Definable flips
Lax185574.DefinableFlips · concepts/Lax185574/DefinableFlips.lean · lax-185574
No public endorsements yet.
Loading review…
Sign in with ORCIDNatural Language Statement
Definition
For s ∈ ℕ, an s-tuple p̄ = (p₁, …, p_s) of vertices of a graph G and a symmetric relation R ⊆ Π^s × Π^s, the graph G ⊕ (p̄, R) is the {p₁, …, p_s}-flip H of G obtained by partitioning V(G) into atomic types over p̄ and then flipping according to R. More precisely, for uv ∈ (V choose 2),
uv ∈ E(H) △ E(G) ⟺ (atp^G(u/p̄), atp^G(v/p̄)) ∈ R.
Such a flip is called s-definable. Equivalently, an s-definable flip is an S-flip for some S ⊆ V(G) of size at most s. An s-definable flip is also s'-definable for every s' ≥ s.
Concept map
In the paper
- page 20 of this submission's paper
Lean source view on GitHub
| 1 | import Lax185574.AtomicTypes |
| 2 | import Lax185574.Flips |
| 3 | import Lax185574.Partitions |
| 4 | |
| 5 | /-! |
| 6 | --- |
| 7 | title: Definable flips |
| 8 | type: definition |
| 9 | --- |
| 10 | For s ∈ ℕ, an s-tuple p̄ = (p₁, …, p_s) of vertices of a graph G and a |
| 11 | symmetric relation R ⊆ Π^s × Π^s, the graph G ⊕ (p̄, R) is the |
| 12 | {p₁, …, p_s}-flip H of G obtained by partitioning V(G) into atomic types |
| 13 | over p̄ and then flipping according to R. More precisely, for |
| 14 | uv ∈ (V choose 2), |
| 15 | |
| 16 | uv ∈ E(H) △ E(G) ⟺ (atp^G(u/p̄), atp^G(v/p̄)) ∈ R. |
| 17 | |
| 18 | Such a flip is called *s-definable*. Equivalently, an s-definable flip |
| 19 | is an S-flip for some S ⊆ V(G) of size at most s. An s-definable flip is |
| 20 | also s'-definable for every s' ≥ s. |
| 21 | |
| 22 | # Formalization notes |
| 23 | |
| 24 | An S-flip of G is a flip for the partition into atomic types over S, |
| 25 | `Flips.IsFlip G (Partitions.atomicTypes G S) H`. |
| 26 | |
| 27 | The graph G ⊕ (p̄, R) is `flipOver G p R`. It is built with |
| 28 | `SimpleGraph.fromRel`, which removes the diagonal and symmetrizes. For a |
| 29 | symmetric relation R the relation that is symmetrized is already |
| 30 | symmetric, so two distinct vertices are adjacent in `flipOver G p R` |
| 31 | exactly when their adjacency in G differs from membership of their |
| 32 | atomic types in R. Every statement about `flipOver` assumes R symmetric. |
| 33 | |
| 34 | An s-definable flip, `IsDefinableFlip G s H`, is defined by the second, |
| 35 | equivalent formulation: an S-flip for a finite set S of at most s |
| 36 | vertices. The two formulations agree on every graph with at least one |
| 37 | vertex, by listing S with repetitions as an s-tuple, or, if S is empty |
| 38 | and s ≥ 1, by a constant tuple, since every ∅-flip is also a {v}-flip. On the empty graph |
| 39 | there is no s-tuple for s ≥ 1, while the empty set still works. The |
| 40 | formulation by sets makes "s-definable implies s'-definable for s' ≥ s" |
| 41 | hold without exception. |
| 42 | -/ |
| 43 | |
| 44 | namespace Lax185574.DefinableFlips |
| 45 | |
| 46 | open Lax185574.AtomicTypes Lax185574.Flips Lax185574.Partitions |
| 47 | |
| 48 | universe u |
| 49 | |
| 50 | variable {V : Type u} |
| 51 | |
| 52 | /-- `G ⊕ (p̄, R)`: two distinct vertices are adjacent if and only if |
| 53 | their adjacency in `G` differs from membership of the pair of their |
| 54 | atomic types over `p̄` in the symmetric relation `R`. -/ |
| 55 | noncomputable def flipOver (G : SimpleGraph V) {s : ℕ} (p : Fin s → V) |
| 56 | (R : (Fin s → PairType) → (Fin s → PairType) → Prop) : SimpleGraph V := |
| 57 | SimpleGraph.fromRel fun u v => |
| 58 | Xor (G.Adj u v) (R (atomicTypeOver G p u) (atomicTypeOver G p v)) |
| 59 | |
| 60 | /-- `H` is an `s`-definable flip of `G`: an `S`-flip of `G` for some set |
| 61 | `S` of at most `s` vertices. -/ |
| 62 | def IsDefinableFlip (G : SimpleGraph V) (s : ℕ) (H : SimpleGraph V) : Prop := |
| 63 | ∃ S : Finset V, S.card ≤ s ∧ IsFlip G (atomicTypes G (S : Set V)) H |
| 64 | |
| 65 | end Lax185574.DefinableFlips |
| 66 |
Formalization notes
An S-flip of G is a flip for the partition into atomic types over S, .
The graph G ⊕ (p̄, R) is . It is built with , which removes the diagonal and symmetrizes. For a symmetric relation R the relation that is symmetrized is already symmetric, so two distinct vertices are adjacent in exactly when their adjacency in G differs from membership of their atomic types in R. Every statement about assumes R symmetric.
An s-definable flip, , is defined by the second, equivalent formulation: an S-flip for a finite set S of at most s vertices. The two formulations agree on every graph with at least one vertex, by listing S with repetitions as an s-tuple, or, if S is empty and s ≥ 1, by a constant tuple, since every ∅-flip is also a {v}-flip. On the empty graph there is no s-tuple for s ≥ 1, while the empty set still works. The formulation by sets makes "s-definable implies s'-definable for s' ≥ s" hold without exception.
Discussion
Ask a question or add context. Endorsements and structured flags are kept in the review panel above.
0 comments