Takeaki Yamazaki (Toyo Univ.) 山崎丈明 ( 東洋大学 ) Oct. 24, RIMS

Σχετικά έγγραφα
Chapter 6: Systems of Linear Differential. be continuous functions on the interval

A Two-Sided Laplace Inversion Algorithm with Computable Error Bounds and Its Applications in Financial Engineering

Chapter 6: Systems of Linear Differential. be continuous functions on the interval

A General Note on δ-quasi Monotone and Increasing Sequence

Ordinal Arithmetic: Addition, Multiplication, Exponentiation and Limit

Reminders: linear functions

5. Choice under Uncertainty

Uniform Convergence of Fourier Series Michael Taylor

Riemannian metrics on positive definite matrices related to means (joint work with Dénes Petz)

2 Composition. Invertible Mappings

Lecture 21: Properties and robustness of LSE

SCITECH Volume 13, Issue 2 RESEARCH ORGANISATION Published online: March 29, 2018

Fourier Series. MATH 211, Calculus II. J. Robert Buchanan. Spring Department of Mathematics

Απόκριση σε Μοναδιαία Ωστική Δύναμη (Unit Impulse) Απόκριση σε Δυνάμεις Αυθαίρετα Μεταβαλλόμενες με το Χρόνο. Απόστολος Σ.

Math 446 Homework 3 Solutions. (1). (i): Reverse triangle inequality for metrics: Let (X, d) be a metric space and let x, y, z X.

What happens when two or more waves overlap in a certain region of space at the same time?

Problem Set 3: Solutions

Lecture 2: Dirac notation and a review of linear algebra Read Sakurai chapter 1, Baym chatper 3

New bounds for spherical two-distance sets and equiangular lines

DiracDelta. Notations. Primary definition. Specific values. General characteristics. Traditional name. Traditional notation

Coefficient Inequalities for a New Subclass of K-uniformly Convex Functions

Homomorphism in Intuitionistic Fuzzy Automata

Tridiagonal matrices. Gérard MEURANT. October, 2008

Inverse trigonometric functions & General Solution of Trigonometric Equations

Divergence for log concave functions

2. Let H 1 and H 2 be Hilbert spaces and let T : H 1 H 2 be a bounded linear operator. Prove that [T (H 1 )] = N (T ). (6p)

Congruence Classes of Invertible Matrices of Order 3 over F 2

Other Test Constructions: Likelihood Ratio & Bayes Tests

Heisenberg Uniqueness pairs

Statistical Inference I Locally most powerful tests

Meta-Learning and Universality

PROPERTIES OF CERTAIN INTEGRAL OPERATORS. a n z n (1.1)

Some new generalized topologies via hereditary classes. Key Words:hereditary generalized topological space, A κ(h,µ)-sets, κµ -topology.

Access Control Encryption Enforcing Information Flow with Cryptography

EXISTENCE OF POSITIVE SOLUTIONS FOR SINGULAR FRACTIONAL DIFFERENTIAL EQUATIONS

Finite Field Problems: Solutions

The k-α-exponential Function

AN ESTIMATE OF QUASI-ARITHMETIC MEAN FOR CONVEX FUNCTIONS. Received March 25, 2011

ΕΠΑΝΑΛΗΠΤΙΚΟ ΔΙΑΓΩΝΙΣΜΑ ΜΑΘΗΜΑΤΙΚΩΝ ΓΕΝΙΚΗΣ ΠΑΙΔΕΙΑΣ Γ ΛΥΕΙΟΥ

ST5224: Advanced Statistical Theory II

Nowhere-zero flows Let be a digraph, Abelian group. A Γ-circulation in is a mapping : such that, where, and : tail in X, head in

MINIMAL CLOSED SETS AND MAXIMAL CLOSED SETS

k A = [k, k]( )[a 1, a 2 ] = [ka 1,ka 2 ] 4For the division of two intervals of confidence in R +

