Trace function convex ici
HermitianMat.trace_function_convex_ici
Plain-language statement
Convexity of trace functions: if g is convex on ℝ₊, then A ↦ Tr[g(A)] is convex on PSD matrices. This is Theorem 2.10 of Carlen.
Exact Lean statement
theorem trace_function_convex_ici {g : ℝ → ℝ} (hg : ConvexOn ℝ (Set.Ici 0) g) :
ConvexOn ℝ {A : HermitianMat d ℂ | 0 ≤ A} (fun A => (A.cfc g).trace)Formal artifact
Lean source
theorem trace_function_convex_ici {g : ℝ → ℝ} (hg : ConvexOn ℝ (Set.Ici 0) g) : ConvexOn ℝ {A : HermitianMat d ℂ | 0 ≤ A} (fun A => (A.cfc g).trace) := by refine ⟨convex_Ici 0, ?_⟩ intro A hA B hB a b ha hb hab; -- Let $C = aA + bB$. set C : HermitianMat d ℂ := a • A + b • B; -- By the properties of the trace and the convexity of $g$, we have: have h_trace : (C.cfc g).trace = ∑ i, g (C.H.eigenvalues i) := by exact trace_cfc_eq C g; -- By the properties of the trace and the convexity of $g$, we have $\sum_{i} g(C_{ii}) \leq a \sum_{i} g(A_{ii}) + b \sum_{i} g(B_{ii})$. have h_sum : ∑ i, g (C.H.eigenvalues i) ≤ a * ∑ i, g ((A.conj (star C.H.eigenvectorUnitary.val)).mat i i |> Complex.re) + b * ∑ i, g ((B.conj (star C.H.eigenvectorUnitary.val)).mat i i |> Complex.re) := by have h_sum : ∀ i, g (C.H.eigenvalues i) ≤ a * g ((A.conj (star C.H.eigenvectorUnitary.val)).mat i i |> Complex.re) + b * g ((B.conj (star C.H.eigenvectorUnitary.val)).mat i i |> Complex.re) := by intro i have h_eigenvalue : C.H.eigenvalues i = a * ((A.conj (star C.H.eigenvectorUnitary.val)).mat i i |> Complex.re) + b * ((B.conj (star C.H.eigenvectorUnitary.val)).mat i i |> Complex.re) := by have h_eigenvalue : (C.conj (star C.H.eigenvectorUnitary.val)).mat i i = a * (A.conj (star C.H.eigenvectorUnitary.val)).mat i i + b * (B.conj (star C.H.eigenvectorUnitary.val)).mat i i := by simp +zetaDelta only [mat_add, mat_smul, map_add, mat_apply] simp only [conj, AddMonoidHom.coe_mk, ZeroHom.coe_mk, mat_smul, Algebra.mul_smul_comm, Algebra.smul_mul_assoc] rfl have h_eigenvalue : (C.conj (star C.H.eigenvectorUnitary.val)) = (diagonal ℂ C.H.eigenvalues).conj 1 := by have h_eigenvalue : C = (diagonal ℂ C.H.eigenvalues).conj C.H.eigenvectorUnitary := by exact eq_conj_diagonal C; convert congr_arg ( fun x => ( conj ( star C.H.eigenvectorUnitary.val ) ) x ) h_eigenvalue using 1; simp [ HermitianMat.conj_conj ]; simp_all [ HermitianMat.conj ]; convert congr_arg Complex.re ‹ ( diagonal ℂ _ ) i i = _ › using 1; · exact Eq.symm ( by erw [ show ( diagonal ℂ _ : HermitianMat d ℂ ) i i = ( C.H.eigenvalues i : ℂ ) by exact if_pos rfl ] ; norm_cast ); · norm_num [ Complex.ext_iff ]; rw [h_eigenvalue] refine hg.2 ?_ ?_ ha hb hab · simp exact (Complex.le_def.mp (((zero_le_iff.mp (conj_nonneg _ hA)).diag_nonneg (i := i)))).1 · simp exact (Complex.le_def.mp (((zero_le_iff.mp (conj_nonneg _ hB)).diag_nonneg (i := i)))).1 simpa only [ Finset.mul_sum _ _ _, Finset.sum_add_distrib ] using Finset.sum_le_sum fun i _ => h_sum i; -- By the properties of the trace and the convexity of $g$, we have $\sum_{i} g(A_{ii}) \leq \text{tr}(g(A))$ and $\sum_{i} g(B_{ii}) \leq \text{tr}(g(B))$. have h_trace_A : ∑ i, g ((A.conj (star C.H.eigenvectorUnitary.val)).mat i i |> Complex.re) ≤ (A.cfc g).trace := by have hA' : 0 ≤ A.conj (star C.H.eigenvectorUnitary.val) := A.conj_nonneg _ hA calc ∑ i, g ((A.conj (star C.H.eigenvectorUnitary.val)).mat i i |> Complex.re) ≤ ((A.conj (star C.H.eigenvectorUnitary.val)).cfc g).trace := peierls_inequality_ici _ _ hg hA' _ = (A.cfc g).trace := trace_cfc_conj_unitary A g ⟨star C.H.eigenvectorUnitary.val, by rw [Matrix.mem_unitaryGroup_iff, star_star]; exact C.H.eigenvectorUnitary.prop.1⟩ have h_trace_B : ∑ i, g ((B.conj (star C.H.eigenvectorUnitary.val)).mat i i |> Complex.re) ≤ (B.cfc g).trace := by have hB' : 0 ≤ B.conj (star C.H.eigenvectorUnitary.val) := B.conj_nonneg _ hB calc ∑ i, g ((B.conj (star C.H.eigenvectorUnitary.val)).mat i i |> Complex.re) ≤ ((B.conj (star C.H.eigenvectorUnitary.val)).cfc g).trace := peierls_inequality_ici _ _ hg hB' _ = (B.cfc g).trace := trace_cfc_conj_unitary B g ⟨star C.H.eigenvectorUnitary.val, by rw [Matrix.mem_unitaryGroup_iff, star_star]; exact C.H.eigenvectorUnitary.prop.1⟩ simpa only [ h_trace ] using h_sum.trans ( add_le_add ( mul_le_mul_of_nonneg_left h_trace_A ha ) ( mul_le_mul_of_nonneg_left h_trace_B hb ) )- Project
- quantumInfo
- License
- MIT
- Commit
- 56e83a9288a3
- Source
- QuantumInfo/Finite/Entropy/DPI.lean:285-337
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.