Physlib

QuantumInfo.ForMathlib.HayataGroup.TraceInequality.LiebAndoTrace

14 declarations

instance

The opposite algebra B(H)op\mathcal{B}(\mathcal{H})^{op} has non-negative spectra for its positive elements.

Let H\mathcal{H} be a complex Hilbert space and B(H)\mathcal{B}(\mathcal{H}) be the algebra of bounded linear operators on H\mathcal{H}. Let B(H)op\mathcal{B}(\mathcal{H})^{op} denote the multiplicative opposite algebra of B(H)\mathcal{B}(\mathcal{H}). This theorem states that B(H)op\mathcal{B}(\mathcal{H})^{op} is a non-negative spectrum class over R\mathbb{R}, which implies that for any positive element in the algebra, its spectrum is contained in the interval [0,)[0, \infty).

instance

B(H)opB(\mathcal{H})^{op} has an Isometric Continuous Functional Calculus for Star-Normal Elements over C\mathbb{C}

Let H\mathcal{H} be a complex inner product space and B(H)B(\mathcal{H}) be the space of bounded linear operators on H\mathcal{H}. The multiplicative opposite algebra B(H)opB(\mathcal{H})^{op} admits an isometric continuous functional calculus over the complex numbers C\mathbb{C} for star-normal elements.

instance

B(H)op\mathcal{B}(\mathcal{H})^{op} has a continuous functional calculus for self-adjoint elements over R\mathbb{R}

Let H\mathcal{H} be a complex Hilbert space and let B(H)\mathcal{B}(\mathcal{H}) denote the algebra of bounded linear operators on H\mathcal{H}. The opposite algebra B(H)op\mathcal{B}(\mathcal{H})^{op} admits a continuous functional calculus for self-adjoint elements with respect to real-valued continuous functions.

definition

Real part of the trace Re(Tr(T))\text{Re}(\text{Tr}(T))

Given a bounded linear operator TB(H)T \in \mathcal{B}(\mathcal{H}) on a finite-dimensional complex inner product space H\mathcal{H}, this function returns the real part of its trace, denoted as Re(Tr(T))\text{Re}(\text{Tr}(T)).

definition

Lieb trace map Re(Tr(AsKB1sK))\text{Re}(\text{Tr}(A^s K^* B^{1-s} K))

Given a complex inner product space H\mathcal{H} and the space of bounded linear operators B(H)\mathcal{B}(\mathcal{H}), this function takes a real number sRs \in \mathbb{R} and three bounded linear operators K,A,BB(H)K, A, B \in \mathcal{B}(\mathcal{H}), and returns the real part of the trace of the operator product AsKB1sKA^s K^* B^{1-s} K, denoted as: Re(Tr(AsKB1sK)) \text{Re}(\text{Tr}(A^s K^* B^{1-s} K)) where KK^* is the adjoint of the operator KK. This functional is the primary trace functional utilized in Lieb's concavity theorem.

definition

Trace functional for Lieb's extension theorem Re(Tr(AqKBpK))\text{Re}(\text{Tr}(A^q K^* B^p K))

Let H\mathcal{H} be a finite-dimensional complex inner product space and B(H)\mathcal{B}(\mathcal{H}) be the space of bounded linear operators on H\mathcal{H}. For real numbers q,pRq, p \in \mathbb{R} and operators K,A,BB(H)K, A, B \in \mathcal{B}(\mathcal{H}), this functional is defined as the real part of the trace: Re(Tr(AqKBpK))\text{Re}(\text{Tr}(A^q K^* B^p K)) where KK^* denotes the adjoint of the operator KK, and the operator powers AqA^q and BpB^p are defined via the continuous functional calculus. This functional appears in the context of Lieb's extension theorem.

definition

Real part of the Lieb-Ando trace functional Re(Tr(AqKB1rK))\text{Re}(\text{Tr}(A^q K^* B^{1-r} K))

Given real numbers qq and rr, and bounded linear operators A,B,KB(H)A, B, K \in \mathcal{B}(\mathcal{H}) on a complex inner product space H\mathcal{H}, this function computes the real part of the trace of the operator product AqKB1rKA^q K^* B^{1-r} K, where KK^* is the adjoint of KK. The powers of the operators AqA^q and B1rB^{1-r} are defined via the continuous functional calculus. This specific functional appears in the context of Lieb's and Ando's trace inequalities (specifically Corollary 1.3).

definition

Ando trace map Re(Tr(AqKBrK))\text{Re}(\text{Tr}(A^q K^* B^{-r} K))

