Physlib

QuantumInfo.ForMathlib.HermitianMat.Peierls

5 declarations

theorem

Unitary invariance of the trace of functional calculus: Tr(g(UAU))=Tr(g(A))\text{Tr}(g(U A U^\dagger)) = \text{Tr}(g(A))

Let dd be a finite type and AA be a d×dd \times d Hermitian matrix over the complex numbers C\mathbb{C}. For any function g:RRg : \mathbb{R} \to \mathbb{R} and any unitary matrix UU of size dd, the trace of the matrix obtained by applying the continuous functional calculus of gg to the unitarily conjugated matrix UAUU A U^\dagger is equal to the trace of the matrix obtained by applying the continuous functional calculus of gg directly to AA. That is, Tr(g(UAU))=Tr(g(A))\text{Tr}(g(U A U^\dagger)) = \text{Tr}(g(A)).

theorem

Peierls Inequality: ig(Re(Aii))Tr(g(A))\sum_i g(\text{Re}(A_{ii})) \le \text{Tr}(g(A))

Let dd be a finite index set and AA be a d×dd \times d Hermitian matrix with complex entries. If g:RRg: \mathbb{R} \to \mathbb{R} is a convex function, then the sum of gg applied to the real parts of the diagonal entries of AA is less than or equal to the trace of the matrix g(A)g(A) (the result of the continuous functional calculus of gg applied to AA): idg(Re(Aii))Tr(g(A))\sum_{i \in d} g(\text{Re}(A_{ii})) \le \text{Tr}(g(A))

theorem

Peierls's inequality for positive semidefinite matrices: ig(Aii)Tr(g(A))\sum_i g(A_{ii}) \le \text{Tr}(g(A))

Let AA be a d×dd \times d complex Hermitian matrix that is positive semidefinite (A0A \ge 0). Let g:RRg: \mathbb{R} \to \mathbb{R} be a function that is convex on the interval [0,)[0, \infty). Then the sum of gg applied to the real parts of the diagonal entries of AA is less than or equal to the trace of the matrix g(A)g(A) obtained via continuous functional calculus: ig(Re(Aii))Tr(g(A))\sum_{i} g(\text{Re}(A_{ii})) \le \text{Tr}(g(A)) Note that since AA is Hermitian, its diagonal entries AiiA_{ii} are already real, so Re(Aii)=Aii\text{Re}(A_{ii}) = A_{ii}.

theorem

Convexity of Atr(g(A))A \mapsto \text{tr}(g(A)) for convex gg

Let g:RRg: \mathbb{R} \to \mathbb{R} be a convex function. The mapping Atr(g(A))A \mapsto \text{tr}(g(A)) is convex on the space of d×dd \times d complex Hermitian matrices, where g(A)g(A) is defined by the continuous functional calculus of gg applied to AA.

theorem

Convexity of ATr(g(A))A \mapsto \text{Tr}(g(A)) on Positive Semidefinite Matrices

Let g:RRg : \mathbb{R} \to \mathbb{R} be a function that is convex on the interval [0,)[0, \infty). Then the map ATr(g(A))A \mapsto \text{Tr}(g(A)), which assigns to each positive semidefinite Hermitian matrix AHermd(C)A \in \text{Herm}_d(\mathbb{C}) the trace of the matrix obtained by applying the functional calculus of gg to AA, is convex on the set of positive semidefinite matrices {AHermd(C)A0}\{A \in \text{Herm}_d(\mathbb{C}) \mid A \ge 0\}.