A whole infinite world inside a finite circle
The Poincaré disk is a model of hyperbolic geometry — a self-consistent geometry where the parallel postulate fails, and through any point not on a given line there pass infinitely many lines never meeting it — packed entirely inside the open unit disk. Points of the hyperbolic plane are ordinary points strictly inside the disk boundary; the boundary circle itself represents points 'at infinity' and is never actually reached. The trick that makes this work is a non-Euclidean metric: distances near the boundary are stretched enormously compared to distances near the centre, so that a hyperbolic straight line of genuinely infinite length fits inside a region of finite Euclidean area, its endpoints only ever approaching, never touching, the rim.
Straight lines that look curved
A hyperbolic straight line — a geodesic, the locally shortest path between two points — is drawn in the Poincaré disk as either a diameter of the disk, or an arc of a circle that meets the boundary circle at exactly 90°. This perpendicularity condition is not a stylistic choice; it falls directly out of the model's underlying conformal map (the model preserves angles, though not distances or straight-line shapes, relative to ordinary Euclidean geometry). Two geodesics that look like they curve away from each other in the flat Euclidean picture are, within the model's own hyperbolic metric, perfectly straight — every apparent curve is an artifact of viewing an infinite non-Euclidean space through a finite Euclidean window.
Möbius transformations build the geodesics
Computing a hyperbolic geodesic through two given points reduces to a classical piece of circle inversion geometry: reflect one of the two points across the unit circle (invert it, sending a point at distance r from the centre to one at distance 1/r along the same ray), then the unique Euclidean circle through the two original points and that inverted point — restricted to its arc inside the disk — is the geodesic, and it automatically meets the boundary at right angles. The isometries (distance-preserving transformations) of the whole disk model are exactly the Möbius transformations that map the unit disk to itself, a fact that also explains why hyperbolic tilings drawn in this model — like the famous circle-limit woodcuts M. C. Escher made after being shown the construction by the mathematician H. S. M. Coxeter — always appear to shrink toward the boundary: the tiles are all congruent in the true hyperbolic metric, and only the model's distance distortion makes them look smaller near the rim.
geodesic through points P, Q inside the unit disk: 1. invert P across the unit circle: P' = P / |P|^2 (maps P to distance 1/|P| from centre) 2. the Euclidean circle through P, Q, P' (or the diameter, if P, Q, O are colinear) is the hyperbolic geodesic -- and it meets the unit circle at exactly 90 deg
Triangles whose angles never reach 180 degrees
The single most visible departure from Euclidean geometry in the model is that a hyperbolic triangle's three interior angles always sum to less than 180° — never equal to it. The shortfall, called the angular defect, is directly proportional to the triangle's hyperbolic area (Gauss–Bonnet again, the same theorem that governs the torus's curvature in the companion article): bigger hyperbolic triangles have a bigger defect, approaching a limiting defect of a full 180° for an 'ideal' triangle whose three vertices sit exactly on the boundary at infinity. That degenerate ideal triangle, despite having all three vertices infinitely far from each other in the hyperbolic metric, still has a strictly finite hyperbolic area — one of the more startling facts hyperbolic geometry hands you once you sit with the model for a while.
Why this particular model, out of several
The Poincaré disk is one of at least four standard models of the same hyperbolic plane — alongside the Poincaré half-plane, the Klein disk, and the hyperboloid model — and they are all mathematically equivalent, related by explicit coordinate transformations, differing only in which properties they preserve visually. The Poincaré disk's specific advantage is that it is conformal: angles measured in the picture match true hyperbolic angles exactly, which is why circles in the hyperbolic plane are drawn as genuine (if off-centre) Euclidean circles in this model, and why it is the model of choice for anything relying on angle-preserving intuition, from Escher's tessellations to visualising the boundary at infinity of certain physical and cosmological models.
Frequently asked questions
Why do 'straight' hyperbolic lines look like curved arcs in the disk?
The Poincaré disk model is conformal, meaning it preserves angles but not the shape of straight lines as we're used to seeing them. A genuine hyperbolic geodesic is drawn as either a diameter or an arc that meets the boundary circle at exactly 90°, and that arc is the model's faithful, if visually distorted, picture of a truly straight, infinitely long line.
Why do tiles in a hyperbolic tiling look smaller near the edge of the disk?
They are not actually smaller — every tile is congruent in the true hyperbolic metric. The model itself compresses distances near the boundary so severely that an infinite hyperbolic plane fits inside a finite Euclidean disk, and that compression is what makes equally-sized tiles appear to shrink as they approach the rim.
Do the angles of a hyperbolic triangle always sum to less than 180 degrees?
Yes, always, and the shortfall (the angular defect) is directly proportional to the triangle's hyperbolic area, by the Gauss–Bonnet theorem. An 'ideal' triangle with all three vertices on the boundary at infinity reaches the maximum possible defect, with an interior angle sum of exactly 0.
Try it live
Everything above runs in your browser — open Poincare Disk (Hyperbolic Geometry) and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.
▶ Open Poincare Disk (Hyperbolic Geometry) simulation