Key Varieties in Algebraic Geometry

Updated 6 July 2026

Key Varieties are explicit ambient spaces with structured equations, graded rings, and symmetry groups that enable the construction of new algebraic varieties via pullbacks and linear sections.
They are universal construction models in algebraic geometry, illustrated by methods such as diptych varieties, weighted flag formats, and Freudenthal triple system approaches.
Their applications include constructing prime Q-Fano 3-folds, organizing toric and homogeneous ambient models, and facilitating birational transformations in complex algebraic settings.

Searching arXiv for recent and foundational papers on "key varieties" to ground the encyclopedia article. Key varieties are explicitly presented ambient varieties used to construct, characterize, or organize other algebraic varieties. In the literature considered here, the term is applied to affine or projective models with controlled equations, symmetries, and graded structures, from which target varieties are obtained by regular pullback, linear section, weighted complete intersection, or Sarkisov-link constructions. Brown and Reid formulate this role in birational geometry through diptych varieties, weighted flag constructions use “key varieties (Format)” as ambient spaces for polarized $3$-folds, and Takagi’s series uses affine varieties and their weighted projectivizations as key varieties for prime $\mathbb Q$ -Fano $3$-folds of anti-canonical codimension $4$ (Brown et al., 2012, Qureshi, 2014, Takagi, 2024).

1. Terminology and constructional role

In Brown and Reid’s formulation, a key variety is an ambient universal model from which other varieties are obtained by regular pullback, linear section, or specialization of parameters. Their guiding comparison is explicit: diptych varieties are “designed for use as ambient spaces or key varieties in constructing other spaces, much as toric varieties” (Brown et al., 2012). In the weighted-flag setting, the same role is called a “Format”: a weighted flag variety supplies a projective embedding into a weighted projective space, an explicit graded ring with generators and relations, a Hilbert series, and defining equations, after which quasilinear sections produce the required $3$-folds (Qureshi, 2014).

Takagi’s work makes the constructional meaning even more explicit. There, a key variety is an affine variety, or a suitable weighted projectivization of it, such that weighted complete intersections produce prime complex $\mathbb Q$ -Fano $3$-folds of anti-canonical codimension $4$. The ambient variety is required to have a structured equation theory, and it is typically large, normal, factorial, and Gorenstein, with explicit group actions and a fibration geometry such as a $\mathbb P^2\times\mathbb P^2$ - or $\mathbb P^1\times\mathbb P^1\times\mathbb P^1$ -fibration (Takagi, 2024, Takagi, 2024, Takagi, 2023).

This suggests that “key variety” is primarily a functional designation. Across these papers, it denotes not a single intrinsic class defined by one invariant, but a family of highly structured ambient models whose main significance lies in their effectiveness as construction spaces.

2. Diptych varieties and the birational-geometry origin

Diptych varieties were introduced by Brown and Reid as a new class of affine Gorenstein $\mathbb Q$ 0-folds obtained by smoothing the $\mathbb Q$ 1-dimensional singular locus of a reducible affine toric surface, with applications as key varieties and as local models of Mori flips of Type $\mathbb Q$ 2 (Brown et al., 2012). The basic input is a tent

$\mathbb Q$ 3

a reducible affine surface made of four toric surface pieces glued cyclically along coordinate axes. The components $\mathbb Q$ 4 and $\mathbb Q$ 5 are cyclic quotient singularities of types $\mathbb Q$ 6 and $\mathbb Q$ 7, while $\mathbb Q$ 8. The singular locus is a union of four axes of transverse ordinary double points.

A diptych is a compatible pair of toric $\mathbb Q$ 9-fold deformations

$3$0

where $3$1 smooths the top axes and $3$2 smooths the bottom axes. The main theorem asserts that any such compatible pair extends to a single affine Gorenstein $3$3-fold

$3$4

carrying an action of the $3$5-torus $3$6, with a regular sequence of eigenfunctions

$3$7

satisfying

$3$8

Thus $3$9 is a flat $4$0-parameter deformation of the tent (Brown et al., 2012).

The construction is controlled by explicit toric geometry and serial Kustin–Miller Gorenstein unprojection. The toric panels are encoded by long rectangles, the classification is indexed by $4$1-step recurrent continued fractions $4$2 with $4$3 and $4$4 in the main argument, and the $4$5-fold is built from a codimension-$4$6 complete intersection by adjoining variables in a specific order determined by the continued-fraction combinatorics and convexity of the monomial cone (Brown et al., 2012).

