Download lean/OAI/Computability/DegreeRigidity/Effective/GuardedCoding.lean from roseking/openAI-math-dataset: direct link, hf CLI and curl.
- Browser
- Download file 3.11 kB
-
https://huggingface.co/datasets/roseking/openAI-math-dataset/resolve/main/lean/OAI/Computability/DegreeRigidity/Effective/GuardedCoding.lean
- Command line
-
hf download hf://datasets/roseking/openAI-math-dataset/lean/OAI/Computability/DegreeRigidity/Effective/GuardedCoding.lean
-
curl -L -o GuardedCoding.lean https://huggingface.co/datasets/roseking/openAI-math-dataset/resolve/main/lean/OAI/Computability/DegreeRigidity/Effective/GuardedCoding.lean
3.11 kB
| import OAI.Computability.DegreeRigidity.Effective.GuardedExtraction | |
| import OAI.Computability.DegreeRigidity.Computability.NormalizedRecovery | |
| import OAI.Computability.DegreeRigidity.Effective.AntichainParameters | |
| namespace OAI | |
| namespace TuringRigidity.GuardedCoding | |
| open EncodedForcing StageNormalization AntichainParameters | |
| theorem effective_coding (A : Oracle) : ∃ G₀ G₁ : Oracle, | |
| Reduces (join G₀ G₁) (OracleJump.jump A) ∧ | |
| (∀ k, ∃ C : Oracle, Reduces C (join G₀ (columns A k)) ∧ | |
| Reduces C (join G₁ (columns A k)) ∧ ¬ Reduces C (columns A k)) ∧ | |
| ∀ Y Z : Oracle, Reduces Y A → Reduces Z (join G₀ Y) → | |
| Reduces Z (join G₁ Y) → ¬ Reduces Z Y → ∃ k, Reduces (columns A k) Y := by | |
| let B := normalized A | |
| obtain ⟨hbound,hfirst⟩ := GuardedLimits.effective_first_clause B | |
| (fun k => (normalized_spec A k).2.1) | |
| refine ⟨GuardedLimits.G₀ B,GuardedLimits.G₁ B, | |
| reduces_trans hbound (OracleJump.jump_mono (normalized_reduces A)),?_,?_⟩ | |
| · intro k | |
| obtain ⟨C,h₀,h₁,hn⟩ := hfirst k | |
| obtain ⟨hBA,hAB⟩ := (degree_eq_iff _ _).mp (normalized_spec A k).1 | |
| exact ⟨C,reduces_trans h₀ (join_mono (reduces_refl _) hBA), | |
| reduces_trans h₁ (join_mono (reduces_refl _) hBA),fun h => hn (reduces_trans h hAB)⟩ | |
| · intro Y Z hY h₀ h₁ hn | |
| obtain ⟨k,hk⟩ := GuardedExtraction.common_lower_computes_column | |
| (fun k => (normalized_spec A k).2.2) | |
| (reduces_trans hY (NormalizedRecovery.normalized_inverse A)) | |
| (reduces_trans h₀ (AntichainCoding.swap_join_reduces _ _)) | |
| (reduces_trans h₁ (AntichainCoding.swap_join_reduces _ _)) hn | |
| exact ⟨k,reduces_trans (((degree_eq_iff _ _).mp (normalized_spec A k).1).2) hk⟩ | |
| theorem effective_antichain_parameters (A : Oracle) | |
| (hAnti : ∀ i j, degree (columns A i) ≤ degree (columns A j) → | |
| degree (columns A i) = degree (columns A j)) : | |
| ∃ g₀ g₁ : Degree, g₀ ⊔ g₁ ≤ OracleJump.degreeJump (degree A) ∧ | |
| ∀ x : Degree, CodedAntichain (degree A) g₀ g₁ x ↔ ∃ k, degree (columns A k) = x := by | |
| obtain ⟨G₀,G₁,hbound,hfirst,hsecond⟩ := effective_coding A | |
| have hm : ∀ k, NontrivialCommonLower (degree A) (degree G₀) (degree G₁) (degree (columns A k)) := by | |
| intro k | |
| obtain ⟨C,h₀,h₁,hn⟩ := hfirst k | |
| exact ⟨CodingExtraction.column_projection_reduces A k,degree C,h₀,h₁,hn⟩ | |
| have hc : ∀ x, NontrivialCommonLower (degree A) (degree G₀) (degree G₁) x → | |
| ∃ k, degree (columns A k) ≤ x := by | |
| intro x hx | |
| obtain ⟨Y,rfl⟩ := degree_surjective x | |
| obtain ⟨z,h₀,h₁,hn⟩ := hx.2 | |
| obtain ⟨Z,rfl⟩ := degree_surjective z | |
| exact hsecond Y Z hx.1 h₀ h₁ hn | |
| refine ⟨degree G₀,degree G₁,hbound,fun x => ?_⟩ | |
| constructor | |
| · intro hx | |
| obtain ⟨k,hk⟩ := hc x hx.1 | |
| exact ⟨k,le_antisymm hk (hx.2 _ (hm k) hk)⟩ | |
| · rintro ⟨k,rfl⟩ | |
| refine ⟨hm k,fun y hy hyk => ?_⟩ | |
| obtain ⟨j,hj⟩ := hc y hy | |
| exact hAnti j k (hj.trans hyk) ▸ hj | |
| end TuringRigidity.GuardedCoding | |
| end OAI | |