F alpha le at optimizer
f_alpha_le_at_optimizer
Plain-language statement
Step 1c: H_hat is a maximizer: for all H ≥ 0, f_α(H) ≤ f_α(H_hat). This uses the trace Young inequality: for PSD A, B and conjugate exponents p, q > 1, ⟪A, B⟫ ≤ Tr[A^p]/p + Tr[B^q]/q. Applied with A = σ^γ ρ σ^γ, B = σ^{-γ} H σ^{-γ}, p = α, q = α/(α-1), the inner product identity ⟪ρ, H⟫ = ⟪A, B⟫ (under the support condition) yield...
Exact Lean statement
theorem f_alpha_le_at_optimizer (hα : 1 < α) (ρ σ : MState d)
(H : HermitianMat d ℂ) (hH : 0 ≤ H) (hker : σ.M.ker ≤ ρ.M.ker) :
f_alpha α H ρ σ ≤ f_alpha α (H_hat α ρ σ) ρ σFormal artifact
Lean source
theorem f_alpha_le_at_optimizer (hα : 1 < α) (ρ σ : MState d) (H : HermitianMat d ℂ) (hH : 0 ≤ H) (hker : σ.M.ker ≤ ρ.M.ker) : f_alpha α H ρ σ ≤ f_alpha α (H_hat α ρ σ) ρ σ := by rw [f_alpha_at_optimizer hα] -- Goal: f_alpha α H ρ σ ≤ Q̃_α(ρ‖σ) set γ : ℝ := (1 - α) / (2 * α) with hγ_def have hγ : γ ≠ 0 := by intro h; have h1 : (1 - α) / (2 * α) = 0 := hγ_def ▸ h have h2 : (2 : ℝ) * α ≠ 0 := by positivity rw [div_eq_zero_iff] at h1; rcases h1 with h1 | h1 <;> linarith set A := ρ.M.conj (σ.M ^ γ).mat set B := H.conj (σ.M ^ (-γ)).mat have hA_nn : 0 ≤ A := HermitianMat.conj_nonneg _ ρ.nonneg have hB_nn : 0 ≤ B := HermitianMat.conj_nonneg _ hH have h_inner : ⟪ρ.M, H⟫_ℝ = ⟪A, B⟫_ℝ := inner_eq_inner_conj_of_ker_le ρ σ H hker γ hγ have hpq : 1 / α + 1 / (α / (α - 1)) = 1 := by field_simp; ring have h_young := HermitianMat.trace_young A B hA_nn hB_nn α (α / (α - 1)) hα hpq have hα_pos : (0 : ℝ) < α := by linarith have hαm1_pos : (0 : ℝ) < α - 1 := by linarith -- Multiply h_young by α and simplify have h_scaled : α * ⟪A, B⟫_ℝ ≤ (A ^ α).trace + (α - 1) * (B ^ (α / (α - 1))).trace := by have := mul_le_mul_of_nonneg_left h_young hα_pos.le have h_simp : α * ((A ^ α).trace / α + (B ^ (α / (α - 1))).trace / (α / (α - 1))) = (A ^ α).trace + (α - 1) * (B ^ (α / (α - 1))).trace := by field_simp linarith -- Goal is definitionally: α * ⟪ρ.M, H⟫ - (α-1) * (B ^ (α/(α-1))).trace ≤ (A ^ α).trace -- which follows from h_scaled and h_inner change α * ⟪ρ.M, H⟫_ℝ - (α - 1) * (B ^ (α / (α - 1))).trace ≤ (A ^ α).trace have h_inner_scaled : α * ⟪ρ.M, H⟫_ℝ = α * ⟪A, B⟫_ℝ := by rw [h_inner] linarith [h_scaled, h_inner_scaled]- Project
- quantumInfo
- License
- MIT
- Commit
- 56e83a9288a3
- Source
- QuantumInfo/Finite/Entropy/DPI.lean:598-630
Reuse this declaration
Bring the exact result into your workflow
The import identifies the source module. Your project still needs the pinned package dependency shown on this page.
What this badge means
This completion status comes from the project or community source. It has not yet been represented here as an independent rebuild and axiom audit.
Continue in this project
Related declarations
Conj Transpose isometry mul isometry le one
conjTranspose_isometry_mul_isometry_le_one
Project documentation
The operator norm of the conjugate transpose is equal to the operator norm. -/ theorem Matrix.opNorm_conjTranspose_eq_opNorm {m n : Type*} [Fintype m] [Fintype n] [DecidableEq m] [DecidableEq n] (A : Matrix m n 𝕜) : Matrix.opNorm Aᴴ = Matrix.opNorm A := by unfold Matrix.opNorm rw [← ContinuousLinearMap.adjoint.norm_map (toEuclideanLin A).toContinuousLine...
Source project: quantumInfo
Person-level attribution pending.
Convex roof of pure
convex_roof_of_pure
Plain-language statement
The convex roof extension of g : KetUpToPhase d → ℝ≥0 applied to a pure state ψ is g (KetUpToPhase.mk ψ).
Source project: quantumInfo
Person-level attribution pending.
Id achieves Rate log dim
CPTPMap.id_achievesRate_log_dim
Plain-language statement
The identity channel on D dimensional space achieves a rate of log2(D).
Source project: quantumInfo
Person-level attribution pending.