Numerical Ranges in Modern Times 14th WONRA at Man-Duen Choi

Slides:



Advertisements
Similar presentations
5.4 Basis And Dimension.
Advertisements

CS 450: COMPUTER GRAPHICS LINEAR ALGEBRA REVIEW SPRING 2015 DR. MICHAEL J. REALE.
1 A camera is modeled as a map from a space pt (X,Y,Z) to a pixel (u,v) by ‘homogeneous coordinates’ have been used to ‘treat’ translations ‘multiplicatively’
Quantum Computing MAS 725 Hartmut Klauck NTU
Introduction to Quantum Information Processing CS 467 / CS 667 Phys 467 / Phys 767 C&O 481 / C&O 681 Lecture 11 (2005) Richard Cleve DC 3524
Principal Component Analysis CMPUT 466/551 Nilanjan Ray.
Chapter 5 Orthogonality
3.II.1. Representing Linear Maps with Matrices 3.II.2. Any Matrix Represents a Linear Map 3.II. Computing Linear Maps.
Advanced Computer Architecture Lab University of Michigan Quantum Operations and Quantum Noise Dan Ernst Quantum Noise and Quantum Operations Dan Ernst.
Quantum Error Correction Michele Mosca. Quantum Error Correction: Bit Flip Errors l Suppose the environment will effect error (i.e. operation ) on our.
Chapter Two: Vector Spaces I.Definition of Vector Space II.Linear Independence III.Basis and Dimension Topic: Fields Topic: Crystals Topic: Voting Paradoxes.
The Terms that You Have to Know! Basis, Linear independent, Orthogonal Column space, Row space, Rank Linear combination Linear transformation Inner product.
Advanced Computer Architecture Lab University of Michigan Quantum Noise and Distance Patrick Cassleman More Quantum Noise and Distance Measures for Quantum.
Quantum Mechanics from Classical Statistics. what is an atom ? quantum mechanics : isolated object quantum mechanics : isolated object quantum field theory.
Daniel A. Pitonyak Lebanon Valley College
T he Separability Problem and its Variants in Quantum Entanglement Theory Nathaniel Johnston Institute for Quantum Computing University of Waterloo.
CSEP 590tv: Quantum Computing Dave Bacon June 29, 2005 Today’s Menu Administrivia Complex Numbers Bra’s Ket’s and All That Quantum Circuits.
Linear Algebra and Image Processing
MA5209 Algebraic Topology Wayne Lawton Department of Mathematics National University of Singapore S ,
Presented by: Erik Cox, Shannon Hintzman, Mike Miller, Jacquie Otto, Adam Serdar, Lacie Zimmerman.
1 February 24 Matrices 3.2 Matrices; Row reduction Standard form of a set of linear equations: Chapter 3 Linear Algebra Matrix of coefficients: Augmented.
A Study of The Applications of Matrices and R^(n) Projections By Corey Messonnier.
Signal Processing and Representation Theory Lecture 1.
CHAPTER FIVE Orthogonality Why orthogonal? Least square problem Accuracy of Numerical computation.
Kansas State University Department of Computing and Information Sciences CIS 736: Computer Graphics Wednesday, January 19, 2000 William H. Hsu Department.
Section 4.1 Vectors in ℝ n. ℝ n Vectors Vector addition Scalar multiplication.
KINEMATICS of a ROLLING BALL Wayne Lawton Department of Mathematics National University of Singapore Lecture based on my student’s MSc.
AN ORTHOGONAL PROJECTION
Can the Numerical Range of a Nilpotent Operator be a Disc? Akio Arimoto Department of Mathematics Musashi Institute of Technology.
4 © 2012 Pearson Education, Inc. Vector Spaces 4.4 COORDINATE SYSTEMS.
EECS 274 Computer Vision Affine Structure from Motion.
Review on Graphics Basics. Outline Polygon rendering pipeline Affine transformations Projective transformations Lighting and shading From vertices to.
Properties of Kernels Presenter: Hongliang Fei Date: June 11, 2009.
8.4.2 Quantum process tomography 8.5 Limitations of the quantum operations formalism 量子輪講 2003 年 10 月 16 日 担当:徳本 晋
Quantum Two 1. 2 Angular Momentum and Rotations 3.
Voronoi Diagrams and a Numerical Estimation of a Quantum Channel Capacity 1,2 Kimikazu Kato, 3 Mayumi Oto, 1,4 Hiroshi Imai, and 5 Keiko Imai 1 Department.
Ch 6 Vector Spaces. Vector Space Axioms X,Y,Z elements of  and α, β elements of  Def of vector addition Def of multiplication of scalar and vector These.
1 Objective To provide background material in support of topics in Digital Image Processing that are based on matrices and/or vectors. Review Matrices.
 Matrix Operations  Inverse of a Matrix  Characteristics of Invertible Matrices …
Lecture XXVII. Orthonormal Bases and Projections Suppose that a set of vectors {x 1,…,x r } for a basis for some space S in R m space such that r  m.
Quantum Two 1. 2 Angular Momentum and Rotations 3.
Twirling states to simplify separability
Postulates of Quantum Mechanics
MATH2999 Directed Studies in Mathematics Quantum Process on One Qubit System Name: Au Tung Kin UID:
Matrices and Vectors Review Objective
Linear Transformations
ECE 638: Principles of Digital Color Imaging Systems
Quantum One.
Jianping Fan Dept of CS UNC-Charlotte
Quantum Two.
ECE 638: Principles of Digital Color Imaging Systems
Quantum One.
Lecture on Linear Algebra
Principal Component Analysis
Chap 4 Quantum Circuits: p
3D Transformation CS380: Computer Graphics Sung-Eui Yoon (윤성의)
Chapter 3 Linear Algebra
Objective To provide background material in support of topics in Digital Image Processing that are based on matrices and/or vectors.
Affine Spaces Def: Suppose
Quantum Two.
I.4 Polyhedral Theory (NW)
CSCI B609: “Foundations of Data Science”
Quantum Foundations Lecture 3
Linear Transformations
I.4 Polyhedral Theory.
On a Geometric Structure of Pure Multi-qubit Quantum States and Its Applicability to a Numerical Computation 1,2Kimikazu Kato, 3Mayumi Oto, 1,4Hiroshi.
Vector Spaces RANK © 2012 Pearson Education, Inc..
Linear Vector Space and Matrix Mechanics
THE DIMENSION OF A VECTOR SPACE
A quantum machine learning algorithm based on generative models
Presentation transcript:

