Mori dream space¶

A normal projective variety whose nef and movable divisor classes are controlled by finitely many semiample cones on small Q-factorial models.

Version: v1 · 2026-10-17 · History
Domain-specific #: 14227
Domain group: Formal Sciences
Origin domain: Mathematics
Subdomain: Algebraic Geometry → Mathematics

Core Idea¶

A Mori dream space, in Hu and Keel's sense, is a normal projective variety for which divisor geometry admits a finite description. Its nef cone is generated by finitely many semiample line-bundle classes, and finitely many small Q-factorial modifications have nef cones whose pullbacks cover the movable cone. The definition also requires Q-factoriality and equality of the rational Picard and numerical divisor spaces. These are membership conditions, not just benefits one hopes the variety will have.^[1]

The name does not mean that every calculation is easy. Hu and Keel show that, under the stated projective, Q-factorial and Picard/numerical hypotheses, finite generation of the Cox ring is equivalent to the Mori dream space condition. This is a powerful alternative test under those hypotheses, not a rule that any finitely generated ring makes any variety a Mori dream space.^[1]

Structural Signature¶

Sig role-phrases:

Normal projective carrier. Start with a normal projective variety \(X\). Its closed projective embedding fills the live Projective Variety parent, without requiring smoothness or a fixed equation.^[1]^[2]
Divisor-class eligibility. Require Q-factoriality and \(\operatorname{Pic}(X)_{\mathbb Q}=N^1(X)_{\mathbb Q}\) so that the divisor-class conditions of Hu and Keel's definition apply in the stated setting.^[1]
Finite semiample nef cone. \(\operatorname{Nef}(X)\) is generated by finitely many semiample line-bundle classes. Nefness alone does not supply the required morphisms from semiample divisors.^[1]
Finite small-model family. There are finitely many small Q-factorial modifications \(f_i:X\dashrightarrow X_i\), each with the same finite semiample-nef property. The identity model can suffice in a simple case; a larger family is allowed.^[1]
Movable-cone coverage. The union of the pulled-back \(\operatorname{Nef}(X_i)\) covers \(\operatorname{Mov}(X)\). A finite list that leaves movable classes uncovered does not meet the definition.^[1]

The number of chambers, exact models, Cox-ring presentation and toric status may vary. Remove divisor-class eligibility, finite semiample nef generation or the finite movable cover, and the defining identity fails. Hu and Keel's ruled-surface example has individually finitely generated section rings but a nef non-semiample divisor, illustrating one such failure.^[1]

What It Is Not¶

Not merely projective or Q-factorial. Those conditions identify the carrier but do not supply finite semiample nef and movable-cone structure.^[1]
Not finite generation of one section ring. Hu and Keel give a case in which even all individual section rings are finitely generated while a nef divisor is not semiample; this is not the Mori dream space condition.^[1]
Not synonymous with a toric variety. Smooth toric projective examples qualify under Hu and Keel's hypotheses, but the degree-five del Pezzo surface \(X_4\) provides a non-toric Mori dream space through its finitely generated, non-polynomial Cox ring.^[1]^[3]
Not a Grassmannian identity. Batyrev and Popov identify the Cox ring of \(X_4\) with a Grassmannian homogeneous-coordinate ring. They do not identify the surface \(X_4\) itself with the Grassmannian.^[3]
Not an unrestricted image or log-Fano inheritance rule. The inspected sources do not establish the seed's all-surjective-images claim. Hu and Keel's stated log-Fano corollary is dimension-limited to at most three in this paper.^[1]

Scope of Application¶

The identity is literal in birational algebraic geometry when a normal projective variety satisfies the three parts of Hu and Keel's definition. It lets one organize movable divisor classes through finitely many nef cones on small models. Hu and Keel then derive termination of divisor-directed Mori programs and a finite chamber description of birational contractions; those are consequences of membership, not separate definitions.^[1]

For a Q-factorial projective variety with \(\operatorname{Pic}(X)_{\mathbb Q}=N^1(X)_{\mathbb Q}\), one may instead establish finite generation of the Cox ring and invoke Hu and Keel's equivalence. Batyrev and Popov's degree-one generators for the Cox ring of the degree-five del Pezzo surface \(X_4\) make that route concrete. The equivalence establishes its Mori dream space roles without supplying an explicit list of all \(X_4\) nef rays or small modifications.^[1]^[3]

Outside this divisor-geometric setting, “finite options make planning easy” is an analogy, not a literal use of this name.