Iterated trilinear fourier integrals with arbitrary symbols

1. Introduction and Preliminaries.

Matrices and Determinants

Example Sheet 3 Solutions

F A S C I C U L I M A T H E M A T I C I

The ε-pseudospectrum of a Matrix

= λ 1 1 e. = λ 1 =12. has the properties e 1. e 3,V(Y

Partial Differential Equations in Biology The boundary element method. March 26, 2013

SOME PROPERTIES OF FUZZY REAL NUMBERS

EE512: Error Control Coding

SCHOOL OF MATHEMATICAL SCIENCES G11LMA Linear Mathematics Examination Solutions

C.S. 430 Assignment 6, Sample Solutions

Nonlinear Fourier transform for the conductivity equation. Visibility and Invisibility in Impedance Tomography

Intuitionistic Fuzzy Ideals of Near Rings

POSITIVE SOLUTIONS FOR A FUNCTIONAL DELAY SECOND-ORDER THREE-POINT BOUNDARY-VALUE PROBLEM

On New Subclasses of Analytic Functions with Respect to Conjugate and Symmetric Conjugate Points

Practice Exam 2. Conceptual Questions. 1. State a Basic identity and then verify it. (a) Identity: Solution: One identity is csc(θ) = 1

J. of Math. (PRC) Banach, , X = N(T ) R(T + ), Y = R(T ) N(T + ). Vol. 37 ( 2017 ) No. 5

Fractional Colorings and Zykov Products of graphs

Global nonlinear stability of steady solutions of the 3-D incompressible Euler equations with helical symmetry and with no swirl

arxiv: v3 [math.ca] 4 Jul 2013

4.6 Autoregressive Moving Average Model ARMA(1,1)

Homework 8 Model Solution Section

THE SECOND ISOMORPHISM THEOREM ON ORDERED SET UNDER ANTIORDERS. Daniel A. Romano

ORDINAL ARITHMETIC JULIAN J. SCHLÖDER

Numerical Analysis FMN011

Jesse Maassen and Mark Lundstrom Purdue University November 25, 2013

Solution Series 9. i=1 x i and i=1 x i.

Research Article Existence of Positive Solutions for m-point Boundary Value Problems on Time Scales

Arithmetical applications of lagrangian interpolation. Tanguy Rivoal. Institut Fourier CNRS and Université de Grenoble 1

Palestine Journal of Mathematics Vol. 2(1) (2013), Palestine Polytechnic University-PPU 2013

TMA4115 Matematikk 3

Section 8.3 Trigonometric Equations

SOLUTIONS TO MATH38181 EXTREME VALUES AND FINANCIAL RISK EXAM

The Pohozaev identity for the fractional Laplacian

Memoirs on Differential Equations and Mathematical Physics

6.1. Dirac Equation. Hamiltonian. Dirac Eq.

Commutative Monoids in Intuitionistic Fuzzy Sets

Math 248 Homework 1. Edward Burkard. Exercise 1. Prove the following Fourier Transforms where a > 0 and c R: f (x) = b. f(x c) = e.

Econ 2110: Fall 2008 Suggested Solutions to Problem Set 8 questions or comments to Dan Fetter 1

ON NEGATIVE MOMENTS OF CERTAIN DISCRETE DISTRIBUTIONS

Fixed point theorems of φ convex ψ concave mixed monotone operators and applications

Affine Weyl Groups. Gabriele Nebe. Summerschool GRK 1632, September Lehrstuhl D für Mathematik

STRONG DIFFERENTIAL SUBORDINATIONS FOR HIGHER-ORDER DERIVATIVES OF MULTIVALENT ANALYTIC FUNCTIONS DEFINED BY LINEAR OPERATOR