Given a complex inner product space H\mathcal{H}, let B(H)\mathcal{B}(\mathcal{H}) denote the space of bounded linear operators on H\mathcal{H}. For real numbers q,rRq, r \in \mathbb{R} and operators K,A,BB(H)K, A, B \in \mathcal{B}(\mathcal{H}), this function calculates the real part of the trace of the operator product AqKBrKA^q K^* B^{-r} K. The expression is given by: Re(Tr(AqKBrK)) \text{Re}\left(\text{Tr}\left(A^q K^* B^{-r} K\right)\right) where KK^* is the adjoint of KK, and the powers AqA^q and BrB^{-r} are defined via the continuous functional calculus. This functional appears in the context of Ando's convexity theorem.

theorem

Convex combinations preserve strictly positive operators A,B>0A, B > 0

Let H\mathcal{H} be a complex inner product space and B(H)\mathcal{B}(\mathcal{H}) be the space of bounded linear operators on H\mathcal{H}. Let AA and BB be strictly positive operators in B(H)\mathcal{B}(\mathcal{H}) (meaning they are self-adjoint and their spectra are contained in (0,)(0, \infty)). For any real number tt such that 0t10 \le t \le 1, the convex combination (1t)A+tB(1 - t)A + tB is also a strictly positive operator.

theorem

Joint concavity of the Lieb trace map for 0<s<10 < s < 1

Let H\mathcal{H} be a complex inner product space and B(H)\mathcal{B}(\mathcal{H}) be the space of bounded linear operators on H\mathcal{H}. Let PB(H)\mathcal{P} \subseteq \mathcal{B}(\mathcal{H}) be the set of strictly positive operators (self-adjoint operators with spectrum in (0,)(0, \infty)). For any real number ss such that 0<s<10 < s < 1 and any operator KB(H)K \in \mathcal{B}(\mathcal{H}), the Lieb trace map (A,B)Re(Tr(AsKB1sK))(A, B) \mapsto \text{Re}(\text{Tr}(A^s K^* B^{1-s} K)) is jointly concave on P×P\mathcal{P} \times \mathcal{P}. That is, for all A1,A2,B1,B2PA_1, A_2, B_1, B_2 \in \mathcal{P} and θ[0,1]\theta \in [0, 1], the following inequality holds: (1θ)Re(Tr(A1sKB11sK))+θRe(Tr(A2sKB21sK))Re(Tr(((1θ)A1+θA2)sK((1θ)B1+θB2)1sK)) (1 - \theta) \text{Re}(\text{Tr}(A_1^s K^* B_1^{1-s} K)) + \theta \text{Re}(\text{Tr}(A_2^s K^* B_2^{1-s} K)) \le \text{Re}(\text{Tr}(((1 - \theta) A_1 + \theta A_2)^s K^* ((1 - \theta) B_1 + \theta B_2)^{1-s} K))

theorem

Re(Tr(AsKB1sK))\text{Re}(\text{Tr}(A^s K^* B^{1-s} K)) is Jointly Convex for 1s21 \le s \le 2

Let H\mathcal{H} be a complex inner product space and L(H)L(\mathcal{H}) be the space of bounded linear operators on H\mathcal{H}. Let P+\mathcal{P}_+ be the set of strictly positive operators in L(H)L(\mathcal{H}), which are self-adjoint operators with spectrum contained in (0,)(0, \infty). For any real number ss such that 1s21 \le s \le 2 and any fixed operator KL(H)K \in L(\mathcal{H}), the Lieb trace map defined by (A,B)Re(Tr(AsKB1sK)) (A, B) \mapsto \text{Re}(\text{Tr}(A^s K^* B^{1-s} K)) is jointly convex for A,BP+A, B \in \mathcal{P}_+. That is, for all A1,A2,B1,B2P+A_1, A_2, B_1, B_2 \in \mathcal{P}_+ and θ[0,1]\theta \in [0, 1], the inequality Φ((1θ)A1+θA2,(1θ)B1+θB2)(1θ)Φ(A1,B1)+θΦ(A2,B2) \Phi((1 - \theta)A_1 + \theta A_2, (1 - \theta)B_1 + \theta B_2) \le (1 - \theta)\Phi(A_1, B_1) + \theta\Phi(A_2, B_2) holds, where Φ(A,B)=Re(Tr(AsKB1sK))\Phi(A, B) = \text{Re}(\text{Tr}(A^s K^* B^{1-s} K)).

theorem

Joint concavity of the Lieb extension trace map (A,B)Re(Tr(AqKBpK))(A, B) \mapsto \text{Re}(\text{Tr}(A^q K^* B^p K)) for p,q>0p, q > 0 and p+q1p + q \le 1