Numerical Ranges in Modern Times 14th WONRA at Man-Duen Choi choi@math.toronto.edu 14th WONRA at

In the last century, Toeplitz-Hausdorff Theorem, … Ando-Arveson Dilation Theorem, … In the 21st century of Quantum Computer

What a great impact if Quantum Channels look like

Quantum Computer with 2x2 matrices A qubit (quantum bit) is a unit of quantum computer memory. Mathematically, each 1-qubit is regarded as an element in = {all 2 X 2 rank-1 projection matrices} = {all vector states acting on C2 } = {one-dimensional complex linear subspaces of C2 } = {special two-dimensional real subspaces of R4}

with misleading pictures?

Thus, S2 need not be a material surface. Physically, a 1-qubit is a superposition of the spherical surface (called the Bloch sphere), because an “electron” can move freely to any direction from the origin of R3. Thus, S2 need not be a material surface. There are uncountably many 1-qubits, to make S2 symmetry with continuity for approximation, while 1-bits are just the north pole and the south pole.

--- Million points of codes are not feasible by construction May try to achieve the effect of 1-qubits with conventional computer, using million sensitive codes on a spherical surface, but ---How to put millions points uniformly on S2 --- Million points of codes are not feasible by construction --- Million points are still discrete and independent --- No way of approximation in terms of continuity.

However, S2 is a mathematical simple object However, S2 is a mathematical simple object. The combinatorial effect of S2 symmetry is not comparable to 2n when n > 30 . So, a 1-qubit computer cannot replace the conventional computer, as used in digital photo and music. But, we should look further in n-qubit computers, with physical meanings and mathematical ideas in terms of continuity and dimensions and analogs. Def. An n-qubit (= a vector state) is regarded as a 1-dimensional complex linear subspace of the dim 2n Hilbert space, which can also be identified as a rank-1 projection in the form as a 2n X 2n complex matrix.

Usual computer produces a finite 0-1 sequences. 1-qubit computer produces 2 x 2 projections, as photons (instead of electrons that cannot be isolated) realized by laser. A more general quantum computer can produces larger-size matrices.

Numerical Ranges through Quantum Computers For an n x n matrix T, the numerical range W(T) = {x*Tx: xεCn , ‖x ‖ = 1} ={ trace Txx*: ||x|| =1} ={trace TP: for all rank-1 projections P}. Thus the numerical range W(T) is the BASIC feature of the matrix T, as needed/provided for quantum information.