Lecture 15 - Root System Axiomatics

Solutions to Exercise Sheet 5

Μαθηματικοί Διαγωνισμοί για Μαθητές Λυκείου Α ΤΕΥΧΟΣ ΑΛΓΕΒΡΑ

Trigonometric Formula Sheet

arxiv: v1 [math-ph] 4 Jun 2016

STABILITY FOR RAYLEIGH-BENARD CONVECTIVE SOLUTIONS OF THE BOLTZMANN EQUATION

The strong semilattice of π-groups

Generating Set of the Complete Semigroups of Binary Relations

MATHEMATICS. 1. If A and B are square matrices of order 3 such that A = -1, B =3, then 3AB = 1) -9 2) -27 3) -81 4) 81

A Bonus-Malus System as a Markov Set-Chain. Małgorzata Niemiec Warsaw School of Economics Institute of Econometrics

EE434 ASIC & Digital Systems Arithmetic Circuits

Solvability of a Maximum Quadratic Integral Equation of Arbitrary Orders

Lecture 13 - Root Space Decomposition II

Transcript:

Takeaki Yamazaki (Toyo Univ.) 山崎丈明 ( 東洋大学 ) Oct. 24, 2017 @ RIMS

Contents Introduction Generalized Karcher equation Ando-Hiai inequalities Problem

Introduction PP: The set of all positive definite operators on a Hilbert space. Operator mean Let σσ: PP 22 PP. If σσ satisfies the following conditions, then σσ is called an operator mean. 1. σσ AA, BB σσ CC, DD if AA CC and BB DD, 2. XX σσ AA, BB XX σσ(xx AAAA, XX BBBB) for all bounded linear operator XX, 3. σσ is upper semi-continuous on PP 22, 4. σσ II, II = II. F. Kubo and T. Ando, Math. A., 246 (1980), 205-224.

Introduction M: The set of all operator monotone functions on 0,. Representing function of an operator mean Let σσ be an operator mean. Then ff MM such that ff 11 = 11 and for all AA, BB PP. σσ AA, BB = AA 11 22ff AA 11 22BBAA 11 22 AA 11 22 ff xx = σσ(1, xx) F. Kubo and T. Ando, Math. A., 246 (1980), 205-224.

Examples of operator means Weighted geometric mean. For λλ [0,1] and AA, BB PP, a AA λλ BB = AA 1 2 AA 1 2BBAA 1 2 λλ AA 1 2. Representing function is gg λλ xx = xx λλ. Weighted power mean. For λλ 0,1, tt [ 1,1] and AA, BB PP, a PP tt λλ; AA, BB = AA 1 1 2 (1 λλ)i + λλ AA 1 2BBAA 1 2 tt tt 1 AA 2. Representing function is pp tt,λλ xx = 1 λλ + λλxx tt 1 tt.

Contents Introduction Generalized Karcher equation Ando-Hiai inequalities Problem

The generalized Karcher equation (GKE) L = {gg M: gg 1 = 0 aaaaaa gg 1 = 1}, Δ = { ww 1,, ww 0,1 : ww ii = 1} Definition. Let gg LL. For AA = (AA 11,, AA ) PP and ωω = (ww 11,, ww ) ΔΔ, the generalized Karcher equation (GKE) is defined by Theorem P1. ww ii gg(xx 11 22AA ii XX 11 22) = 00. 1 The GKE has a unique solution XX PP. M. Palfia, Adv. Math., 289 (2016), 951 1007

Properties of the solution of GKE Theorem P2. Let gg LL. For AA = (AA 11,, AA ) PP and ωω = (ww 11,, ww ) ΔΔ, the solution of GKE XX = σσ gg (ωω; AA) satisfies the properties. (1) σσ gg ωω; XXAAXX = XXσσ gg ωω; AA XX for XX PP, (2) σσ gg ωω; AA σσ gg ωω; BB if AA ii BB ii, where BB = (BB 11,, BB ) PP, (3) σσ gg ωω; AA σσ ff ωω; AA if gg(xx) ff(xx) for all xx > 00. Especially, σσ gg 11 ww, ww; AA, BB is a ww-weighted operator mean. M. Palfia, Adv. Math., 289 (2016), 951 1007

