Approximation by juntas in the symmetric group, and forbidden intersection problems

A family of permutations $\mathcal{F} \subset S_{n}$ is said to be $t$-intersecting if any two permutations in $\mathcal{F}$ agree on at least $t$ points. It is said to be $(t-1)$-intersection-free if no two permutations in $\mathcal{F}$ agree on exactly $t-1$ points. If $S,T \subset \{1,2,\ldots,n\}$ with $|S|=|T|$, and $\pi: S \to T$ is a bijection, the $\pi$-star in $S_n$ is the family of all permutations in $S_n$ that agree with $\pi$ on all of $S$. An $s$-star is a $\pi$-star such that $\pi$ is a bijection between sets of size $s$. Friedgut and Pilpel, and independently the first author, showed that if $\mathcal{F} \subset S_n$ is $t$-intersecting, and $n$ is sufficiently large depending on $t$, then $|\mathcal{F}| \leq (n-t)!$; this proved a conjecture of Deza and Frankl from 1977. Equality holds only if $\mathcal{F}$ is a $t$-star. In this paper, we give a more `robust' proof of a strengthening of the Deza-Frankl conjecture, namely that if $n$ is sufficiently large depending on $t$, and $\mathcal{F} \subset S_n$ is $(t-1)$-intersection-free, then $|\mathcal{F} \leq (n-t)!$, with equality only if $\mathcal{F}$ is a $t$-star. The main ingredient of our proof is a `junta approximation' result, namely, that any $(t-1)$-intersection-free family of permutations is essentially contained in a $t$-intersecting {\em junta} (a `junta' being a union of a bounded number of $O(1)$-stars). The proof of our junta approximation result relies, in turn, on a weak regularity lemma for families of permutations, a combinatorial argument that `bootstraps' a weak notion of pseudorandomness into a stronger one, and finally a spectral argument for pairs of highly-pseudorandom fractional families. Our proof employs four different notions of pseudorandomness, three being combinatorial in nature, and one being algebraic.