Quantum Computing A Gentle Introduction Rieffel Eleanorauthor instant download
Quantum Computing A Gentle Introduction Rieffel Eleanorauthor instant download
Eleanorauthor download
https://ebookbell.com/product/quantum-computing-a-gentle-
introduction-rieffel-eleanorauthor-56399420
https://ebookbell.com/product/quantum-computing-a-pathway-to-quantum-
logic-design-hafiz-md-hasan-babu-49165146
Quantum Computing A Shift From Bits To Qubits 1st Edition Rajiv Pandey
https://ebookbell.com/product/quantum-computing-a-shift-from-bits-to-
qubits-1st-edition-rajiv-pandey-49438170
https://ebookbell.com/product/quantum-computing-a-pathway-to-quantum-
logic-design-2nd-edition-2nd-hafiz-md-hasan-babu-50856750
https://ebookbell.com/product/quantum-computing-a-new-era-of-
computing-kuldeep-singh-kaswan-50994996
Quantum Computing A Pathway To Quantum Logic Design Hafiz Hasan Babu
https://ebookbell.com/product/quantum-computing-a-pathway-to-quantum-
logic-design-hafiz-hasan-babu-37754396
https://ebookbell.com/product/quantum-computing-a-short-course-from-
theory-to-experiment-1st-edition-joachim-stolze-1188372
https://ebookbell.com/product/essential-mathematics-for-quantum-
computing-a-beginners-guide-to-just-the-math-you-need-without-
needless-complexities-leonard-s-woody-iii-42782302
https://ebookbell.com/product/the-brain-fuzzy-arithmetic-to-quantum-
computing-a-f-rocha-eduardo-massad-4126534
https://ebookbell.com/product/highlevel-structures-in-quantum-
computing-jaroslaw-a-miszczak-5214584
QUANTUM COMPUTING
Scientific and Engineering Computation
William Gropp and Ewing Lusk, editors; Janusz Kowalik, founding editor
A complete list of the books in this series can be found at the back of this book.
QUANTUM COMPUTING
A Gentle Introduction
All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including
photocopying, recording, or information storage and retrieval) without permission in writing from the publisher.
This book was set in Syntax and Times Roman by Westchester Book Group. Printed and bound in the United States of
America.
10 9 8 7 6 5 4 3 2 1
Contents
Preface xi
1 Introduction 1
3 Multiple-Qubit Systems 31
3.1 Quantum State Spaces 32
3.1.1 Direct Sums of Vector Spaces 32
3.1.2 Tensor Products of Vector Spaces 33
3.1.3 The State Space of an n-Qubit System 34
3.2 Entangled States 38
3.3 Basics of Multi-Qubit Measurement 41
3.4 Quantum Key Distribution Using Entangled States 43
3.5 References 44
3.6 Exercises 44
APPENDIXES 329
Bibliography 353
Notation Index 365
Index 369
Preface
Quantum computing is a beautiful combination of quantum physics, computer science, and infor-
mation theory. The purpose of this book is to make this exciting research area accessible to a
broad audience. In particular, we endeavor to help the reader bridge the conceptual and notational
barriers that separate quantum computing from conventional computing.
The book is concerned with theory: what changes when the classical model underpinning
conventional computing is replaced with a quantum one. It contains only a brief discussion of
the ongoing efforts to build quantum computers, an active area which is still so young that it is
impossible even for experts to predict which approaches will be most successful. While this book
is about theory, it is important to ground the discussion of quantum computation in the physics that
motivates it. For this reason, the text includes discussions of quantum physics and experiments
that illustrate why the theory is defined the way it is.
We precisely define concepts used in quantum computation and emphasize subtle distinc-
tions. This rigor is motivated in part by our experience working with members of the joint
FXPAL1 /PARC2 reading group and with reviewing papers by authors new to the field. Mistakes
commonly arise due to a lack of precision. For example, we take care to distinguish a quantum
state from a vector that represents it. We make clear which notions are basis dependent (e.g.,
superposition) and which are not (e.g., entanglement), and emphasize the dependence of certain
notions (e.g., entanglement) on a particular tensor decomposition. The distinction between tensor
decompositions and direct sum decompositions, both used extensively in quantum mechanics,
is discussed explicitly in both quantum mechanical and classical probabilistic settings. Defini-
tions are carefully motivated. For example, instead of starting with axioms for density operators
or mixed states, the definitions of these concepts are motivated by a discussion of what can be
deduced about a subsystem from measurements of the subsystem alone.
