Relative entropy as a tensor perspective
ProvedTroppMatrixConcentration.ch8_joint_tensor_representationmatrix-analysisrelative-entropy
For positive-definite complex matrices of the same dimension,
where and is the matrix perspective. This identifies relative entropy with a positive functional applied to a matrix perspective.
Preamble
import Definitions.Def_TroppMatrixConcentration_ch8_joint_tensor open scoped Matrix.Norms.L2Operator MatrixOrder ComplexOrder
Formal statement
namespace TroppMatrixConcentration
theorem ch8_joint_tensor_representation {d : ℕ} [NeZero d]
(A H : Matrix (Fin d) (Fin d) ℂ) (hA : A.PosDef) (hH : H.PosDef) :
ch8_relativeEntropy A H =
ch8_joint_eval (ch8_perspective (fun x => -Real.log x)
(ch8_joint_left A) (ch8_joint_right H)) - (Matrix.trace (A - H)).re := by sorry
end TroppMatrixConcentration
Source
Tropp, An Introduction to Matrix Concentration Inequalities, arXiv:1501.01571v1, Sections 8.7–8.8, equation (8.8.1); uses the transpose required by complex vectorization conventions.