Segre Class¶

A graded intersection-theoretic cycle class of a cone or closed embedding that records its projective directions and reduces to the inverse Chern class of the normal bundle for a regular embedding.

Version: v1 · 2026-08-30 · History
Domain-specific #: 2741
Origin domain: algebraic geometry
Subdomain: intersection theory
Aliases: Segre class of a cone, Segre class of an embedding

Core Idea¶

A Segre class assigns a graded cycle class to a cone over a scheme, or equivalently to the normal cone of a closed embedding. It packages how the cone points away from its base into the Chow group of that base. For a cone \(C\to X\), one projectively completes it as \(\mathbf P(C\oplus 1)\), uses the first Chern class of its tautological quotient line bundle, and pushes powers of that class back to \(X\). William Fulton's treatment makes this construction a basic ingredient of modern intersection theory.^[1]

For a closed embedding \(X\hookrightarrow Y\), the relevant cone is the normal cone \(C_XY\), and the resulting class is written \(s(X,Y)=s(C_XY)\). When the embedding is regular, the normal cone is the normal vector bundle \(N_{X/Y}\), so the definition collapses to

\[ s(X,Y)=c(N_{X/Y})^{-1}\cap [X]. \]

The point of the abstraction is precisely that it continues to make sense for singular or nonregular embeddings, where no ordinary normal bundle captures all the excess-intersection data.^[1]

Structural Signature¶

Base: a scheme \(X\) carrying the final cycle class.
Cone: a cone \(C\to X\), commonly the normal cone of \(X\hookrightarrow Y\).
Projective completion: \(q:\mathbf P(C\oplus 1)\to X\), which makes cone directions proper over the base.
Tautological class: \(\xi=c_1(\mathcal O(1))\) on the projective completion.
Graded pushforward: powers of \(\xi\) cap the fundamental class and are pushed along \(q\).
Dimension bookkeeping: each component lands in the appropriate Chow group of \(X\).
Regular-embedding reduction: vector-bundle cones reproduce inverse total Chern classes.
Functorial controls: flat pullback and proper birational modification preserve the intended class under stated hypotheses.^[1]

A proposed object qualifies only when the cone, base, tautological projective geometry, grading, and pushforward are present. A class merely named after Segre, or a generic inverse Chern expression with no cone or embedding, does not pass.

What It Is Not¶

It is not a Chern class. Chern classes are characteristic data of vector bundles; a Segre class is inverse-Chern data for a vector bundle and a cone-based extension for general embeddings. It is not the Segre embedding of products of projective spaces, the Segre variety, a Segre number, or a Segre symbol.

It is not the fundamental class \([X]\) alone. Higher components record how \(X\) sits in \(Y\), and two embeddings of the same abstract \(X\) can have different normal cones. It is not an arbitrary cycle supported on \(X\): its components arise from one specified cone construction. Nor is it the normal cone itself; the cone is geometric input, while the Segre class is its intersection-theoretic output.

Scope of Application¶

Segre classes appear wherever intersection theory must account for nontransverse or singular behavior. Deformation to the normal cone replaces an intersection by geometry on a normal cone, and Segre classes convert that cone into cycles entering refined Gysin maps, excess-intersection formulas, residual intersections, and localized characteristic classes.^[1]

They also support computations through blowups. If \(\pi:\operatorname{Bl}_X Y\to Y\) is the blowup and \(E\to X\) its exceptional divisor, the class can be recovered by pushing an expression involving \(E\); this is one reason birational invariance is computationally valuable.^[1] Applications extend to multiplicities, singularity invariants, enumerative formulas, and characteristic classes of singular schemes. The definition remains tied to Chow groups or an explicitly substituted oriented theory; it should not be generalized informally to every homology-valued characteristic class.

Clarity¶

For a rank-\(r\) vector bundle \(E\to X\), let \(\pi:\mathbf P(E)\to X\) and \(\xi=c_1(\mathcal O_E(1))\). Projective-bundle relations imply that pushforwards of powers of \(\xi\) are the cohomological Segre components \(s_i(E)\), assembled as

\[ s^\bullet(E)=1+s_1(E)+s_2(E)+\cdots=c(E)^{-1}. \]

The corresponding cycle-valued total Segre class is \(s(E)=s^\bullet(E)\cap[X]\). Thus \(c(E)s^\bullet(E)=1\). Under the quotient-line convention, \(s_1(E)=-c_1(E)\) and \(s_2(E)=c_1(E)^2-c_2(E)\). Sign conventions can change if a source projectivizes lines rather than quotients; a reference-grade calculation must declare its convention.

For a regular embedding, replace \(E\) with \(N_{X/Y}\) and cap with \([X]\). For a general embedding, replace the bundle with \(C_XY\). That substitution is not cosmetic: the normal cone retains nilpotent, embedded, and excess-dimensional structure that a naïve normal-bundle formula would discard.

Manages Complexity¶

The abstraction compresses embedding geometry into a graded class participating in the algebra of Chow groups. Instead of carrying the entire cone through every intersection calculation, one transports its Segre class, multiplies by Chern classes, and pushes or pulls components under controlled morphisms. The regular case provides a consistency check, while the general cone case handles singularities.

This compression is disciplined rather than lossless. Distinct cones can share a Segre class, so the class is not a complete invariant. Its value is that it preserves exactly the information demanded by many intersection formulas. Computational strategies can switch among projective completion, blowup, and regular-embedding presentations without changing the target invariant.

