Circular Triangle¶

A simple plane figure bounded by three cyclic circular arcs, with corner angles measured between the arcs' tangents.

Version: v1 · 2026-10-17 · History
Domain-specific #: 14086
Domain group: Formal Sciences
Origin domain: Mathematics
Subdomains: Plane Geometry, Circular Arc Geometry → Mathematics
Aliases: Arc Triangle

Core Idea¶

A circular triangle has a simple closed boundary made from three circular arcs joined cyclically at three vertices. Each corner angle is measured between the tangents of the two meeting arcs, on the interior side. The defining pattern is the three-arc boundary and its incidence, not a particular radius, construction, convexity, or angle triple.^[1] A Reuleaux triangle and a suitable facial three-cycle in a curved graph drawing both meet this geometric test, but they do so for different reasons.^[2]^[1]

The main cited mathematical treatment calls this an arc-triangle and permits a straight segment as a degenerate circular arc.^[1] This entry uses circular triangle for the ordinarily curved figure; whenever a theorem or example uses the paper's wider convention, it says so. In particular, an ordinary all-straight triangle is only a limiting case under that convention, not an unqualified curved example.

The paper gives an exact existence test for a simple arc-triangle with proposed tangent angles. If the three angles are \(\theta_1,\theta_2,\theta_3\) in \([0,2\pi]\), set \(\psi=(\pi-\sum_i\theta_i)/2\) and \(\phi_i=\psi+\theta_i\). Such a triangle exists under the paper's convention exactly when every \(\phi_i\) satisfies \(-\pi<\phi_i<\pi\).^[1] This is a constraint on possible forms, not a replacement for the three-arc definition.

Structural Signature¶

Ambient plane and ordered vertices — Three junctions are visited cyclically. A fourth distinct junction changes the side count.^[1]
Three arc edges — A portion of a circle joins each successive vertex pair. Straight segments count only when the broader degenerate-arc convention is declared.^[1]
Simple closure — Consecutive sides meet at their shared endpoint, and nonconsecutive portions do not cross. This gives one closed boundary with an inside and outside.^[1]
Tangent corner angles — At each vertex, the two side tangents define the local interior angle. Chord angles or angles between circle centers are different quantities.^[1]
Optional construction constraints — Equal radii, opposite-vertex centers, constant width, or graph-drawing angle rules pick subclasses or applications; none defines every circular triangle.^[2]^[1]

The first four roles recognize the simple geometric type. The fifth is deliberately conditional: adding a construction can make new theorems available, but cannot be read backward as a universal property of the type.

What It Is Not¶

Not necessarily a Reuleaux triangle. Intersecting three disks whose radii equal the side length of the equilateral triangle joining their centers produces a Reuleaux triangle of constant width.^[2] A circular triangle may have unequal radii, mixed convex and concave arcs, or no constant-width property.

Not an ordinary straight-sided triangle without qualification. A standard polygon has line-segment sides. Eppstein, Frishberg, and Osegueda deliberately include segments as degenerate arcs in their formal arc-polygon definition.^[1] A theorem proved under that convention can include all-straight limiting cases; the everyday curved sense does not silently inherit them.

Not every three-arc walk. Three arcs can cross, or one can run through the opposite vertex. The admitted simple type requires a closed non-self-crossing boundary. Local corner angles alone do not guarantee that global condition.^[1]

Not a two-arc lens or an arbitrary curvilinear triangle. The former has the wrong number of sides; the latter may use noncircular curves. Nor does an isoperimetric optimizer define the whole type: optimization adds its own area, point, and radius constraints.

Scope of Application¶

The literal scope is plane geometry of boundaries made from circle arcs. It includes convex and nonconvex figures when the sides form a simple closed curve. In constant-width geometry, the Reuleaux triangle is a strong special case.^[2] In graph drawing, three circular-arc edges may bound a face of a planar Lombardi drawing, where incident edges meet with equal angular spacing.^[1]

Angle realizability needs its precise formal scope. The cited theorem allows tangent angles in \([0,2\pi]\), tests simplicity with the derived \(\phi_i\), and counts straight segments as degenerate arcs.^[1] A statement about physically curved sides only, or about an arc path that self-intersects, needs separate proof. The frozen candidate's printed pairwise inequalities are not the cited theorem's test; the source-exact \(\phi_i\) condition governs this entry.

The frozen article also mentions a constrained three-point isoperimetric problem. Its detailed area regimes are not used here because the cited book pages were not directly verified for this release. That omission does not narrow the geometric definition.

Clarity¶

Four questions classify a proposed figure. Are there exactly three cyclic side edges? Is each edge circular under the declared convention? Do they form a simple closed boundary? Are stated interior angles measured between tangents on the inside?^[1] An answer about only the shape's visual resemblance to a triangle leaves these conditions unresolved.