One advantage of dealing only with theory, and not with the efforts to build quantum computers,
is that the amount of quantum physics and supporting mathematics needed is reduced. We are
able to develop all of the necessary quantum mechanics within the book; no previous exposure to
quantum physics is required. We give careful and precise descriptions of fundamental concepts—
such as quantum state spaces, quantum measurement, and entanglement—before covering the
xii Preface
standard quantum algorithms and other quantum information processing tasks such as quantum
key distribution and quantum teleportation.
The intent of this book is to make quantum computing accessible to a wide audience of computer
scientists, engineers, mathematicians, and anyone with a general interest in the subject who knows
sufficient mathematics. Basic concepts from college-level linear algebra such as vector spaces,
linear transformations, eigenvalues, and eigenvectors are used throughout the book. Afew sections
require more mathematics; familiarity with group theory is required for sections 8.6.1 and 8.6.2,
appendix B, and much of chapter 11. Group theory is reviewed in boxes, but readers who have
never seen group theory should consult a book on the subject or skip those sections.
While we hope our book lives up to the gentle of its title, reading it will require effort. Many of
the concepts are subtle and unintuitive, and much of the notation unfamiliar. Readers will need to
spend time working with the concepts and notations to develop a level of fluency at each stage.
For example, even readers with significant mathematical background may not have worked much
with tensor products and may not be familiar with the relation of tensor product spaces to their
component spaces. The early chapters of the book develop these notions carefully, since they are
absolutely fundamental to quantum information processing. It is well worth the effort to master
them, as well as the concise Dirac notation in which they are generally expressed, but mastery will
require effort. The precise nature of these mathematical formalisms provides a means of working
with quantum concepts before fully understanding them. Intuition for quantum mechanics and
quantum information processing will develop from playing with the formal mathematics.
The book emphasizes features of quantum mechanics that give quantum computation its power
and are responsible for its limitations. Neither the extent of the power of quantum computation
nor its limitations have been fully understood. Research challenges remain not only in build-
ing quantum computers and developing novel algorithms and protocols, but also in answering
fundamental questions as to the source of quantum computing’s power and the reasons for its
limitations. This book examines what is known about what quantum computers can and cannot
do, and also explores what is known about why.
The focus on the reasons underlying quantum computing’s effectiveness results in the inclusion
of topics frequently left out of other expositions of the subject. For example, one theme of the
book is the relationship of quantum information processing to probability. That many quantum
algorithms are nonprobabilistic is emphasized. A section is devoted to modifications of Grover’s
original algorithm that preserve the speed-up but return a solution with certainty. On the other
hand, the strong formal resemblance between quantum theory and probability theory is described
in detail and distinctions are highlighted, illuminating, for example, how entanglement differs
from correlation, and the difference between a superposition and a mixture.
As another example, while quantum entanglement is the most common explanation given for
why quantum information processing works, multipartite entanglement remains poorly under-
stood. Bipartite entanglement is much better understood but has limited use for understanding
quantum computation. The book includes sections on multipartite entanglement, a topic often left
Preface xiii
Acknowledgments
We are enormously indebted to Michael B. Heaney and Paul McEvoy, both of whom read multiple
versions of many of the chapters and provided valuable comments each time. It is largely due to
their steadfast belief in this project that the book reached completion. The FXPAL/PARC reading
group enabled us to discover which expository approaches worked and which did not. The group’s
comments, struggles, and insights spurred substantial improvements in the book. We are grateful
to all of the members of that group, particularly Dirk Balfanz, Stephen Jackson, and Michael Plass.
