arXiv++ Combinatorics

We ask whether most Boolean functions are determined by their low frequencies. We show a partial result: for almost every function $f: \{-1,1\}^p \to \{-1,1\}$ there exists a function $f': \{-1,1\}^p \to (-1,1)$ that has the same frequencies as $f$ up to dimension $(1/2-o(1))p$.