Let Ω {\displaystyle \Omega } be a matrix with entries a i j = ω i j , 1 ≤ i , j ≤ n {\displaystyle a_{ij}=\omega ^{ij},1\leq i,j\leq n} , where ω = e 2 i π / n , n ∈ N {\displaystyle \omega =e^{2\mathrm {i} \pi /n},n\in \mathbb {N} } . If n {\displaystyle n} is prime then any minor of Ω {\displaystyle \Omega } is non-zero.
Equivalently, all submatrices of a DFT matrix of prime length are invertible.
In signal processing,4 the theorem was used by T. Tao to extend the uncertainty principle.5
Stevenhagen et al., 1996 ↩
P.E. Frenkel, 2003 ↩
J. Dieudonné, 1970 ↩
Candès, Romberg, Tao, 2006 ↩
T. Tao, 2003 ↩