The significance of diptych varieties for the theory of key varieties lies in the combination of three features. First, they are explicit and combinatorial. Second, they contain natural sections $4$7, $4$8, and $4$9. Third, they are intended as ambient models for further construction, especially in local birational geometry. Brown and Reid identify Mori’s “continued division” algorithm for flips of Type $3$0 as the main motivation and interpret the canonical cover of a Type $3$1 Mori flip as a regular pullback from a diptych variety (Brown et al., 2012).

3. Homogeneous, weighted, and toric ambient models

Weighted flag varieties furnish another important class of key varieties. The construction begins with a flag variety $3$2, a weight datum $3$3, and the weighted flag variety $3$4. The weighted flag variety is embedded in a weighted projective space, its Hilbert series is computed from Lie-theoretic data, and suitable hypersurface sections then produce polarized $3$5-folds with controlled canonical class and singularities (Qureshi, 2014).

Two flagship cases are the weighted Lagrangian Grassmannian $3$6, coming from $3$7, and the weighted partial $3$8 flag variety $3$9, coming from $\mathbb Q$ 0. The former has dimension $\mathbb Q$ 1 and codimension $\mathbb Q$ 2 in $\mathbb Q$ 3, while the latter has dimension $\mathbb Q$ 4 and codimension $\mathbb Q$ 5 in $\mathbb Q$ 6. Their Hilbert series determine the canonical class, and quasilinear sections yield families of Calabi–Yau or Fano $\mathbb Q$ 7-folds. The paper explicitly uses these weighted flag varieties as “key varieties (Format)” to construct polarized $\mathbb Q$ 8-folds in codimensions $\mathbb Q$ 9 and $3$0 (Qureshi, 2014).

Richardson varieties provide a different use of the term. In the flag variety $3$1, a Richardson variety

$3$2

is presented as a key variety in representation theory and algebraic geometry, organizing Schubert calculus, torus actions, cell decompositions, and Weyl-group combinatorics. The paper on toric Richardson varieties asks when such a key variety is toric under the natural maximal-torus action. It proves that, under the hypothesis $3$3, the following are equivalent: $3$4 is toric, the Bruhat interval $3$5 contains no subinterval isomorphic to the Bruhat order on $3$6, and $3$7 is a lattice. It also gives an equivalent root-theoretic criterion in terms of linear independence of the special roots associated with Deodhar’s positive subexpression (Can et al., 2023).

These examples show that key varieties need not be affine unprojection formats. They may also be homogeneous or almost homogeneous varieties whose representation-theoretic structure makes them effective as universal ambient spaces.

Takagi’s work develops a large family of key varieties adapted to prime complex $4$1-Fano $4$2-folds of anti-canonical codimension $4$3. The first construction uses the affine varieties $4$4 and $4$5, defined in affine $4$6-space by Tom-format Pfaffian equations together with two extra unprojection relations. $4$7 is irreducible, reduced, Gorenstein, prime, and of codimension $4$8, with a $4$9-action and a $\mathbb P^2\times\mathbb P^2$ 0-action. Its partial projectivization has a $\mathbb P^2\times\mathbb P^2$ 1-fibration over $\mathbb P^2\times\mathbb P^2$ 2, and weighted projectivizations of $\mathbb P^2\times\mathbb P^2$ 3 and $\mathbb P^2\times\mathbb P^2$ 4 produce prime $\mathbb P^2\times\mathbb P^2$ 5-Fano $\mathbb P^2\times\mathbb P^2$ 6-folds in $\mathbb P^2\times\mathbb P^2$ 7 classes of the graded ring database (Takagi, 2021).