Many thanks to Tad Hogg and Marc Rieffel for their feedback on some of the most technical and
notationally heavy sections. Thanks also go to Gene Golovchinsky for suggestions that clarified
and streamlined the writing of an early draft, to Livia Polanyi for suggestions that positively
impacted the flow and emphasis, to Garth Dales for comments on an early draft that improved
our wording and use of notation, and to Denise Greaves for extensive editorial assistance. Many
people provided valuable comments on drafts of the tutorial3 that was the starting point for this
book. Their comments improved this book as well as the tutorial. We gratefully acknowledge
the support of FXPAL for part of this work. We are grateful to our friends, to our family, and
especially to our spouses for their support throughout the years it took us to write this book.
Notes
In the last decades of the twentieth century, scientists sought to combine two of the century’s most
influential and revolutionary theories: information theory and quantum mechanics. Their success
gave rise to a new view of computation and information. This new view, quantum information
theory, changed forever how computation, information, and their connections with physics are
thought about, and it inspired novel applications, including some wildly different algorithms and
protocols. This view and the applications it spawned are the subject of this book.
Information theory, which includes the foundations of both computer science and communica-
tions, abstracted away the physical world so effectively that it became possible to talk about the
major issues within computer science and communications, such as the efficiency of an algorithm
or the robustness of a communication protocol, without understanding details of the physical
devices used for the computation or the communication. This ability to ignore the underlying
physics proved extremely powerful, and its success can be seen in the ubiquity of the computing
and communications devices around us. The abstraction away from the physical had become such
a part of the intellectual landscape that the assumptions behind it were almost forgotten. At its
heart, until recently, information sciences have been firmly rooted in classical mechanics. For
example, the Turing machine is a classical mechanical model that behaves according to purely
classical mechanical principles.
Quantum mechanics has played an ever-increasing role in the development of new and more
efficient computing devices. Quantum mechanics underlies the working of traditional, classical
computers and communication devices, from the transistor through the laser to the latest hardware
advances that increase the speed and power and decrease the size of computer and communications
components. Until recently, the influence of quantum mechanics remained confined to the low-
level implementation realm; it had no effect on how computation or communication was thought
of or studied.
In the early 1980s, a few researchers realized that quantum mechanics had unanticipated impli-
cations for information processing. Charles Bennett and Gilles Brassard, building on ideas of
Stephen Wiesner, showed how nonclassical properties of quantum measurement provided a prov-
ably secure mechanism for establishing a cryptographic key. Richard Feynman, Yuri Manin,
and others recognized that certain quantum phenomena—phenomena associated with so-called
2 1 Introduction
over classical algorithms could be proved, give the correct answer with certainty. They improve
upon classical algorithms by solving in polynomial time with certainty a problem that can be
solved in polynomial time only with high probability using classical techniques. Such a result is
of no direct practical interest, since the impossibility of building a perfect machine reduces any
practical machine running any algorithm to solving a problem only with high probability. But
such results were of high theoretical interest, since they showed for the first time that quantum
computation is theoretically more powerful than classical computation for certain computational
problems.
These results caught the interest of various researchers, including Peter Shor, who in 1994 sur-
prised the world with his polynomial-time quantum algorithm for factoring integers. This result
provided a solution to a well-studied problem of practical interest. A classical polynomial-time
solution had long been sought, to the point where the world felt sufficiently confident that no
such solution existed that many security protocols, including the widely used RSA algorithm,
base their security entirely on the computational difficulty of this problem. It is unknown whether
an efficient classical solution exists, so Shor’s result does not prove that quantum computers can
solve a problem more efficiently than a classical computer. But even in the unlikely event that a
polynomial-time classical algorithm is found for this problem, it would be an indication of the ele-
gance and effectiveness of the quantum information theory point of view that a quantum algorithm,
in spite of all the unintuitive aspects of quantum mechanics, was easier to find.
While Shor’s result sparked a lot of interest in the field, doubts as to its practical significance
remained. Quantum systems are notoriously fragile. Key properties, such as quantum entangle-
ment, are easily disturbed by environmental influences that cause the quantum states to decohere.
Properties of quantum mechanics, such as the impossibility of reliably copying an unknown
quantum state, made it look unlikely that effective error-correction techniques for quantum compu-
tation could ever be found. For these reasons, it seemed unlikely that reliable quantum computers
could be built.