Let H\mathcal{H} be a finite-dimensional complex inner product space and L(H)L(\mathcal{H}) be the space of bounded linear operators on H\mathcal{H}. Let PL(H)\mathcal{P} \subset L(\mathcal{H}) be the set of strictly positive operators (self-adjoint operators with strictly positive spectrum). For any real numbers p,qp, q such that p>0,q>0p > 0, q > 0, and p+q1p + q \le 1, and for any fixed operator KL(H)K \in L(\mathcal{H}), the map (A,B)Re(Tr(AqKBpK)) (A, B) \mapsto \text{Re}(\text{Tr}(A^q K^* B^p K)) is jointly concave for A,BPA, B \in \mathcal{P}. That is, for any A1,A2,B1,B2PA_1, A_2, B_1, B_2 \in \mathcal{P} and θ[0,1]\theta \in [0, 1], (1θ)Re(Tr(A1qKB1pK))+θRe(Tr(A2qKB2pK))Re(Tr(((1θ)A1+θA2)qK((1θ)B1+θB2)pK)). (1 - \theta) \text{Re}(\text{Tr}(A_1^q K^* B_1^p K)) + \theta \text{Re}(\text{Tr}(A_2^q K^* B_2^p K)) \le \text{Re}(\text{Tr}(( (1 - \theta) A_1 + \theta A_2 )^q K^* ( (1 - \theta) B_1 + \theta B_2 )^p K)).

theorem

Joint Convexity of (A,B)Re(Tr(AqKBrK))(A, B) \mapsto \text{Re}(\text{Tr}(A^q K^* B^{-r} K)) for 1q21 \le q \le 2 and 0r10 \le r \le 1 with qr1q-r \ge 1

Let H\mathcal{H} be a complex inner product space and B(H)\mathcal{B}(\mathcal{H}) be the space of bounded linear operators on H\mathcal{H}. Let P+\mathcal{P}_+ denote the set of strictly positive operators in B(H)\mathcal{B}(\mathcal{H}), consisting of self-adjoint operators whose spectrum is contained in (0,)(0, \infty). For any real numbers qq and rr satisfying the conditions 1q21 \le q \le 2, 0r10 \le r \le 1, and qr1q - r \ge 1, and for any fixed operator KB(H)K \in \mathcal{B}(\mathcal{H}), the Ando trace map defined by (A,B)Re(Tr(AqKBrK))(A, B) \mapsto \text{Re}(\text{Tr}(A^q K^* B^{-r} K)) is jointly convex on P+×P+\mathcal{P}_+ \times \mathcal{P}_+. That is, for all A1,A2,B1,B2P+A_1, A_2, B_1, B_2 \in \mathcal{P}_+ and θ[0,1]\theta \in [0, 1], the following inequality holds: Re(Tr(((1θ)A1+θA2)qK((1θ)B1+θB2)rK))(1θ)Re(Tr(A1qKB1rK))+θRe(Tr(A2qKB2rK)).\text{Re}(\text{Tr}(((1 - \theta)A_1 + \theta A_2)^q K^* ((1 - \theta)B_1 + \theta B_2)^{-r} K)) \le (1 - \theta)\text{Re}(\text{Tr}(A_1^q K^* B_1^{-r} K)) + \theta \text{Re}(\text{Tr}(A_2^q K^* B_2^{-r} K)).

theorem

Joint Convexity of (A,B)Re(Tr(AqKB1rK))(A, B) \mapsto \text{Re}(\text{Tr}(A^q K^* B^{1-r} K)) for 1<rq21 < r \le q \le 2

Let H\mathcal{H} be a complex inner product space and B(H)\mathcal{B}(\mathcal{H}) be the space of bounded linear operators on H\mathcal{H}. Let P+B(H)\mathcal{P}_+ \subset \mathcal{B}(\mathcal{H}) denote the set of strictly positive operators (self-adjoint operators whose spectrum is contained in (0,)(0, \infty)). For any real numbers qq and rr such that 1<rq21 < r \le q \le 2, and for any fixed operator KB(H)K \in \mathcal{B}(\mathcal{H}), the functional (A,B)Re(Tr(AqKB1rK))(A, B) \mapsto \text{Re}(\text{Tr}(A^q K^* B^{1-r} K)) is jointly convex for A,BP+A, B \in \mathcal{P}_+. Here, KK^* denotes the adjoint of KK, and the operator powers AqA^q and B1rB^{1-r} are defined via the continuous functional calculus.