Takagi subsequently recast this program in the language of cubic Jordan algebras. The $\mathbb P^2\times\mathbb P^2$ 8-dimensional affine variety $\mathbb P^2\times\mathbb P^2$ 9, also written $\mathbb P^1\times\mathbb P^1\times\mathbb P^1$ 0, is constructed as the null-locus of the $\mathbb P^1\times\mathbb P^1\times\mathbb P^1$ 1-mapping of a $\mathbb P^1\times\mathbb P^1\times\mathbb P^1$ 2-dimensional nondegenerate quadratic Jordan algebra of a cubic form, after coordinatizing the algebra with $\mathbb P^1\times\mathbb P^1\times\mathbb P^1$ 3 parameters using three complementary primitive idempotents and the Peirce decomposition. $\mathbb P^1\times\mathbb P^1\times\mathbb P^1$ 4 is $\mathbb P^1\times\mathbb P^1\times\mathbb P^1$ 5-dimensional, irreducible, normal, factorial, and terminal, with a $\mathbb P^1\times\mathbb P^1\times\mathbb P^1$ 6-action, a $\mathbb P^1\times\mathbb P^1\times\mathbb P^1$ 7-fibration over the parameter space $\mathbb P^1\times\mathbb P^1\times\mathbb P^1$ 8, and an ideal generated by $\mathbb P^1\times\mathbb P^1\times\mathbb P^1$ 9 equations with $\mathbb Q$ 00 relations. Weighted projectivizations of $\mathbb Q$ 01 and its subvarieties are key varieties for prime $\mathbb Q$ 02-Fano $\mathbb Q$ 03-folds belonging to $\mathbb Q$ 04 classes in the GRDB, and the genus- $\mathbb Q$ 05 family No. $\mathbb Q$ 06 is characterized as arising precisely from a codimension- $\mathbb Q$ 07 linear section of the weighted projectivization $\mathbb Q$ 08 (Takagi, 2024).

Part II enlarges the Jordan-algebraic repertoire by introducing the $\mathbb Q$ 09-dimensional $\mathbb Q$ 10 and the $\mathbb Q$ 11-dimensional $\mathbb Q$ 12, constructed again as null-loci of $\mathbb Q$ 13-maps but now from coordinatizations with $\mathbb Q$ 14 and $\mathbb Q$ 15 parameters. The $\mathbb Q$ 16-parameter construction uses a fixed primitive idempotent and the associated Peirce decomposition; the $\mathbb Q$ 17-parameter construction uses Petersson’s quadratic Jordan subalgebra generated by $\mathbb Q$ 18-products of two elements. Both varieties are irreducible, normal, factorial, Gorenstein, and terminal, their ideals are generated by $\mathbb Q$ 19 equations with $\mathbb Q$ 20 syzygies, and each is related to a $\mathbb Q$ 21-fibration (Takagi, 2024).

Variety	Construction principle	Stated role
$\mathbb Q$ 22, $\mathbb Q$ 23	Tom-format Pfaffians and unprojection relations	Weighted projectivizations produce prime $\mathbb Q$ 24-Fano $\mathbb Q$ 25-folds in $\mathbb Q$ 26 GRDB classes
$\mathbb Q$ 27	Null-locus of the Jordan $\mathbb Q$ 28-mapping with $\mathbb Q$ 29 parameters	Key varieties for the genus- $\mathbb Q$ 30 No. $\mathbb Q$ 31 family and for $\mathbb Q$ 32 GRDB classes
$\mathbb Q$ 33, $\mathbb Q$ 34	Jordan $\mathbb Q$ 35-mapping with $\mathbb Q$ 36- and $\mathbb Q$ 37-parameter coordinatizations	New key varieties related with $\mathbb Q$ 38-fibrations

Taken together, these constructions establish a characteristic $\mathbb Q$ 39-fibration branch of the theory. The ambient variety is typically factorial or a UFD, its equations are explicit, and the resulting weighted projectivizations serve as universal construction spaces for codimension- $\mathbb Q$ 40 prime $\mathbb Q$ 41-Fano $\mathbb Q$ 42-folds.

5. Freudenthal-triple-system key varieties and $\mathbb Q$ 43-fibrations

A parallel branch of the theory replaces cubic Jordan algebras by Freudenthal triple systems. In this setting, the equations of the key varieties are derived from the condition that an element be strictly regular in an FTS. For a decomposition

$\mathbb Q$ 44

strict regularity is characterized by the equations

$\mathbb Q$ 45

together with

$\mathbb Q$ 46

These equations define the affine scheme $\mathbb Q$ 47 of strictly regular elements (Takagi, 2023).

When $\mathbb Q$ 48, the strictly regular locus in an $\mathbb Q$ 49-dimensional Freudenthal triple system is isomorphic to the affine cone over the Segre embedding of

$\mathbb Q$ 50

This produces the geometric background for key varieties related with $\mathbb Q$ 51-fibrations. The universal affine scheme $\mathbb Q$ 52 is then specialized to produce several concrete key varieties, most notably the $\mathbb Q$ 53-dimensional $\mathbb Q$ 54 (Takagi, 2023).