Examples of a solutions of GKE 1. Let AA = (AA 1,, AA ) PP and ωω = (ww 1,, ww ) Δ. a The Karcher mean XX = Λ(ωω; AA) (gg xx = log xx L). ww ii log(xx 1 2AA ii XX 1 2) = 0 Relative operator entropy SS AA BB = AA 1 2 log(aa 1 2BBAA 1 2) AA 1 2 a a Λ 1 ww, ww; AA 1, AA 2 = AA 1 ww AA 2 Rep. ft. xx ww ww xxww ww=0 = log xx

Examples of a solutions of GKE 2. Let AA = AA 1,, AA PP, ωω = (ww 1,, ww ) Δ and tt 1,1. a The power mean P t (ωω; AA) (gg xx = xxtt 1 1 ww ii XX 1 2AA ii XX 1 2 tt II tt tt L). = 0 a a PP tt 1 ww, ww; AA 1, AA 2 Rep. ft. (1 ww) + wwxx tt 1 tt ww Tsallis Relative operator entropy TT tt AA BB = AA 1 2 1 AA (1 ww) + wwxxtt 2BBAA 1 2 tt II 1 tt tt ww=0 AA 1 2 = xxtt 1 tt

Operator means via GKE. Let AA = (AA 1,, AA ) PP, ωω = (ww 1,, ww ) Δ and gg L. a The solution of GKE XX = σσ gg (ωω; AA). Prop. 1 ww ii gg(xx 1 2AA ii XX 1 2) = 0 Generalized relative operator entropy A a σσ gg 1 ww, ww; AA 1, AA 2 Rep. ft. ff ww (xx) Prop. 2 GG AA BB = AA 1 2gg(AA 1 2BBAA 1 2)AA 1 2 ww ff ww xx ww=00 = gg(xx)

Representing function Let AA, BB PP, ww (0,1), aa > 0 and gg L. GKE 11 ww gg XX 11 22AAXX 11 22 + wwww XX 11 22BBXX 11 22 = 00 Solution (Operator mean) XX = σσ gg (11 ww, ww; AA, BB) GKE 11 ww gg 11 xx + wwww aa Solution (representing function) xx = 00 xx = ff ww aa = σσ gg 11 ww, ww; 11, aa

Representing function L = {gg M: gg 1 = 0 aaaaaa gg 1 = 1} Proposition 1. Let gg LL. For aa > 00, ww (00, 11), and let xx = ff ww (aa) be a solution of 11 ww gg 11 xx + wwww aa xx = 00. Then the inverse function of ff ww is ff ww 11 xx = xxgg 11 11 ww ww gg xx 11. cf. M. Palfia, ArXiv 1208.5603

Proof GKE 11 ww gg 11 xx + wwww aa Solution (representing function) xx = 00 xx = ff ww aa = σσ gg 11 ww, ww; 11, aa 1 ww gg 1 xx + wwww aa xx = 0 gg aa xx = 1 ww ww gg 1 xx ff ww 1 xx = aa = xxgg 1 1 ww ww gg 1 xx

Generalized relative operator entropy Proposition 2. Let gg LL. For aa > 00, ww 00, 11, let xx = ff ww (aa) be a solution of 11 ww gg 11 xx + wwww aa xx = 00. Then ww ff ww(xx) ww=00 = gg xx. J.I. Fujii and Kamei, Math. Japon. 34 (1989), 541 547. Paalfia and Petz, Linear Algebra Appl. 463 (2014), 134 153.

Proof 1 ww gg 1 ff ww (aa) + wwww aa ff ww (aa) = 0 ww 1 ww gg 1 ff ww aa + wwww aa ff ww aa = 0 Let ww = 0, then by gg 1 = 0, gg 1 = 1 and ff 0 xx = 1, ff ww(xx) ww=0 = gg xx.

