Kakeya in three and four dimensions

The explanation

Picture a thin straw of length 1 and radius 0.01. Its volume is π times 0.01 squared, about 0.000314. Now take 10,000 such straws, each pointing in a different direction. Their volumes add up to about 3.14. If no two straws overlapped, their union would fill that much space. The question is how much they can overlap, so that the union fills far less.

A set that contains a line segment of length 1 in every direction is called a Kakeya set. Overlap can be extreme. Besicovitch's constructions, from 1928, show that a Kakeya set can have zero volume. So volume is too blunt a measure here. Mathematicians use Hausdorff dimension instead. It measures size by how the number of small covering pieces grows as the pieces shrink. A line has dimension 1, a flat sheet 2, a solid block 3. A set can have zero volume and still have full dimension.

The Kakeya conjecture says that every Kakeya set in n-dimensional space has dimension n. In other words, the segments may overlap, but never so much that the set becomes thinner than space itself. Davies proved the flat, two-dimensional case in 1971. In 1995, Wolff proved dimension at least (n+2)/2. That is 5/2 in three dimensions and 3 in four. Katz and Zahl later pushed past 5/2 in three dimensions. Wang and Zahl then settled the three-dimensional set problem, in preprints from 2022, 2024 and 2025.

The first paper asks a harder question in three dimensions, called the Kakeya maximal conjecture. Go back to the straws. Suppose each straw keeps only a fraction of itself, call it λ, and the rest is thrown away. The maximal conjecture says the union still fills about λ³ times the total, with only a tiny loss. With λ = 1/2, λ³ is 1/8, so the 10,000 half-straws should fill about 0.39. The tiny loss is a factor like 0.01 raised to a small power. For the power 0.01, that factor is about 0.955. A fixed constant factor is also allowed, so 0.39 sets the scale, not an exact floor.

Wang and Zahl's theorem gives a bound of this shape. But it raises λ to some unnamed power, not 3. Their own paper states the need to replace that power by 3. That was the one obstacle. The difficulty is keeping exactly the cube while the argument cuts, rescales and restricts the straws. The first paper claims to keep it at every stepSource 1. It then extends the bound from kept pieces of straws to arbitrary functions, which is the full maximal conjectureSource 2.

The second paper moves to four-dimensional space and to the set question. Before it, the best lower bounds for dimension sat just above 3. Guth and Zahl, with work of Zahl and of Katz and Rogers, gave 3 + 1/40 for bounded, closed sets. That is 3.025. Katz and Zahl's planebrush method gave at least 3.059. The paper names the central difficulty plainly. Many segments can overlap heavily even when their directions are well spread outSource 3. The paper claims to prove that every Kakeya set in four dimensions has dimension 4Source 4. It uses an estimate from the first paper as an inputSource 5.

In the straw picture, the four-dimensional claim says this. Pack unit segments in every direction as cleverly as you like. The union may have zero volume, but its dimension is still 4. The claim covers every such set. No assumption is placed on the set, such as being closed or boundedSource 4. It is a statement about sets only. It does not give the four-dimensional maximal estimate for kept pieces of straws. In five or more dimensions, it yields only dimension at least 4, while the conjecture asks for the full dimensionSource 6.

How sure is this? Both results are claims in preprints written by the model. Neither paper has a Lean formalization listed in the release, so no computer proof checker has checked either argument. No specialist review has been published.

What it doesn't show. This does not prove the maximal conjecture in four dimensions, or full dimension in five or more dimensions.