Voronoi Diagrams: How to Divide Space by Who's Closest

Stand anywhere on Earth and ask a simple question: which pizza place is closest to me? Answer that question for every point in space at once, and you've just built a Voronoi diagram — one of the most quietly ubiquitous structures in mathematics, biology, and engineering. It shows up in the cracks of dried mud, the cells of your own skin, the coverage maps of cell phone towers, and the way Uber decides which driver gets dispatched to your curb. It is, at heart, nothing more than the geometry of "nearest neighbor" — but that simple rule turns out to carve the world into shapes with astonishing regularity.

The Concept

Here's the setup. Scatter a handful of points on a flat plane — call them "seeds" or "sites." Maybe they're coffee shops, cell towers, or raindrop gauges. Now, for every other point on the plane, ask: which seed is closest? Color that point according to its nearest seed. Do this for every point simultaneously, and the plane splits into a patchwork of regions — one region per seed, each region being exactly the set of points for which that seed is the closest one.

That patchwork is a Voronoi diagram, and each individual region is called a Voronoi cell. The boundaries between cells are perpendicular bisectors between pairs of neighboring seeds — the set of points exactly equidistant from two sites. Where three or more cell boundaries meet, you get a Voronoi vertex, a point equidistant from three or more seeds.

The shapes that fall out of this simple rule are often convex polygons (in two dimensions), and seeds that are tightly clustered get small, cramped cells, while isolated seeds command enormous territories. It's the mathematical embodiment of "spheres of influence."

There's also a secret dance partner hiding behind every Voronoi diagram: the Delaunay triangulation, named after Russian mathematician Boris Delaunay. If you connect every pair of seeds whose Voronoi cells share a border, you get a network of triangles — and this triangulation has a famously elegant property: for every triangle in it, the circle drawn through its three corner points contains no other seed. Voronoi diagrams and Delaunay triangulations are mathematical duals — each one can be constructed directly from the other, and algorithms that build one often build the other as a byproduct.

Why It Matters

The concept is named for Georgy Voronoy, a Ukrainian mathematician born in 1868 in Zhuravka, in what is now Ukraine, who formally defined and studied these diagrams in a 1908 paper on quadratic forms. But like a lot of mathematics, the idea had already been quietly at work before it had a name. German mathematician Peter Gustav Lejeune Dirichlet had studied the same partitioning of space in the context of number theory decades earlier, around 1850, which is why the regions are sometimes called Dirichlet domains or Dirichlet cells in older or more formal texts.

Even more strikingly, one of the most famous uses of Voronoi-style reasoning predates the formal mathematics entirely. In 1854, London physician John Snow was investigating a cholera outbreak that killed roughly 500 people in the Soho area within about ten days. Snow plotted every death on a map alongside the locations of the neighborhood's water pumps, then drew a boundary marking which pump was closest to each residence — effectively hand-drawing a Voronoi cell around the infamous Broad Street pump. The deaths clustered overwhelmingly inside that cell. Snow convinced local authorities to remove the pump handle, and the outbreak subsided — a landmark moment in both epidemiology and, in retrospect, applied geometry, decades before Voronoi or Dirichlet gave the method a name.

Meteorologists found their own independent use for it. In 1911, American meteorologist Alfred H. Thiessen needed a way to estimate rainfall across a region using scattered, unevenly spaced measuring stations. His solution — now called a Thiessen polygon — assigns every point on a map to its nearest weather station and averages rainfall by the area of each station's territory. It's the identical idea to a Voronoi cell, discovered independently for an entirely different reason.

The Details

Nature, it turns out, builds Voronoi diagrams constantly, without knowing any geometry at all. Whenever something grows outward at a roughly uniform rate from multiple separate starting points, and growth simply stops wherever two expanding fronts meet, the result is a Voronoi tessellation almost by definition. That's why the cracked, polygonal pattern of dried mud looks the way it does, and it's why a dragonfly's wing is crisscrossed with irregular, puzzle-piece cells — each vein grew outward from a point and stopped when it collided with a neighbor's growth. Giraffe coat patterns emerge from a similar process: during embryonic development, scattered pigment-producing cells each claim a patch of skin, and the boundary between two giraffe spots falls roughly where two Voronoi cells would meet. Soap bubbles packed into a box behave the same way — each bubble occupies the set of points closer to its own center than to any neighbor's, flattening into shared planar walls exactly where two Voronoi cells border each other. Even the tightly packed epithelial cells lining your own skin, intestines, and blood vessels approximate three-dimensional Voronoi cells, a geometric fact that led researchers in 2018 to describe an entirely new category of cell shape, the "scutoid," that lets a layer of cells curve smoothly while still packing efficiently.

Human engineering leans on the same structure deliberately. Cellular phone networks are modeled as Voronoi diagrams with towers as seeds, since a phone generally connects to whichever tower is nearest — the honeycomb-like coverage maps phone companies publish are literal Voronoi diagrams. Delivery and ride-share platforms use Voronoi partitioning to decide which driver or warehouse should serve a given customer, since the nearest option is usually the cheapest. In computer graphics, Voronoi textures generate convincingly organic-looking stone, scales, and cellular surfaces because — per the previous paragraph — that's genuinely how many organic surfaces form. Robots and self-driving cars use a close cousin, the "generalized Voronoi diagram," to navigate safely through clutter by preferring paths that stay maximally far from every obstacle at once, since the Voronoi skeleton of a room is, by construction, the set of safest possible routes through it. Crystallographers use Voronoi cells (sometimes called Wigner-Seitz cells in that context) to describe the natural territory each atom occupies inside a crystal lattice, which governs how the material packs, conducts, and fractures.

Picture the diagram itself: it looks like a cracked windshield or a turtle's shell, and that visual is not a coincidence — the stress patterns that crack a windshield or the plates that tile a tortoise's carapace both follow the same "nearest flaw wins the territory" logic that generates a mathematical Voronoi partition. Zoom into any single cell and it's a convex polygon; zoom into any vertex and exactly three or more boundary lines converge, each one a perpendicular bisector silently enforcing a tie in distance between two seeds.

Takeaways

A Voronoi diagram partitions space into regions based purely on "which seed point is nearest," producing convex cells whose boundaries are perpendicular bisectors between neighboring seeds.
The idea has three independent origin stories: Dirichlet's number theory work around 1850, John Snow's 1854 hand-drawn cholera map, and Georgy Voronoy's formal 1908 definition — proof that a good idea tends to get reinvented wherever it's needed.
Its mathematical twin, the Delaunay triangulation, connects the same seeds into empty-circle triangles and can be derived directly from the Voronoi diagram, and vice versa.
Nature builds Voronoi patterns automatically whenever growth radiates outward from scattered points and stops on contact — giraffe spots, dragonfly wings, soap foam, and mud cracks are all real-world instances.
The same geometry now routes robots around obstacles, assigns customers to delivery drivers, maps cell tower coverage, and describes how atoms pack inside a crystal — a 19th-century abstraction doing extremely practical 21st-century work.

Resources: MacTutor biography of Georgy Voronoy · AMS Feature Column: Voronoi Diagrams and a Day at the Beach · Scientific American: Voronoi Tessellations and Scutoids Are Everywhere