typed input roles, pair weights, three-term closure and a useful derived inference. That future-Prime proposal fails if the commonality is only a three-term equality or if the weights play no constitutive role.

Domain accent: compact Riemann surface, normalized periods, Abel integrals, Jacobian theta, prime form and consistent paths are not replaceable decorations. They determine which substitutions are legal and what the equation asserts. Once those are removed, many unrelated addition theorems share the bare “three-term relation” shape. That bare shape is insufficient to call a case Fay's trisecant identity or to promote this named entry to a cross-domain Prime.^[1]^[2]

This entry is a kind of Mathematical Relation.

Mathematical Relation (Mathematical Relation) is a broader relation type instantiated here: Fay's identity is its curve-theta specialization with a three-term satisfaction condition. The Prime Relation names a broader association pattern but adds tuple-set and relational-algebra roles that are not independent constituents of this equation. The Jacobian Variety is a related curve-associated carrier rather than a genus of equations; its divisor-class group law is not an additional step in the formula itself. Theta Representation concerns a Heisenberg-group representation rather than this four-point identity.

Relationships to Other Abstractions¶

Current abstraction Fay's trisecant identity Domain-specific

Parents (1) — more general patterns this builds on

Fay's trisecant identity is a kind of Mathematical Relation Domain-specific

Fay’s identity is an exact mathematical relation among typed theta evaluations and prime-form weights, specialized by four curve points, Abel shifts and compatible path conditions.

Hierarchy path (1) — routes to 1 parentless root

Fay's trisecant identity → Mathematical Relation

Neighborhood in Abstraction Space¶

Fay's trisecant identity sits in a sparse region of the domain-specific corpus (83^rd percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (2721 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-17

Not to Be Confused With¶

A Jacobian variety. It supplies a geometric carrier; it is not itself Fay's theta/prime-form equation.^[1]
A geometric Kummer trisecant or the trisecant characterization theorem. Fay's identity implies the former in the Jacobian setting; Krichever's converse uses indecomposability and its trisecant hypotheses.^[3]
Any theta addition theorem or arbitrary bilinear identity. Similar notation and term count do not supply the required four-point Abel shifts and pair weights.^[1]
A KP-type or Ernst equation. Klein and colleagues describe the KP-type connection and use Fay's identity with further formulas in their Ernst analysis.^[2] These equations are application targets or derived relatives, whose additional steps must be named.
A literal equation obtained by mixing two presentations. Reordered points, signs and integration conventions have to be converted together.^[1]^[2]

References¶

[1] NIST Digital Library of Mathematical Functions, “§21.7 Riemann Surfaces,” especially §21.7(ii), equations 21.7.7–10 and the paragraph after 21.7.10, https://dlmf.nist.gov/21.7 . Full authoritative formula and surrounding definitions consulted. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r

[2] C. Klein, D. Korotkin and V. Shramchenko, “Ernst equation, Fay identities and variational formulas on hyperelliptic curves,” arXiv:math-ph/0401055v1, abstract, §1 and §2.1 equation (2.4), https://arxiv.org/pdf/math-ph/0401055 . Full original preprint consulted. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q

[3] Igor Krichever, “Characterizing Jacobians via trisecants of the Kummer variety,” Annals of Mathematics 172 (2010), 485–516, abstract and introduction pp. 485–487, https://www.math.columbia.edu/department/krichever/pdfs/2010-2014/2010-CJVTOTKV.pdf . Full original paper consulted. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g

[4] Cédric Boutillier, David Cimasoni and Béatrice de Tilière, “Elliptic dimers on minimal graphs and genus 1 Harnack curves,” author-hosted PDF dated 9 December 2022, §2.3 Proposition 7, equation (6), Corollaries 8–10, and §3, https://www.unige.ch/math/folks/cimasoni/elliptic_dimers.pdf . Full original paper consulted. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j