Clarity¶

The definition has two finite requirements that must be kept together. Finite semiample generators control nef classes on each relevant model; finitely many qualifying small models then cover movable classes on \(X\). A list of chambers without semiample generators, or a list of generators without movable coverage, leaves a gap.^[1]

Cox-ring finite generation is a conditional equivalent test. Before using it, check the projective and Q-factorial carrier and the rational Picard-to-numerical equality. Conversely, a theorem-mediated proof of \(X_4\) membership should be described as such: Batyrev and Popov supply finite Cox generators; Hu and Keel's equivalence supplies the full definition. Neither paper's inspected passage enumerates \(X_4\)'s complete nef-cone/SQM map.^[1]^[3]

Manages Complexity¶

A projective variety can have many effective divisors, linear systems and birational models. This abstraction compresses that behavior into a finite audit: identify the carrier and Picard setting, inspect semiample generators on its nef cone, find finitely many qualifying small models, and test whether their nef cones cover the movable cone. Hu and Keel's chamber theorem makes the resulting finite organization useful for rational contractions.^[1]

The compression does not erase the geometry of a specific surface. For \(\mathbb P^2\), one nef ray and the identity model suffice by an elementary rank-one argument; for \(X_4\), finite Cox generation supplies a theorem-based certificate while its Plücker relations show a different algebraic presentation. Those implementations fill the same definition by different evidence routes.^[1]^[3]

Abstract Reasoning¶

First verify that \(X\) is a normal projective Q-factorial variety and that \(\operatorname{Pic}(X)_{\mathbb Q}=N^1(X)_{\mathbb Q}\). Then ask whether the nef cone has finitely many semiample line-bundle generators and whether finitely many small Q-factorial models, each meeting that condition, cover the movable cone by pulled-back nef cones. A failed role is a failed classification, even if another ring or cone has a finite description.^[1]

Alternatively, after verifying the same carrier hypotheses, test finite generation of \(\operatorname{Cox}(X)\) and apply Hu and Keel's Proposition 2.9. Keep this theorem-mediated route distinct from a direct enumeration of the nef and movable structures. The ruled elliptic surface in their Remark 1.12 is a useful rejection test: individually finite section rings do not compensate for a nef class that is not semiample.^[1]

Knowledge Transfer¶

Within algebraic geometry, the definition transfers from a toric rank-one plane to a non-toric del Pezzo surface because both satisfy the projective carrier and finite divisor-control conditions. The proof route changes: the plane's geometry is elementary, while the del Pezzo case can be certified by its Cox ring and Hu and Keel's conditional equivalence. Its exact Plücker relations do not become required features of other Mori dream spaces.^[1]^[3]

Outside algebraic geometry, finite states or finite plans may suggest an analogy. The named class does not transfer without projective varieties, divisor classes, semiample nef cones, small Q-factorial modifications and movable-cone coverage. The live Projective Variety parent also remains domain-specific; no prime status is inferred from the word dream or from generic finiteness.

Examples¶

Canonical: the projective plane¶

\(\mathbb P^2\) is a smooth projective toric surface. Its rational Picard and numerical divisor spaces are one-dimensional, generated by the ample class of \(\mathcal O(1)\). Its nef and movable cones are the same ray, generated by that semiample class, so the identity modification alone covers the movable cone. Hu and Keel identify toric varieties as Mori dream space examples and, under their smooth projective Picard hypothesis, show that the toric Cox ring is polynomial. The explicit one-ray/identity-model mapping here is an elementary deduction from that setting, not a case table printed in their paper.^[1]

Mapped back: the normal projective carrier is smooth \(\mathbb P^2\); divisor-class eligibility comes from smooth Q-factoriality and its rank-one Picard/numerical space; the finite semiample nef cone is the ray generated by \(\mathcal O(1)\); the finite small-model family is the identity; and movable-cone coverage holds because that one nef ray is the movable ray.

Unlike case: degree-five del Pezzo surface \(X_4\)¶

Let \(X_4\) be the blow-up of \(\mathbb P^2\) at four general-position points, a smooth projective del Pezzo surface. Batyrev and Popov prove finite degree-one generators for its Cox ring and identify that ring with a Grassmannian Plücker coordinate ring carrying quadratic relations. The smooth rational blow-up has the required Q-factorial and rational Picard/numerical properties; combined with Cox finite generation, Hu and Keel's Proposition 2.9 certifies Mori dream space membership. Under their smooth-projective hypotheses, the non-polynomial Cox presentation distinguishes this example from a toric variety. This is a cross-source deduction, not a claim that \(X_4\) equals the Grassmannian.^[1]^[3]

Mapped back: the normal projective carrier is smooth \(X_4\); divisor-class eligibility follows from its smooth rational blow-up and Picard/numerical condition; the finite Cox-ring theorem, through Hu and Keel's equivalence, certifies the finite semiample nef cone, finite small-model family and movable-cone coverage. The inspected papers do not enumerate all those cones or models for \(X_4\); the ring theorem is the evidence route.