Luckily, in spite of serious and widespread doubts as to whether quantum information process-
ing could ever be practical, the theory itself proved so tantalizing that researchers continued to
explore it. As a result, in 1996 Shor and Robert Calderbank, and independently Andrew Steane,
saw a way to finesse the seemingly show-stopping problems of quantum mechanics to develop
quantum error correction techniques. Today, quantum error correction is arguably the most mature
area of quantum information processing.
How practical quantum computing and quantum information will turn out is still unknown. No
fundamental physical principles are known that prohibit the building of large-scale and reliable
quantum computers. Engineering issues, however, remain. As of this writing, laboratory exper-
iments have demonstrated quantum computations with several quantum bits performing dozens
of quantum operations. Myriad promising approaches are being explored by theorists and exper-
imentalists around the world, but much uncertainty remains as to how, when, or even whether, a
quantum computer capable of carrying out general quantum computations on hundreds of qubits
will be built.
4 1 Introduction
Quantum computational approaches improve upon classical methods for a number of special-
ized tasks. The extent of quantum computing’s applicability is still being determined. It does not
provide efficient solutions to all problems; neither does it provide a universal way of circumvent-
ing the slowing of Moore’s law. Strong limitations on the power of quantum computation are
known; for many problems, it has been proven that quantum computation provides no significant
advantage over classical computation. Grover’s algorithm, the other major algorithm of the mid-
1990s, provides a small speedup for unstructured search algorithms. But it is also known that this
small speedup is the most that quantum algorithms can attain. Grover’s search algorithm applies
to unstructured search. For other search problems, such as searching an ordered list, quantum
computation provides no significant advantage over classical computation. Simulation of quan-
tum systems is the other significant application of quantum computation known in the mid-1990s.
Of interest in its own right, the simulation of increasingly larger quantum systems may provide
a bootstrap that will ultimately lead to the building of a scalable quantum computer.
After Grover’s algorithm, there was a hiatus of more than five years before a significantly new
algorithm was discovered. During that time, other areas of quantum information processing, such
as quantum error correction, advanced significantly. In the early 2000s, several new algorithms
were discovered. Like Shor’s algorithm, these algorithms solve specific problems with narrow,
if important, applications. Novel approaches to constructing quantum algorithms also devel-
oped. Investigations of quantum simulation from a quantum-information-processing point of view
have led to improved classical techniques for simulating quantum systems, as well as novel quan-
tum approaches. Similarly, the quantum-information-processing point of view has led to novel
insights into classical computing, including new classical algorithms. Furthermore, alternatives to
the standard circuit model of quantum computation have been developed that have led to new quan-
tum algorithms, breakthroughs in building quantum computers, new approaches to robustness,
and significant insights into the key elements of quantum computation.
However long it takes to build a scalable quantum computer and whatever the breadth of
applications turns out to be, quantum information processing has changed forever the way in which
quantum physics is understood. The quantum information processing view of quantum mechanics
has done much to clarify the character of key aspects of quantum mechanics such as quantum
measurement and entanglement. This advancement in knowledge has already had applications
outside of quantum information processing to the creation of highly entangled states used for
microlithography at scales below the wavelength limit and for extraordinarily accurate sensors.
The precise practical consequences of this increased understanding of nature are hard to predict,
but the unification of the two theories that had the most profound influence on the technological
advances of the twentieth century can hardly fail to have profound effects on technological and
intellectual developments throughout the twenty-first.
Part I of this book covers the basic building blocks of quantum information processing: quan-
tum bits and quantum gates. Physical motivation for these building blocks is given and tied to the
key quantum concepts of quantum measurement, quantum state transformations, and entangle-
ment between quantum subsystems. Each of these concepts is explored in depth. Quantum key
1 Introduction 5
distribution, quantum teleportation, and quantum dense coding are introduced along the way. The
final chapter of part I shows that anything that can be done on a classical computer can be done
with comparable efficiency on a quantum computer.
Part II covers quantum algorithms. It begins with a description of some of the most common
elements of quantum computation. Since the advantage of quantum computation over classical
computation is all about efficiency, part II carefully defines notions of complexity. Part II also
discusses known bounds on the power of quantum computation. A number of simple algorithms
are described. Full chapters are devoted to Shor’s algorithm and Grover’s algorithm.