For proposed angles, the paper's test is more discriminating than a casual “curved triangles can have many angles” rule. Take \((3\pi/2,3\pi/2,\pi/2)\). The paper gives \(\psi=-5\pi/4\) and \(\phi=(\pi/4,\pi/4,-3\pi/4)\), all strictly between \(-\pi\) and \(\pi\), so the triple is realizable under its convention.^[1] One inequality printed in the frozen article would instead reject it because \(3\pi/2+3\pi/2-\pi/2=5\pi/2\). The contradiction shows why the exact theorem, not a paraphrased formula, must anchor the test.

Manages Complexity¶

The type separates identity from parameters. The number and kind of sides, their cyclic incidence, and simple closure decide whether the figure belongs. Radii, circle centers, arc bulges, angle values, area, and constant width describe or restrict particular members. An analyst can compare two figures as circular triangles even while one is convex and the other is not.

The separation also keeps application-specific constraints in their place. A graph face must satisfy the drawing's equal-angle rule; a Reuleaux triangle must satisfy its disk-intersection and width conditions. Neither rule is imposed on every three-arc boundary.^[1]^[2] When a drawing problem fails, the angle theorem can locate a geometric obstruction rather than merely reporting that a picture could not be sketched.

Abstract Reasoning¶

Start with a candidate cyclic triple of arcs and test the side and simplicity conditions. If its interior tangent angles are prescribed, compute \(\psi\) and each \(\phi_i\) from the paper's formula. An angle outside the open interval \((-\pi,\pi)\) rules out a simple realization under that theorem; satisfying all three values permits one.^[1] This turns a visual construction problem into a small feasibility calculation.

The calculation does not identify a unique Euclidean size or set of radii. Möbius transformations map circles and their degenerate line cases to the same types and preserve intersection angles, but a transformation can exchange the interior and exterior of an arc-polygon; interior-angle statements must respect that qualification.^[1] The transformation property helps compare constructions. It is not part of the three-arc recognition rule, and this entry does not require a full same-angle classification claim.

Knowledge Transfer¶

The roles transfer literally from a disk-intersection construction to a graph-drawing face. In both cases there are three cyclic vertices, three arc sides, a simple enclosed region, and tangent-defined angles. The extra questions change: a geometer may ask about support-line width, while a graph drawer asks whether vertex degrees and embedding can support a prescribed face-angle triple.^[2]^[1]

This is transfer within geometry. Calling a three-party negotiation a “curved triangle” supplies none of the circular edges, tangents, or realizability test. The image can be metaphorically useful, but it does not instantiate the formal object.

Examples¶

Canonical — Reuleaux triangle¶

Begin with an equilateral triangle of side length one and intersect the three radius-one disks centered at its vertices. The resulting boundary consists of three congruent circular arcs joined at those vertices. It is simple and convex; the local tangent interior angles are each \(2\pi/3\), as follows from perpendicularity of a circle tangent to its radius at contact. Hynd's support-line argument shows the additional constant-width property.^[2]

Mapped back: plane/vertices → ordered equilateral vertex triple; arc edges → three equal-radius disk-boundary arcs; simple closure/interior → convex disk intersection; tangent angles → \((2\pi/3,2\pi/3,2\pi/3)\). Equal radii and constant width are subclass conditions, not the general test.

Applied / In Practice — a cactus graph's curved face¶

Take a three-cycle and attach one outward pendant edge at just one of its cycle vertices. The natural embedding makes the cycle a face. Theorem 17 guarantees a planar Lombardi drawing of that embedding; its cycle edges are arcs, with straight segments permitted as degenerate arcs in the paper's convention.^[1] One cycle vertex has degree three, and the other two have degree two. Equal spacing of incident edges therefore gives face angles \((2\pi/3,\pi,\pi)\). For that triple, \(\psi=-5\pi/6\) and \(\phi=(-\pi/6,\pi/6,\pi/6)\), so Theorem 6's simple arc-triangle test is satisfied.^[1] This degree-and-angle specialization is derived from the paper's definitions and theorem; it is not a pictured example in the paper. The angle sum rules out an all-straight triangle, although it does not force every individual side to curve.

Mapped back: plane/vertices → the ordered three cycle vertices; arc edges → Lombardi cycle edges under the stated degenerate convention; simple closure/interior → planar bounded cycle face; tangent angles → \((2\pi/3,\pi,\pi)\). The graph's equal-spacing rule is an added application constraint, distinct from the Reuleaux width rule.

Structural Tensions¶

T1: Local angular freedom vs simple global closure. Circular sides can accommodate corner-angle combinations unavailable to three straight segments. Yet independently chosen arcs can cross or pass through an opposite vertex. Requiring a simple boundary removes some locally plausible triples. More angular flexibility is useful only while the global boundary remains valid. Diagnostic: Do all three derived \(\phi_i\) lie strictly inside \((-\pi,\pi)\)?^[1]

T2: General three-arc type vs powerful subclasses. Keeping the identity at three arcs allows unequal radii and different constructions. Adding equal-radius, opposite-center, and constant-width conditions gives the Reuleaux subtype and additional deductions.^[2] The gain in theorem strength costs generality. Diagnostic: Is the claimed property established by the three-arc definition, or only by the construction added to this instance?