Structural Tensions¶

The inspected definition and case theorems establish conditions, consequences and varying presentations. They do not establish an inherent opposed-pressure tradeoff that every Mori dream space must negotiate. Toric polynomial versus non-toric Plücker Cox rings is an admissible variation; direct cone enumeration versus the Cox-ring equivalence is a choice of proof, not a constitutive tension.^[1]^[3]

Structural–Framed Character¶

Mori dream space is structural within algebraic geometry: the five roles state repeatable conditions on varieties and divisor cones. Its evaluative weight is mathematical tractability of birational classification, not a claim about human or moral value. Human practice supplies the definitions and proof methods, while the projective carrier and its divisor relations are mathematical properties independent of any one institution. Hu and Keel introduced the name as a research classification, not an institutional rule.^[1]

The words finite, model and Γ-space travel widely, but their literal roles here require projective algebraic geometry. Importing a generic finite-state system would lose the divisor and birational content; recognizing another qualified projective variety requires checking the same roles. There is no reviewed substrate-independent prime parent for this named identity. Its character: a sharply typed geometric subclass with a finite organizational structure, framed by algebraic-geometric carriers and proof conditions.

Structural Core vs. Domain Accent¶

The skeletal relation is finite control of nef and movable divisor behavior on a projective variety: finitely generated semiample nef cones on finitely many small Q-factorial models cover movable classes. The one strict live parent is Projective Variety, whose closed projective embedding supplies the broader carrier but not those finite divisor conditions. Its Algebraic Variety ancestor adds no second direct edge.^[1]^[2]

The domain accent is indispensable: Q-factoriality, rational Picard/numerical equality, semiample line bundles, nef and movable cones, Cox rings and small birational modifications are not generic finite-planning roles. Removing them yields an analogy about finitely many options, not a Mori dream space. The exact plane or del Pezzo presentation may vary; the named identity remains domain-specific and does not clear the prime bar. Future-prime question, not an asserted edge: whether finite coverage of a large possibility space by a small family of locally tractable regions recurs with equally strict roles outside algebraic geometry would require independent non-geometric cases and a separate abstraction review. The current two cases establish only this geometric subclass.

This entry is a kind of Projective variety.

The strict edge is subsumption to Projective variety. Every Hu–Keel Mori dream space has the normal projective carrier required by that parent, and its added Q-factorial, Picard and finite divisor conditions narrow the class. Hu and Keel's non-semiample nef example shows why projective geometry alone is insufficient. The broader Algebraic Variety node is reached through Projective Variety; adding a direct edge would bypass that nearer genus.^[1]

Toric Variety is an instance family, not a parent: \(X_4\) is non-toric under the Cox-ring criterion. Cox Ring names an algebraic diagnostic or object, not the Mori dream space itself. A Mori chamber is a consequence or component of the finite organization, not a synonym for the variety. No prime edge is claimed from generic finiteness or planning language.^[1]^[3]

Relationships to Other Abstractions¶

Current abstraction Mori dream space Domain-specific

Parents (1) — more general patterns this builds on

Mori dream space is a kind of Projective variety Domain-specific

Every Mori dream space is a projective variety with further finite nef and movable-cone conditions.

Hierarchy path (1) — routes to 1 parentless root

Mori dream space → Projective variety → Algebraic Variety

Neighborhood in Abstraction Space¶

Mori dream space sits in a sparse region of the domain-specific corpus (64^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Sheaves & Birational Geometry (23 abstractions)

Nearest neighbors

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

Not to Be Confused With¶

Projective Variety supplies the carrier but need not satisfy finite semiample-nef or movable-cover conditions. Toric Variety can supply examples but does not contain the non-toric \(X_4\) case. Cox Ring is a graded algebra used as a conditional equivalent test for membership under Hu and Keel's hypotheses. Mori dream region is a related local cone notion defined separately in their §2.12. Finite generation of one section ring is too weak, as their §1.12 counterexample shows. Grassmannian \(\operatorname{Gr}(3,5)\) appears as the projective spectrum of the \(X_4\) Cox ring, not as the del Pezzo surface itself.^[1]^[3]

References¶

[1] Yi Hu and Sean Keel, “Mori Dream Spaces and GIT,” Michigan Mathematical Journal 48 (2000), 331–348, original author version https://arxiv.org/pdf/math/0004017. Definition 1.10; Proposition 1.11; Remark 1.12; Proposition 2.9; Corollaries 2.10 and 2.16. The log-Fano result in this version is limited to dimension at most three. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r ↩s ↩t ↩u ↩v ↩w ↩x ↩y ↩z ↩27 ↩28 ↩29

[2] The Stacks Project, “Types of Varieties,” Definition 33.26.1, https://stacks.math.columbia.edu/tag/04L0. Projective varieties admit closed immersion into finite-dimensional projective space over the field. registry ↩a ↩b

[3] Victor V. Batyrev and Oleg N. Popov, “The Cox Ring of a Del Pezzo Surface,” original author version https://arxiv.org/pdf/math/0309111. Section 2, Theorem 3.2 and Proposition 4.1. The Cox-ring/Grassmannian identification does not identify the surface with the Grassmannian. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j