Part III explores entanglement and robust quantum computation. A discussion of quantum
subsystems leads into discussions of quantifying entanglement and of decoherence, the environ-
mental errors affecting a quantum system because it is really a part of a larger quantum system.
The elegant and important topic of quantum error correction fills a chapter, followed by a chapter
on techniques to achieve fault tolerance. The book finishes with brief descriptions and pointers
to references for many quantum information processing topics the book could not cover in depth.
These include further quantum algorithms and protocols, adiabatic, cluster state, holonomic, and
topological quantum computing, and the impact quantum information processing has had on
classical computer science and physics.
I QUANTUM BUILDING BLOCKS
Quantum mechanics, that mysterious, confusing discipline, which none of us really understands, but which
we know how to use.
—Murray Gell-Mann [126]
2 Single-Qubit Quantum Systems
Quantum bits are the fundamental units of information in quantum information processing in
much the same way that bits are the fundamental units of information for classical processing.
Just as there are many ways to realize classical bits physically (two voltage levels, lights on or off
in an array, positions of toggle switches), there are many ways to realize quantum bits physically.
As is done in classical computer science, we will concern ourselves only rarely with how the
quantum bits are realized. For the sake of concretely illustrating quantum bits and their properties,
however, section 2.1 looks at the behavior of polarized photons, one of many possible realizations
of quantum bits.
Section 2.2 abstracts key properties from the polarized photon example of section 2.1 to give
a precise definition of a quantum bit, or qubit, and a description of the behavior of quantum bits
under measurement. Dirac’s bra / ket notation, the standard notation used throughout quantum
information processing as well as quantum mechanics, is introduced in this section. Section 2.4
describes the first application of quantum information processing: quantum key distribution. The
chapter concludes with a detailed discussion of the state space of a single-qubit system.
A simple experiment illustrates some of the nonintuitive behavior of quantum systems, behavior
that is exploited to good effect in quantum algorithms and protocols. This experiment can be
performed by the reader using only minimal equipment: a laser pointer and three polaroids
(polarization filters), readily available from any camera supply store. The formalisms of quantum
mechanics that describe this simple experiment lead directly to a description of the quantum bit,
the fundamental unit of quantum information on which quantum information processing is done.
The experiment not only gives a concrete realization of a quantum bit, but it also illustrates key
properties of quantum measurement. We encourage you to obtain the equipment and perform the
experiment yourself.
10 2 Single-Qubit Quantum Systems
Figure 2.1
Single polaroid attenuates unpolarized light by 50 percent.
Figure 2.2
Two orthogonal polaroids block all photons.
2.1 The Quantum Mechanics of Photon Polarization 11
C
B
A
Figure 2.3
Inserting a third polaroid allows photons to pass.
Finally, place polaroid B between polaroids A and C. One might expect that adding another
polaroid will not make any difference; if no light got through two polaroids, then surely no light
will pass through three! Surprisingly, at most polarization angles of B, light shines on the screen.
The intensity of this light will be maximal if the polarization of B is at 45 degrees to both A and
C (figure 2.3).
Clearly the polaroids cannot be acting as simple sieves; otherwise, inserting polaroid B could
not increase the number of photons reaching the screen.
v =a +b
b
Figure 2.4
Measurement of state |v = a|↑ + b|→ by a measuring device with preferred basis {|↑, |→}.
is |v, just as v or v are notations used for vectors in other settings. This notation is part of a
more general notation, Dirac’s notation, that will be explained in more detail in sections 2.2 and
4.1. An arbitrary polarization can be expressed as a linear combination |v = a|↑ + b|→ of the
two basis vectors |↑ and |→. For example, | = √12 |↑ + √12 |→ is a unit vector representing
polarization of 45 degrees. The coefficients a and b in |v = a|↑ + b|→ are called the amplitudes
of |v in the directions |↑ and |→ respectively (see figure 2.4). When a and b are both non-zero,
|v = a|↑ + b|→ is said to be a superposition of |↑ and |→.