T3: Curved everyday sense vs degenerate formal closure. Allowing a straight segment as a limiting arc makes the source's theorems apply uniformly to arc-polygons.^[1] It also risks classifying an ordinary straight triangle as a circular one if the convention is left unstated. Tightening the name to visibly curved cases preserves everyday clarity but no longer states the theorem's exact scope. Diagnostic: Is this claim about genuinely curved sides, or about the source's broader arc-triangle class?

Structural–Framed Character¶

Vocabulary travel: The words side, vertex, and triangle travel broadly, but the operational test here requires planar circle arcs and their tangents. Evaluative weight: The name praises no figure; constant width or optimality is an added property. Human-practice dependence: Graph drawing is a human application, but the existence and angle conditions are geometric claims. Institutional origin: No legal or organizational rule creates the shape; mathematical definitions and conventions specify it. Import versus recognition: The label recognizes an already specified arc boundary, while selecting a degenerate convention changes which limiting cases are admitted.^[1]

Its character: strongly structural within geometry, with little evaluative or institutional framing. The only important framing choice is the declared mathematical convention that admits or excludes straight limiting sides; it changes borderline membership, not the underlying three-side incidence test.

Structural Core vs. Domain Accent¶

What can lift: A finite cyclic assembly of parts forms a closed whole, while constraints at each junction can make that whole feasible or impossible. A bounded future-Prime question is whether this local-compatibility-to-global-closure inference recurs with an operational yes/no test in two materially unlike nongeometric systems, such as a protocol cycle and a chemical reaction cycle, without importing planar arcs, tangent angles, or this paper's formula. Evidence would have to identify corresponding junction constraints and predict which proposed wholes fail to close in both systems. If the comparison offers only three-part visual resemblance or needs the geometric formula to work, it is analogy and does not establish a Prime. The two examples here both remain in planar geometry.

What stays in the domain: Three cyclic circle arcs with a simple planar closure and the tangent-angle convention define this identity; straight segments are admitted only under the stated degenerate-arc convention.^[1] Disk intersection constructs the Reuleaux subtype, while the \(\phi_i\) condition tests whether a proposed tangent-angle triple can be realized; neither is required of every circular triangle.^[1]^[2] Remove the circle-arc carrier and the named circular triangle disappears. That is why the entry is domain-specific despite its use in two mathematical settings.

This entry is part of Circular arc.

The live Prime Invariance is relevant when a Möbius transformation preserves circle/line form and intersection angles, but the circular triangle's identity does not require a transformation or a constant width.^[1] The live Prime Boundary supplies an inside/outside comparison, yet its full catalog account includes operational demarcation and crossing rules beyond a plane figure. These comparisons illuminate aspects of the geometry without becoming necessary parent identities for the figure.

Relationships to Other Abstractions¶

Current abstraction Circular Triangle Domain-specific

Parents (1) — more general patterns this builds on

Circular Triangle is part of Circular arc Domain-specific

A circular triangle contains a circular arc as an identity-bearing side of its simple three-sided boundary.

Hierarchy path (1) — routes to 1 parentless root

Circular Triangle → Circular arc

Neighborhood in Abstraction Space¶

Circular Triangle sits in a sparse region of the domain-specific corpus (88^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (2721 abstractions)

Nearest neighbors

Polygon — 0.84
RAC drawing — 0.81
Sphericon — 0.79
Arrangement of Pseudolines — 0.79
Desargues's Theorem — 0.79

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

Not to Be Confused With¶

Reuleaux polygon: A constant-width curve with an odd number of equal-radius circular arcs; a Reuleaux triangle is a specific three-arc member, while other circular triangles lack those extra conditions.^[2]
Polygon: Its ordinary sides are straight segments; an arc-triangle uses circular sides except for a declared degenerate convention.
Circular arc: One side component, not the closed three-side region.
Curve: A broad continuous path; it does not require exactly three circle arcs or an enclosed plane interior.
Schwarz triangle function: A conformal mapping with a circular-arc triangle as target under further analytic constraints, not the target figure itself.
An impossible Lombardi face: Theorem 18 in the source derives an infeasible forced angle triple for one embedded cactus. That is a nonexistence result, not an example of a circular triangle.^[1]

References¶

[1] David Eppstein, Daniel Frishberg, and Martha C. Osegueda, “Angles of Arc-Polygons and Lombardi Drawings of Cacti,” Computational Geometry 112 (2023), 101982; author manuscript, especially Definition 1 and Definition 2 on PDF p. 3, Corollary 4 and Theorem 6 on PDF pp. 5–6, and Theorems 17–18 on PDF pp. 9–10. https://arxiv.org/pdf/2107.03615. 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

[2] Ryan Hynd, “The Blaschke–Lebesgue Theorem Revisited,” arXiv v2 (2024), PDF pp. 1–2, intersection-of-three-equal-disks construction, Figure 2, and constant-width support-line argument. https://arxiv.org/pdf/2303.04359. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j