$\mathbb Q$ 55 is a factorial affine variety of codimension $\mathbb Q$ 56 in an affine $\mathbb Q$ 57-space, with only Gorenstein terminal singularities. Its defining ideal has a graded minimal free resolution of length $\mathbb Q$ 58 of type $\mathbb Q$ 59, its coordinate ring is a UFD, and its singularities along a specified locus are of type $\mathbb Q$ 60. A weighted projectivization of $\mathbb Q$ 61 in $\mathbb Q$ 62, cut by ten general weight- $\mathbb Q$ 63 hypersurfaces, yields examples of prime $\mathbb Q$ 64-Fano $\mathbb Q$ 65-folds, including No. $\mathbb Q$ 66 in the GRDB (Takagi, 2023).

This FTS-based approach also clarifies the relation with cluster formats. The paper shows that the maximal $\mathbb Q$ 67-cluster variety is isomorphic, after a weighted homogeneous change of coordinates, to a subvariety of $\mathbb Q$ 68. A plausible implication is that FTS equations provide a unifying description for a substantial part of the earlier cluster-variety constructions.

6. Duality, Sarkisov links, and higher-dimensional $\mathbb Q$ 69-Fano key varieties

Another major development treats key varieties not only as ambient spaces for weighted complete intersections but also as higher-dimensional $\mathbb Q$ 70-Fano varieties encoding Sarkisov links. In the duality paper, prime $\mathbb Q$ 71-Fano $\mathbb Q$ 72-folds in five classes—genus $\mathbb Q$ 73, genus $\mathbb Q$ 74, genus $\mathbb Q$ 75 of $\mathbb Q$ 76-type, genus $\mathbb Q$ 77 of $\mathbb Q$ 78-type, and genus $\mathbb Q$ 79—are realized as linear sections of unique rational $\mathbb Q$ 80-Fano varieties $\mathbb Q$ 81 of Picard number $\mathbb Q$ 82. From a vector bundle $\mathbb Q$ 83 on a base $\mathbb Q$ 84, one defines a dual bundle $\mathbb Q$ 85 by

$\mathbb Q$ 86

and obtains a dual variety in the projective space $\mathbb Q$ 87. The paper proves that the Fano side of the Sarkisov link is cut out on $\mathbb Q$ 88, while the curve side is recovered as a complementary linear section of $\mathbb Q$ 89; in the genus- $\mathbb Q$ 90 context this leads to a characterization of a general canonical curve of genus $\mathbb Q$ 91 with a $\mathbb Q$ 92 (2211.06784).

The 2025 paper on genus-one prime $\mathbb Q$ 93-Fano $\mathbb Q$ 94-folds with six $\mathbb Q$ 95-singularities extends this philosophy by constructing higher-dimensional key varieties

$\mathbb Q$ 96

through extensions of Sarkisov links to higher dimensions. The resulting $\mathbb Q$ 97 and $\mathbb Q$ 98 are terminal $\mathbb Q$ 99-factorial $3$00-Fano varieties of Picard number $3$01. General codimension-$3$02 linear sections of $3$03 are exactly the Type $3$04 threefolds, and conversely every Type $3$05 threefold arises in this way. General hyperplane sections of $3$06 are exactly the Type $3$07 threefolds, and conversely every Type $3$08 threefold is such a section (Takagi, 18 Jul 2025).

In both the duality framework and the genus-one construction, the midpoint geometry of the Sarkisov link is central. The duality paper builds its key and dual varieties from projective bundles and their tautological morphisms, while the genus-one paper uses the Segre cubic primal, its double cover, and a weighted cone over $3$09. This makes the key variety a container not only for the target $3$10-fold, but also for the birational transformations governing it (2211.06784, Takagi, 18 Jul 2025).

Across these developments, several recurrent structural features emerge. Key varieties are typically given by explicit equations—Pfaffians of $3$11 skew-symmetric matrices, Plücker relations and their modifications, or null-loci of $3$12-maps. They often carry large symmetry groups, such as $3$13, $3$14, $3$15, $3$16, $3$17, or $3$18. Many are affine, Gorenstein, factorial, or terminal; several weighted projectivizations have Picard number $3$19; and the target varieties are recovered by linear sections, regular pullbacks, or weighted complete intersections (Brown et al., 2012, Takagi, 2021, Takagi, 2024, Takagi, 2024, Takagi, 18 Jul 2025).

In this sense, key varieties occupy a distinctive position in contemporary explicit algebraic geometry. They are neither merely auxiliary embeddings nor only classification devices: they are structured ambient models in which geometry, equations, and birational behavior are made simultaneously visible.