Quantum mechanics models the interaction between a photon and a polaroid as follows. The
polaroid has a preferred axis, its polarization. When a photon with polarization |v = a|↑ +
b|→ meets a polaroid with preferred axis |↑, the photon will get through with probability
|a|2 and will be absorbed with probability |b|2 ; the probability that a photon passes through the
polaroid is the square of the magnitude of the amplitude of its polarization in the direction of the
polaroid’s preferred axis. The probability that the photon is absorbed by the polaroid is the square
of the magnitude of the amplitude in the direction perpendicular to the polaroid’s preferred axis.
Furthermore, any photon that passes through the polaroid will now be polarized in the direction of
the polaroid’s preferred axis. The probabilistic nature of the interaction and the resulting change
of state are features of all interactions between qubits and measuring devices, no matter what their
physical realization.
In the experiment, any photons that pass through polaroid A will leave polarized in the direction
of polaroid A’s preferred axis, in this case horizontal, |→. A horizontally polarized photon has
no amplitude in the vertical direction, so it has no chance of passing through polaroid C, which
was given a vertical orientation. For this reason, no light reaches the screen. Had polaroid C
been in any other orientation, a horizontally polarized photon would have some amplitude in the
direction of polaroid C’s preferred axis, and some photons would reach the screen.
To understand what happens once polaroid B, with preferred axis |, is inserted, it is helpful
to write the horizontally polarized photon’s polarization state |→ as
1 1
|→ = √ | − √ |.
2 2
2.2 Single Quantum Bits 13
Any photon that passes through polaroid A becomes horizontally polarized, so the amplitude
of any such photon’s state |→ in the direction | is √12 . Applying the quantum theory we just
learned tells us that a horizontally polarized photon will pass through polaroid B with probability
= | √12 | . Any photons that have passed through polaroid B now have polarization |. When
1 2
2
these photons hit polaroid C, they do have amplitude in the vertical direction, so some of them
(half ) will pass thorough polaroid C and hit the screen (see figure 2.3). In this way, quantum
mechanics explains how more light can reach the screen when the third polaroid is added, and it
provides a means to compute how much light will reach the screen.
In summary, the polarization state of a photon is modeled as a unit vector. Its interaction with a
polaroid is probabilistic and depends on the amplitude of the photon’s polarization in the direction
of the polaroid’s preferred axis. Either the photon will be absorbed or the photon will leave the
polaroid with its polarization aligned with the polaroid’s preferred axis.
The space of possible polarization states of a photon is an example of a quantum bit, or qubit. A
qubit has a continuum of possible values: any state represented by a unit vector a|↑ + b|→ is
a legitimate qubit value. The amplitudes a and b can be complex numbers, even though complex
amplitudes were not needed for the explanation of the experiment. (In the photon polarization
case, the imaginary coefficients correspond to circular polarization.)
In general, the set of all possible states of a physical system is called the state space of the
system. Any quantum mechanical system that can be modeled by a two-dimensional complex
vector space can be viewed as a qubit. (There is redundancy in this representation in that any
vector multiplied by a modulus one [unit length] complex number represents the same quantum
state. We discuss this redundancy carefully in sections 2.5 and 3.1.) Such systems, called two-
state quantum systems, include photon polarization, electron spin, and the ground state together
with an excited state of an atom. The two-state label for these systems does not mean that the
state space has only two states—it has infinitely many—but rather that all possible states can be
represented as a linear combination, or superposition, of just two states. For a two-dimensional
complex vector space to be viewed as a qubit, two linearly independent states, labeled |0 and |1,
must be distinguished. For the theory of quantum information processing, all two-state systems,
whether they be electron spin or energy levels of an atom, are equally good. From a practical
point of view, it is as yet unclear which two-state systems will be most suitable for physical
realizations of quantum information processing devices such as quantum computers; it is likely
that a variety of physical representation of qubits will be used.
Dirac’s bra / ket notation is used throughout quantum physics to represent quantum states and
their transformations. In this section we introduce the part of Dirac’s notation that is used for
quantum states. Section 4.1 introduces Dirac’s notation for quantum transformations. Familiarity
and fluency with this notation will help greatly in understanding all subsequent material; we
strongly encourage readers to work the exercises at the end of this chapter.
