Intersection Curve¶

An intersection curve is a one-dimensional common locus of two surfaces, with regularity determined by how they meet.

Version: v1 · 2026-10-04 · History
Domain-specific #: 13740
Domain group: Formal Sciences
Origin domain: Mathematics
Subdomains: Differential Geometry, Surface Intersection → Mathematics
Aliases: Surface intersection curve, Curve of intersection of surfaces

Core Idea¶

An intersection curve is a one-dimensional locus common to two surfaces in the same geometric space. Its points satisfy both surface conditions, and its curve character matters: two surfaces can instead have no common points, only an isolated contact point, or a two-dimensional overlapping patch. Calling every surface meeting an intersection curve would erase that distinction.^[1]

The common curve need not be everywhere transverse. At a regular transverse point of two smooth surfaces in three-dimensional space, their normals are linearly independent; the intersection is locally a smooth curve whose tangent runs parallel to the normals' cross product. At tangential contact, that cross product vanishes and different local analysis is needed. A tangential intersection may still contain a curve; the transverse tangent rule is a powerful special case, not the definition of the whole class.^[1]

Structural Signature¶

First surface: contributes one two-dimensional geometric constraint.
Second surface: contributes a second constraint in the same ambient space.
Common locus: points satisfy both constraints; only a one-dimensional component is the intersection curve.
Regularity condition: local normal independence, when present, ensures a regular transverse curve and supplies a tangent direction.
Representation and retrieval: implicit equations, parametric patches or geometric constructions describe the surfaces; analytic or numerical methods locate curve components without defining their identity.

Condensed: two surface constraints → one-dimensional common component; transverse normals → local regularity and tangent rule.^[1]

Sig role-phrases: first surface; second surface; one-dimensional shared locus; conditional transverse normal independence; representation-independent identity.

What It Is Not¶

Not any surface overlap. A common two-dimensional patch is not itself a curve, although its boundary could be one under another specification.
Not an isolated tangency. One shared point has no one-dimensional component.
Not necessarily transverse. Tangent surface pairs can have curve components, but the normal-cross-product formula does not apply at a point where the normals are parallel.^[1]
Not the marching algorithm. Tracing from seed points is one computational route; the locus exists independently of which algorithm finds it.^[2]
Not guaranteed complete after one trace. A numerical path may follow one branch while missing disconnected branches or singular contacts.

Scope of Application¶

In analytic geometry, the planes x = 0 and y = 0 meet in the z-axis, {(0, 0, z)}. Each plane supplies one linear condition, and solving both yields their common one-dimensional set. Their normals (1, 0, 0) and (0, 1, 0) cross to (0, 0, 1), the line's direction. This simple instance makes the roles visible without a numerical tracing method.

A plane–sphere pair gives a differently shaped instance: z = 0 and x² + y² + z² = 1 meet in the unit circle {(cos t, sin t, 0)}. Their normals (0, 0, 1) and (2x, 2y, 2z) are independent at every point of that circle. By contrast, z = 1 meets the same sphere at one point, (0, 0, 1), and therefore yields no intersection curve. These equations are constructed analytic checks, not geometries reported by the cited CAD papers.

In computer-aided geometric design, two smooth parametric or rational surfaces may meet along several curve branches. Surface–surface intersection research treats the computation of such curves through marching, characteristic points, matrix representations or other methods. The CAD representation changes how the locus is found, not what counts as a point on it.^[2]^[3]

Clarity¶

Separate a geometric object from the condition that simplifies it and the algorithm that approximates it. The object is the one-dimensional common locus. Transversality yields a local smoothness guarantee and a tangent direction. Marching uses such local information to generate sample points, then must check them against both surface constraints. A tangential point may require a different method; it does not retroactively make every nearby curve cease to be an intersection curve.^[1]^[2]

Also distinguish a local component from the complete global answer. Establishing a valid tangent at one point does not show that every branch of the common set has been found.

Manages Complexity¶

The pair-of-constraints representation compresses many surface combinations into one question: what is their common set, and which components are curves? The transverse case reduces local direction finding to the normals' cross product. But global topology and singularities remain; a simple local rule cannot enumerate components or certify that a numerical routine has not jumped from one branch to another.^[1]^[3]

Abstract Reasoning¶

Given two surface descriptions, first test whether they share points and whether the shared set has a one-dimensional component. At a smooth common point, compare the normals. If they are independent, the local tangent is parallel to their cross product; an analytic calculation or a numerical step can then follow the curve. Correct the proposed point so that it lies on both surfaces, not merely near one. If normals become dependent, pause the transverse inference and analyze possible tangency, singularity, branch joining or overlap separately.^[1]^[2]

To claim a complete intersection, examine components and endpoints globally. A traced sample is evidence of a curve branch, not by itself evidence that all branches were discovered.

