The Chebotarev theorem on roots of unity was originally a conjecture made by Ostrowski in the context of lacunary series.

Chebotarev was the first to prove it, in the 1930s. This proof involves tools from Galois theory and pleased Ostrowski, who made comments arguing that it "does meet the requirements of mathematical esthetics".^[1] Several proofs have been proposed since,^[2] and it has even been discovered independently by Dieudonné.^[3]

Statement

Let ${\displaystyle \Omega }$ be a matrix with entries ${\displaystyle a_{ij}=\omega ^{ij},1\leq i,j\leq n}$, where ${\displaystyle \omega =e^{2\mathrm {i} \pi /n},n\in \mathbb {N} }$. If ${\displaystyle n}$ is prime then any minor of ${\displaystyle \Omega }$ is non-zero.

Equivalently, all submatrices of a DFT matrix of prime length are invertible.

Applications

In signal processing,^[4] the theorem was used by T. Tao to extend the uncertainty principle.^[5]