Counting permutations that keep a set inside a set #
The expander existence proof needs one combinatorial estimate: of the n!
permutations of Fin n, at most descFactorial s k · (n - k)! map a given
k-element set inside a given s-element set.
The proof is the obvious one, made precise. Restricting a permutation to K
gives an injection into S; there are descFactorial s k of those. Two
permutations with the same restriction differ only outside K, where they are
injections from an (n - k)-set into the complement of the common image —
another (n - k)-set — so each restriction is shared by at most (n - k)!
permutations.
Main results #
Complexity.card_perm_mapsTo_le— the estimateComplexity.card_perm_escape_le— its consequence for the escape count: few permutations move only a small part ofSout ofS
How many points of S the permutation σ sends outside S.
Equations
- Complexity.escape σ S = {v ∈ S | σ v ∉ S}.card
Instances For
Restrictions of a permutation #
The permutations mapping K into S.
Equations
- Complexity.permsInto S K = {σ : Equiv.Perm (Fin n) | ∀ v ∈ K, σ v ∈ S}
Instances For
The restrictions that can occur: injective maps of K into S.
Equations
- Complexity.restrs S K = {f : ↥K → Fin n | Function.Injective f ∧ ∀ (v : ↥K), f v ∈ S}
Instances For
The values of a restriction.
Equations
Instances For
Permutations with little escape #
Few permutations move only a little of S out of S. Such a
permutation keeps a (s - t)-element subset of S inside S, and there are
few subsets and, by card_perm_mapsTo_le, few permutations for each.