Operator means via GKE. Let AA = (AA 1,, AA ) PP, ωω = (ww 1,, ww ) Δ and gg L. a The solution of GKE XX = σσ gg (ωω; AA). Prop. 1 ww ii gg(xx 1 2AA ii XX 1 2) = 0 Generalized relative operator entropy A a σσ gg 1 ww, ww; AA 1, AA 2 Rep. ft. ff ww (xx) Prop. 2 GG AA BB = AA 1 2gg(AA 1 2BBAA 1 2)AA 1 2 ww ff ww xx ww=00 = gg(xx)

Contents Introduction Generalized Karcher equation Ando-Hiai inequalities Problem

Ando-Hiai inequality Theorem AH. Let AA, BB PP and ww (00, 11). Then AA ww BB II AA pp ww BB pp II holds for all pp 11. Ando and Hiai, Linear Algebra Appl. 197/198 (1994), 113 131.

Extension of Ando-Hiai inequality 1 Theorem PLY. For AA = (AA 11,, AA ) PP and ωω = (ww 11,, ww ) ΔΔ, a (1) ΛΛ ωω; AA II ΛΛ ωω; AA pp II, (Karcher mean) (2) for tt (00, 11], PP tt ωω; AA II PP tt/pp ωω; AA pp II a hold for all pp 11. (Power mean) Y., Oper. Matrices, 6 (2012), 577 588. M. Palfia and Y. Lim, J. Funct. Anal. 262 (2012), 1498 1514. Y. Lim and Y., Linear Algebra Appl. 438 (2013), 1293 1304.

Extension of Ando-Hiai inequality 2 Theorem W. Let σσ be a operator mean with a representing function ff MM. TFAE (1) ff xx pp ff xx pp for all xx > 00 and pp 11. (2) For AA, BB PP, σσ(aa, BB) II σσ(aa pp, BB pp ) II for all pp 11. All statements hold for, respectively. S. Wada, Linear Algebra Appl. 457 (2014), 276 286.

Motivation Theorem AH. AA ww BB II AA pp ww BB pp II Rep. Function Karcher equation Theorem W. TFAE (1) ff xx pp ff xx pp (2) σσ(aa, BB) II σσ(aa pp, BB pp ) II Theorem PLY. For tt 00, 11, (1) ΛΛ ωω; AA II ΛΛ ωω; AA pp II, (2) PP tt ωω; AA II PP tt/pp ωω; AA pp II GKE and Rep. Function (Generalized relative operator entropy)

PLY type Ando-Hiai inequality L = {gg M: gg 1 = 0 aaaaaa gg 1 = 1} Theorem 1. Let gg LL. For AA = (AA 11,, AA ) PP and ωω = (ww 11,, ww ) ΔΔ, σσ gg ωω; AA II σσ ggpp ωω; AA pp II for all pp 11, where gg pp xx = pppp xx 11/pp LL. Moreover, if ff is a representing function of σσ gg, then ff pp xx = ff xx 11/pp pp is a representing function of σσ ggpp.

A key lemma Lemma. Let gg LL. For AA = (AA 11,, AA ) PP and ωω = (ww 11,, ww ) ΔΔ, for all pp 11. ww ii gg(aa ii ) 00 σσ gg ωω; AA II 1 mm ww ii gg XX 1 2AA ii XX 1 2 = 0 XX = σσ gg (ωω; AA) mm ww ii gg AA ii = 0 σσ gg ωω; AA = II

Proof ww ii gg(aa ii ) 0 XX II; wwii 2 gg AA ii + 1 gg XX = 0 2 (gg 1 = 0) Then σσ gg ωω 2, 1 ; AA, XX = II. 2 AA = (AA 1,, AA ) ωω = (ww 1,, ww ) Let XX 0 = II, XX kk = σσ gg ωω 2, 1 2 ; AA, XX kk 1.

