Solving problems by adding a dimension

Original Media · 29m · English · YouTube · 3Blue1Brown

TLDR

Grant Sanderson presents five mathematical puzzles that demonstrate how stepping into higher dimensions provides elegant solutions to lower-dimensional problems. The puzzles cover rhombus tilings (solved via 3D cube stacks), covering a circle with strips (solved via hemisphere projection), Monge's theorem on circle tangents (solved via cones in 3D), tetrahedron volume formula (hinting at 4D determinants), and a 4D hypercube projection yielding rhombic dodecahedra. The video concludes with a reflection on the sadness that 4D+ insights remain inaccessible to human intuition.

For you

PRO members can sign in and request more information — or their own version of this digest, tuned to what they care about.

Sign in to unlock

Surprising Insights

•

The area of a spherical strip on a hemisphere depends only on its width (π×d), not its latitude — a consequence of Archimedes' equal-area projection onto a cylinder.

•

The flawed sphere-based proof of Monge's theorem fails when a small circle sits between two larger ones; the fix uses cones, which always have a plane through their apexes regardless of sizes.

•

The rhombic dodecahedron — a 3D shape with 12 rhombic faces — is the exact projection of a 4D hypercube, and it tiles 3D space because hypercubes tile 4D space.

•

The maximum number of moves in the 4D analog puzzle is n⁴, showing the fourth power is 'baked into' the 3D projection just as n³ was baked into the 2D tiling.

Key takeaways

•

A 2D rhombus tiling of a hexagon corresponds exactly to a 3D stack of cubes in an n×n×n frame; rotating a hexagon of three rhombi equals adding/removing a cube, making the maximum moves n³.

•

The Tarski-Plank problem (minimum sum of strip widths covering a unit circle) is solved by projecting strips onto a hemisphere: each strip's area becomes π×width, independent of position, proving the sum cannot be less than 2.

•

Monge's theorem (three intersection points of external tangents to three circles are collinear) is proven by lifting circles to cones with equal apex angles; the plane through the three apexes intersects the original plane in the required line.

•

The volume of a tetrahedron from vertex coordinates is given by a 4×4 determinant; this generalizes the triangle area formula and reveals why the determinant formula is what it is across dimensions.

•

Projecting a 4D hypercube along its main diagonal (1,1,1,1) yields a rhombic dodecahedron, which tiles 3D space; the 4D analog of the rhombus tiling puzzle has maximum moves n⁴.

Notable quotes

“Today I have a small series of puzzles for you, and if you even remotely like math, I can essentially guarantee that the solution to each one will delight you.”

▸ Notable quotes · 0:00

“the three-dimensionality really is baked into the puzzle itself, it's not coming from some bias that we have as three-dimensional creatures.”

▸ Notable quotes · 6:33

“the area of that strip up in three dimensions like this happens to be pi times d. Importantly, it doesn't matter where the strip is, it could be going over the center or closer to the edge, the area only depends on the width.”

▸ Notable quotes · 7:47

“The key is that instead of using spheres, we should really be using cones.”

▸ Notable quotes · 14:45

“this has a special name, it's called a rhombic dodecahedron.”

▸ Notable quotes · 22:14

“What I'm jealous of, and a little bit sad about, is the thought that there might be problems in three dimensions where somehow squinting your eyes as a four-dimensional creature or even higher would offer some guiding light that, to you and me, is inaccessible.”

▸ Notable quotes · 28:52

Chapters

•

0:00 — Introduction and Puzzle 1: Rhombus Tiling: Introduces a series of puzzles where higher-dimensional thinking solves lower-dimensional problems. Puzzle 1: tiling a hexagon with 60°/120° rhombi; moves consist of rotating three rhombi forming a hexagon. Asks if any tiling can reach any other, and the maximum number of moves.

•

4:26 — Puzzle 2: Tarski-Plank Problem: Cover a unit circle with strips (regions between parallel chords). Minimize the sum of strip widths. Trivial parallel strips give sum = 2 (diameter). Can we do better?

•

4:30 — Solution to Puzzle 1: 3D Cube Stacks: Coloring rhombi by orientation reveals a 3D cube stack projection. Each hexagon rotation corresponds to adding/removing a cube. Any tiling connects to any other via the empty configuration; maximum moves = n³ for hexagon side length n.

•

6:44 — Solution to Puzzle 2: Hemisphere Projection: Project strips onto a hemisphere above the circle. Each strip's area = π × width, independent of position. Total area ≥ hemisphere area (2π), so sum of widths ≥ 2. Parallel strips are optimal.

•

10:25 — Puzzle 3: Monge's Theorem (Three Circles): Three circles, external tangents for each pair intersect at three points. Theorem: these three points are collinear. Initial 3D proof uses spheres and a resting plane, but fails for certain configurations.

•

14:41 — Fixing Monge's Theorem with Cones: Replace spheres with similar cones (equal apex angles). The line through two cone apexes passes through the center of similarity (tangent intersection). The plane through all three apexes intersects the original plane in the line containing all three points. Works for all configurations, even nested circles.

•

17:40 — Puzzle 4: Tetrahedron Volume and Determinants: Find volume of tetrahedron from vertex coordinates. Hint: involves a 4×4 determinant. Analogous to triangle area in 2D. Full solution deferred to a future linear algebra video.

•

18:51 — Puzzle 5: 4D Hypercube Projection and Rhombic Dodecahedra: Project a 4D hypercube along (1,1,1,1) to get a rhombic dodecahedron, which tiles 3D space. The 3D analog of the rhombus tiling uses squished cubes (projections of 4D cubic cells). Sliding pieces through the center corresponds to adding/removing hypercubes; maximum moves = n⁴.

•

25:56 — Conclusion: The Limits of Intuition: Higher-dimensional constructs (quaternions in 4D, sphere packing in 24D, random vectors in high dimensions) solve real problems but require analysis over intuition. The sadness: 4D+ insights that could guide 3D proofs are inaccessible to human visualization.

Speakers

•

Grant Sanderson — Host/Creator

⋮