Knowledge Transfer¶

The common-locus test transfers literally among plane, quadric and freeform surface pairs. The tangent cross-product rule transfers only at smooth transverse points, and a numerical tracing method requires representation- and conditioning-specific checks. Beyond geometry, the prime idea of Intersection travels widely, but a shared set of nongeometric objects is not literally an intersection curve.

Examples¶

Two specified planes¶

Take the planes x = 0 and y = 0 in three-dimensional space. Their common solutions are exactly (0, 0, z) for every real z: a line, not a point or a surface patch. The normals (1, 0, 0) and (0, 1, 0) are independent; their cross product is (0, 0, 1), tangent to the line. This is a constructed elementary calculation.

Mapped back: x = 0 + y = 0 → common z-axis → independent normals and a cross-product tangent → exact linear-equation solution.

Plane and sphere¶

Take the plane z = 0 and unit sphere x² + y² + z² = 1. A shared point must have z = 0 and x² + y² = 1, so the common locus is the unit circle {(cos t, sin t, 0): 0 ≤ t < 2π}. On that circle, the plane normal (0, 0, 1) and sphere normal (2x, 2y, 0) are independent; their cross product gives a tangent direction. This is another constructed exact calculation, not an instance taken from the cited CAD papers. Replace the plane by z = 1: the only common point is (0, 0, 1), a tangential near miss with no curve. Freeform CAD intersections use algorithms to approximate the same common-locus type when exact elimination is impractical.^[2]^[3]

Mapped back: plane + sphere → one-dimensional circle → independent normals along the circle → exact constraint solution; the tangent-plane near miss loses the one-dimensional component.

Structural Tensions¶

No intrinsic opposed-cost tension is established for the geometric locus itself. Local regularity versus global completeness and exact formulas versus broad shape coverage are challenges for computing the locus, not conflicting properties constitutive of an intersection curve. The former is a Clarity/Abstract Reasoning check; the latter guides algorithm choice in CAD. Diagnostic: have both surface constraints and the curve component been established, independently of the method used to recover them?

Structural–Framed Character¶

Intersection curve is predominantly structural: common points and local dimension do not depend on an observer's preference. Engineering practice nevertheless chooses representations, precision tolerances and what counts as a usable CAD trace, so an algorithmic report is partly framed by a task. The geometric relation is not created by an institution; a particular CAD software convention merely represents it. The vocabulary travels literally among geometric surface pairs, while calling a nongeometric overlap a “curve” imports a metaphor rather than recognizes this typed identity. The portable skeleton is shared content constrained to a lower-dimensional component, but the named curve requires geometric surfaces. Its character: a geometric relation type with formal identity and task-sensitive computation, not a cross-domain prime.

Structural Core vs. Domain Accent¶

The skeleton is two carriers sharing a constrained locus. The domain accent supplies surfaces in a common ambient geometry, a one-dimensional component, and conditional normal/tangent analysis. Remove the geometric dimension and the remaining common locus may still be an Intersection, but not an intersection curve. The staged strict genus is live Curve, because the admitted result is one-dimensional; Intersection is a presupposed common-membership operation, not necessarily identical to the selected curve component. The named entry itself does not clear the prime bar because its exact recognition test remains geometric.

This entry presupposes Intersection and is a kind of Curve.

The staged strict genus is live Curve: each admitted instance is a one-dimensional geometric locus, though most curves are not surface intersections. The staged composition/presupposes parent is Intersection: common membership in both surfaces is necessary, but the named curve may be only one component of their full intersection, so strict subsumption to that complete set operation would overclaim.

Relationships to Other Abstractions¶

Current abstraction Intersection Curve Domain-specific

Parents (2) — more general patterns this builds on

Intersection Curve is a kind of Curve Domain-specific

An intersection curve is a one-dimensional geometric curve locus.
Intersection Curve presupposes Intersection Prime

The curve is selected from the common locus of two surfaces.

Hierarchy paths (3) — routes to 3 parentless roots

Intersection Curve → Curve → Continuity → Neighborhood → Topology

Show alternative paths (2)

Neighborhood in Abstraction Space¶

Intersection Curve 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 — Surface Geometry & Projective Transforms (6 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With¶

Line–line intersection can yield a point rather than a surface–surface curve. Intersection graph records pairwise overlaps among a family of objects as a graph, not their geometric common locus. Curve alone need not arise from two surfaces. Surface–surface intersection algorithm computes or approximates the locus but is not the locus itself.

References¶

[1] Original research, “Differential geometry of intersection curves of two surfaces” (1999). registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g

[2] Original research, “A marching method for parametric surface/surface intersection” (1990). registry ↩a ↩b ↩c ↩d ↩e

[3] Original CAD research, “Computing the Intersection of Two Rational Surfaces Using Matrix Representations” (2022). registry ↩a ↩b ↩c