Proof XX 0 = II, XX kk = σσ gg ωω 2, 1 2 ; AA, XX kk 1. Then II = σσ gg ωω 2, 1 2 ; AA, XX σσ gg ωω 2, 1 2 ; AA, II σσ gg ωω 2, 1 2 ; AA, XX 1 XX kk > 0 XX 11 XX 22 Hence XX = lim kk XX kk. XX II II XX 1

Proof It satisfies XX = σσ gg ωω 2, 1 2 ; AA, XX ωω II = σσ gg 2, 1 2 ; XX 12 12 AAXX, II wwii 2 gg XX 12 12 AA ii XX + 1 gg II = 0 2 ww ii gg XX 12 AA ii XX 12 = 0 (gg 1 = 0) Hence XX = σσ gg ωω; AA II.

PLY type Ando-Hiai inequality Theorem 1. Let gg LL. For AA = (AA 11,, AA ) PP and ωω = (ww 11,, ww ) ΔΔ, σσ gg ωω; AA II σσ ggpp ωω; AA pp II for all pp 11, where gg pp xx = pppp xx 11/pp LL. Moreover, if ff is a representing function of σσ gg, then ff pp xx = ff xx 11/pp pp is a representing function of σσ ggpp.

Proof of Theorem 1 In the case 1 pp 2. Hansen s inequality gg pp xx : = pppp xx 1/pp By lemma, Let XX = σσ gg ωω; AA II. Then XX 1 II and 0 = pp ww ii gg XX 1 2AA ii XX 1 2 L = pp ww ii gg XX 1 2AA ii XX 1 2 pp ww ii gg XX 1 2AA pp ii XX 1 2 = ww ii gg pp XX 1 2AA pp ii XX 1 2 Hence σσ ggpp ωω; AA pp XX = σσ gg ωω; AA II holds for all 1 pp 2. ww ii gg pp XX 1 2AA pp ii XX 1 2 0 σσ ggpp ωω; XX 1 2AA pp XX 1 2 II pp pp 1 pp

Proof of Theorem 1 Let ff pp be a representing function of σσ ggpp (1 ww, ww; AA, BB). Then ff pp 1 xx = xxgg pp 1 1 ww ww gg pp xx 1 gg pp xx : = pppp xx 11/pp gg pp 11 xx = gg xx/pp pp = xxgg 1 1 pp 1 ww ww = xx 1/pp gg 1 1 ww ww pppp xx 1/pp gg xx 1/pp pp pp = ff 1 xx1/pp pp Hence ff pp xx = ff xx 1/pp pp.

Wada s type Ando-Hiai inequality Theorem 2. Let gg LL. Then TFAE. A (1) ff ww xx pp ff ww xx pp for all pp 11, xx > 00 and ww 00, 11, a (2) gg xx pp pppp(xx) for all pp 11 and xx > 00, a (3) σσ gg ωω; AA II σσ gg ωω; AA pp II for all pp 11, AA = AA 11,, AA PP aaaaaa ωω = (ww 11,, ww ) ΔΔ, a where σσ gg ωω; AA and ff λλ (xx) mean a solution of GKE and representing function of σσ gg, respectively. All statements hold for, respectively.

Proof of Theorem 2 Proof of (1) (2). By Proposition 1, ff ww 1 xx = xxxx ff ww xx pp ff ww xx pp ff ww 1 xx pp ff ww 1 xx pp. 1 ww ww gg xx 1 = xxxx λλgg xx 1 PPPPPP λλ: = 11 ww ww > 00 gg 1 λλ gg xx 1 pp 1 λλ gg 1 λλλλ xx pp 1 λλ λλ +0 pppp(xx 1 ) gg xx pp Hence gg xx pp pppp(xx) holds for all xx > 0.