Abstract Reasoning¶

Segre classes instantiate a recurring mathematical move: extend an invariant from locally free objects to singular objects by replacing a missing quotient bundle with a cone. The extension is constrained by normalization on vector bundles and functorial behavior. This lets arguments prove a formula in a convenient birational model and push the result back.

The inverse identity \(s^\bullet(E)=c(E)^{-1}\) also makes Segre classes natural correction terms when a total Chern class would otherwise overcount normal directions. In excess intersection, the expected transverse quotient is unavailable; a cone and its Segre class keep the failure measurable rather than treating it as an exception.

Knowledge Transfer¶

The construction transfers safely from vector bundles to normal cones because projective completion, tautological line classes, cap products, and proper pushforward survive the change of input. What transfers is not the assertion that a cone is a bundle, but that the same projective-direction recipe defines a class and matches the inverse Chern class where both descriptions exist.

Transfer to other cohomology theories requires new orientation and pushforward assumptions. Likewise, an informal “Segre-type” invariant in another domain is not automatically this abstraction. The decisive test is whether the input is an algebraic cone or embedding and the output is the graded characteristic cycle defined by the projectivized-cone operation.

Examples¶

Zero section of a line bundle. If \(X\hookrightarrow L\) is the zero section of a line bundle \(L\), then \(N_{X/L}\cong L\), so \(s(X,L)=c(L)^{-1}\cap[X]=(1-c_1(L)+c_1(L)^2-\cdots)\cap[X]\).
Effective Cartier divisor. For a Cartier divisor \(D\hookrightarrow Y\), the normal bundle is \(\mathcal O_Y(D)|_D\), hence \(s(D,Y)=c(\mathcal O(D)|_D)^{-1}\cap[D]\).
Singular closed subscheme. If an ideal sheaf defines a nonregular \(X\subset Y\), \(C_XY=\operatorname{Spec}_X\bigoplus I^n/I^{n+1}\) remains available even when \(I/I^2\) is not locally free. Its Segre class is defined while a normal-bundle inverse is not.^[2]
Blowup computation. On \(\operatorname{Bl}_X Y\), the exceptional divisor provides a proper model whose divisor-class series can be pushed to \(X\), replacing a difficult cone computation with intersection on the blowup.^[1]

Structural Tensions¶

Bundle normalization vs. singular extension. The class must agree with \(c(N)^{-1}\cap[X]\) for regular embeddings without pretending every normal cone is a bundle. Diagnostic: verify the regular case separately and retain the cone for the general case.
Convention dependence vs. invariant content. Projectivization and \(\mathcal O(1)\) conventions affect signs and indexing. Diagnostic: state the convention before comparing formulas.
Birational convenience vs. scheme structure. Blowups simplify calculations but can hide components if the pushforward theorem is used outside its hypotheses. Diagnostic: name the morphism, properness, and exact invariance statement.
Compression vs. completeness. A Segre class controls many intersection formulas but does not reconstruct the cone. Diagnostic: do not infer geometric isomorphism from equality of Segre classes.

Structural–Framed Character¶

The structural core is the cone-to-graded-cycle operation. The algebraic-geometric frame supplies schemes, Chow groups, projective completion, tautological Chern classes, and proper pushforward. Removing that frame leaves broad patterns of representation and invariance, not a Segre class.

The class is therefore domain-specific even though its architecture is reusable. Its identity depends on specialized objects and theorems; its role cannot be stated faithfully as generic classification, decomposition, or aggregation.

Structural Core vs. Domain Accent¶

Structural core: input object with directions, canonical completion, graded operator, pushforward, normalization, and functoriality.

Domain accent: normal cones, projective bundles, Chow groups, Chern classes, closed embeddings, and intersection products. The accent is constitutive. A topological characteristic class called “Segre” only by analogy does not automatically share the same admissibility or functorial laws.

Segre Class compositionally presupposes Intersection because its output lives in the intersection-theoretic cycle algebra and corrects nonproper intersections. This is not specialization: a Segre class is not a set-theoretic overlap operation. It is also related to Invariance through flat pullback and birational pushforward properties, and to Representation because it compresses cone geometry into a cycle class. The minimal proposed parent is Intersection; additional relations are informative but unnecessary for placement.

Relationships to Other Abstractions¶

Current abstraction Segre Class Domain-specific

Parents (1) — more general patterns this builds on

Segre Class presupposes Intersection Prime

Segre Class compositionally presupposes Intersection because its output lives in the intersection-theoretic cycle algebra and corrects nonproper intersections.

Hierarchy path (1) — routes to 1 parentless root

Segre Class → Intersection → Set and Membership

Neighborhood in Abstraction Space¶

Segre Class sits in a sparse region of the domain-specific corpus (90^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (1565 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-09-08

Not to Be Confused With¶

Chern class: bundle characteristic class; inverse in the regular case, not identical in general.
Normal cone: geometric input to \(s(X,Y)\), not the resulting cycle.
Normal bundle: only available as the cone in a regular embedding.
Segre embedding: map \(\mathbf P^m\times\mathbf P^n\hookrightarrow\mathbf P^{(m+1)(n+1)-1}\).
Segre number: a numerical value extracted from components, not the total class.
Fundamental class: records the scheme as a cycle but not its embedded normal behavior.

References¶

[1] William Fulton, Intersection Theory, 2^nd ed., Springer, 1998, especially Chapter 4, “Cones and Segre Classes,” DOI: 10.1007/978-1-4612-1700-8. registry ↩a ↩b ↩c ↩d ↩e ↩f

[2] The Stacks Project Authors, “The normal cone of an immersion,” The Stacks Project, Tag 09RM, current online edition. registry ↩