What On earth Does it mean? How could laser be shoot to produce the numerical range of a matrix yields What On earth Does it mean? It means to be quantum information, as a direction to shoot laser, or a 2x2 rank-1 projection .

→ Concrete Example: Consider T = 2 3 0 −2 The numerical range of T is the effect of laser that yields an elliptical disc of major axis of length 5 and minor axis of length 3, with foci at 2 and -2. → This is a two-to-one mapping except from the equator. The equator → the boundary of W(T). Two poles → the center of the elliptical disc Four eigen projections → two foci

deep featureS of 2 x 2 matrices: W(T) determines T up to unitary equivalence. In other words, each ellipse as W(T) determines the norm of T, and the eigenvalues of T . The norm structure is built in hard meaning.

Density Matrices In the formal setting of non-commutative probability, the random position of an n-qubit can be regarded as a density matrix, to be defined as a convex combination of rank-1 projections in M2n. Def: A density matrix is a positive semidefinite matrix of trace 1. Thus, each 2 x 2 density matrix is expressed as with x2 +y2 + z2 < 1; so all density matrices fill up the whole solid sphere with S2 as boundary. Nevertheless, for the case n >2, there is no geometrical picture for the collection of all n x n density matrices.

QUERY: Munich is a subset of the surface of the earth QUERY: Munich is a subset of the surface of the earth. What is a random point of Munich? Answer: A point in the solid sphere, not just a point on the surface. Similarly, I went to 23 places on the surface of earth. Then my average position is a point of the solid earth. Could this point of the solid earth provide you some useful information about my whole locus? Most geographers /computer scientists / experimental physicists could never understand the meaning of random position. How to define the average of projective spaces?

Numerical Ranges as elliptical discs Let  : M2  C be a linear map. Then  restricted to {2 x 2 density matrices} is an affine map (preserving the convex structure). As {2 x 2 density matrices} is same as the solid ball as a convex set, each affine map will send the solid ball onto an elliptical disc of the complex plane. But, the affine image of S2 is an elliptical disc. So the map  also sends all rank-1 projections onto an elliptical disc.

Numerical Ranges as related to density matrices For an n x n matrix T, the numerical range W(T) = {x*Tx: xεCn , ‖x ‖ = 1} ={ trace Txx*: ||x|| =1} ={trace TP: for all rank-1 projections P}. =(trace TR: for all density matrices}. Thus the numerical range W(T) , as the BASIC feature of the matrix T, justifies the need of density matrices for qubits, the need of state space for pure states

Unital Positive Linear Maps vs Numerical Ranges Usually, no way to describe of positive linear maps between 3 x 3 matrices, but there is a graphical way: . Key Fact: If W(T) is a subset of W(S), then the unital linear map sending I to I and S to T is a positive linear map from the span of {I, S, S*} onto the span of {I,T, T*}.

Affine Transforms induces Linear Maps Mn = Mn+ - Mn+ + i Mn+ – i Mn+ Mn+ = R+ x { density matrices} {affine transforms on density matrices} ≈ {trace-preserving positive linear maps}. {affine transforms on density matrices, fixing scalar matrices } ≈ {unital trace-preserving positive linear maps}. {feasible affine transforms on density matrices} ≈ {trace-preserving completely positive linear maps}. 22

Completely positive linear maps Completely positive linear maps serves better than positive linear maps in structure theory as shown in Thm: (Choi, 1975) A linear map : Mn  Mk is completely positive iff [(Eij)] is positive iff when  is of the form (A) =  Vj*AVj for all A in Mn . More details for the structure theory. (This 1975 paper has been cited in more than 1650 research papers, as of 2018 June Google Scholars.)

Completely positive linear maps in circuit theory Each transformer defines a positive linear map A --> V*AV . Thus several transformers in series define a completely positive linear maps. Main question in circuit theory: Besides transformers, what else can make a positive linear map? Concrete linear maps of mathematical expressions are not implementable.

Non-normal Dilations in the easiest setting of the Stinespring Theorem Thm: (Ando,Arveson) If the numerical range of T is a subset of the closed unit disc, then T can be dilated to The proof is non-constructive (not enough for the need of quantum information). Major Problem: Given T, how to get concrete A, B, C so that is unitarily equivalent to

Quantum Channels as Trace-preserving completely positive linear maps A linear map  : Mn  Mn is a quantum channel when  is of the form (A) =  Vj*AVj for all A in Mn with VjVj* = I . Open Question: Let T be an n x n matrix. What are (T) for all possible quantum channels ? Partial answer: Suppose T is positive. Then All (T) are precisely all positive n x n matrices of the same trace as T.