Proof of Theorem 2 Proof of (2) (3). Put XX = σσ gg ωω; AA II. Then XX 1 II. The case 1 pp 2. 0 = pp ww ii gg(xx 1 2AA ii XX 1 2) = pp ww ii gg XX 1 2 AA ii XX 1 2 Hansen s inequality (2) By lemma, gg xx pp pppp(xx) pp ww ii gg XX 1 2AA pp ii XX 1 2 ww ii gg(xx 1 2AA pp ii XX 1 2) Hence σσ gg ωω; AA pp XX = σσ gg ωω; AA II holds for all 1 pp 2. ww ii gg XX 1 2AA pp ii XX 1 2 0 σσ gg ωω; XX 1 2AA pp XX 1 2 II. pp pp 1 pp

Proof of Theorem 2 Proof of (3) (1). By (3), for AA, BB PP and ww 0,1, σσ gg 1 ww, ww; AA, BB II σσ gg 1 ww, ww; AA pp, BB pp II. By Theorem W, ff ww xx pp ff ww xx pp.

Results Theorem AH. AA ww BB II AA pp ww BB pp II Rep. Function Karcher equation Theorem W. TFAE (1) ff xx pp ff xx pp (2) σσ(aa, BB) II σσ(aa pp, BB pp ) II Theorem PLY. (1) ΛΛ ωω; AA II ΛΛ ωω; AA pp II, (2) PP tt ωω; AA II PP tt/pp ωω; AA pp II Theorem 2. Theorem 1.

Weighted power mean case tt (00, 11] ff ww xx = 1 ww + wwxx tt 1 tt, gg xx = xxtt 1 tt Theorem PLY. PP tt ωω; AA II PP tt/pp ωω; AA pp II Theorem 2. TFAE. a (1) ff ww xx pp ff ww (xx pp ) for all pp 11. (2) pppp xx gg(xx pp ) for all pp 11. (3) PP tt ωω; AA II PP tt ωω; AA pp II for all pp 11.

Contents Introduction Generalized Karcher equation Ando-Hiai inequalities Problem

Norm inequality Theorem HP. For AA = (AA 11,, AA ) PP and ωω = (ww 11,, ww ) ΔΔ, ΛΛ ωω; AA eexxxx ww ii log AA ii holds for all unitarily invariant norm. 1 If {A 1,, AA } is commuting, then Λ ωω; AA = exp ww ii log AA ii Hiai and Petz, Linear Algebra Appl. 436 (2012), 2117 2136.

Norm inequality Theorem LY-BLY-DDF. For AA = AA 11,, AA PP and ωω = (ww 11,, ww ) ΔΔ, tt PP tt ωω; AA pp ww ii AA ii 1 holds for all Schatten pp-norm for pp 11. 11 tt pp If {A 1,, AA } is commuting, then PP tt ωω; AA = ww ii AA ii tt 1 tt Lim and Y., Linear Algebra Appl. 438 (2013), 1293 1304. Bhatia, Lim and Y., Linear Algebra Appl. 501 (2016), 112 122. Dinh, Dumitru and Franco, Linear Algebra Appl. 532 (2017) 140-145.

Problem Problem. For AA = (AA 11,, AA ) PP and ωω = (ww 11,, ww ) ΔΔ, σσ gg (ωω; AA) gg 11 ww ii gg AA ii 1 holds for any unitarily invariant norm?

A key lemma Lemma. Let gg LL. For AA = (AA 11,, AA ) PP and ωω = (ww 11,, ww ) ΔΔ, for all pp 11. gg 11 ww ii gg AA ii 1 II σσ gg ωω; AA II L = {gg M: gg 1 = 0 aaaaaa gg 1 = 1}, Δ = { ww 1,, ww 0,1 : ww ii = 1} gg 1 ww ii gg AA ii = II σσ gg ωω; AA = II

Problem Let gg L and σσ gg be a solution of GKE. Problem. For AA = (AA 11,, AA ) PP and ωω = (ww 11,, ww ) ΔΔ, σσ gg (ωω; AA) gg 11 ww ii gg AA ii 1 holds for any unitarily invariant norm? Proposition 4. Let AA, BB PP and ww (00, 11). If AAAA = BBBB, then TFAE. 1. σσ gg 11 ww, ww; AA, BB is a power mean. 2. σσ gg 11 ww, ww; AA, BB = gg 11 11 ww gg(aa) + wwww(bb).

Thanks! Thanks for your attention!