14 2 Single-Qubit Quantum Systems
In Dirac’s notation, a ket such as |x, where x is an arbitrary label, refers to a vector representing
a state of a quantum system. A vector |v is a linear combination of vectors |s1 , |s2 , . . . , |sn if
there exist complex numbers ai such that |v = a1 |s1 + a2 |s2 + · · · + an |sn .
A set of vectors S generates a complex vector space V if every element |v of V can be
written as a complex linear combination of vectors in the set: every |v ∈ V can be written as
|v = a1 |s1 + a2 |s2 + · · · + an |sn for some elements |si ∈ S and complex numbers ai . Given a
set of vectors S, the subspace of all linear combinations of vectors in S is called the span of S and
is denoted span(S). A set of vectors B for which every element of V can be written uniquely as a
linear combination of vectors in B is called a basis for V . In a two-dimensional vector space, any
two vectors that are not multiples of each other form a basis. In quantum mechanics, bases are
usually required to be orthonormal, a property we explain shortly. The two distinguished states,
|0 and |1, are also required to be orthonormal.
An inner product v2 |v1 , or dot product, on a complex vector space V is a complex function
defined on pairs of vectors |v1 and |v2 in V , satisfying
• v|v is non-negative real,
• v2 |v1 = v1 |v2 , and
• (a v2 | + b v3 |)|v1 = a v2 |v1 + b v3 |v1 ,
processing, classical bit values of 0 and 1 will be encoded in the distinguished states |0 and |1.
This encoding enables a direct comparison between bits and qubits: bits can take on only two
values, 0 and 1, while qubits can take on not only the values |0 and |1 but also any superposition
of these values, a|0 + b|1, where a and b are complex numbers such that |a|2 + |b|2 = 1.
Vectors and linear transformations can be written using matrix notation once a basis has been
a
specified. That is, if basis {|β1 , |β2 } is specified, a ket |v = a|β1 + b|β2 can be written ;
b
a ket |v corresponds to a column vector v, where v is simply a label, a name for this vector. The
conjugate transpose v † of a vector
⎛ ⎞
a1
⎜ ⎟
v = ⎝ ... ⎠ is v † = ( a1 , . . . , an ) .
an
In Dirac’s notation, the conjugate transpose of a ket |v is called a bra and is written v|, so
⎛ ⎞
a1
⎜ ⎟
|v = ⎝ ... ⎠ and v| = ( a1 , . . . , an ) .
an
When a = |a and b = |b are real vectors, this inner product is the same as the standard dot
product on the n dimensional real vector space Rn : a|b = a1 b1 + · · · + an bn = a · b. Dirac’s
choice of bra and ket arose as a play on words: an inner product a|b of a bra a| and a ket |b
is sometimes called a bracket. The following relations hold, where v = a|0 + b|1: 0|0 = 1,
1|1 = 1, 1|0 = 0|1 = 0, 0|v = a, and 1|v = b.
In the standard basis, with ordering {|0, |1}, the basis elements |0 and |1 can be expressed
1 0 a
as and , and a complex linear combination |v = a|0 + b|1 can be written .
0 1 b
16 2 Single-Qubit Quantum Systems
This choice of basis and order of the basis vectors are mere convention. Representing |0 as
1 0 1 1
and |1 as or representing |0 as √12 and |1 as √12 would be equally
0 1 −1 1
good as long as it is done consistently. Unless otherwise specified, all vectors and matrices in this
book will be written with respect to the standard basis {|0, |1} in this order.
A quantum state |v is a superposition of basis elements {|β1 , |β2 } if it is a nontrivial linear
combination of |β1 and |β2 , if |v = a1 |β1 + a2 |β2 where a1 and a2 are non-zero. For the
term superposition to be meaningful, a basis must be specified. In this book, if we say “super-
postion” without explicitly specifying the basis, we implicitly mean with respect to the standard
basis.
Initially the vector/matrix notation will be easier for many readers to use because it is familiar.
Sometimes matrix notation is convenient for performing calculations, but it always requires the
choice of a basis and an ordering of that basis. The bra / ket notation has the advantage of being
independent of basis and the order of the basis elements. It is also more compact and suggests
correct relationships, as we saw for the inner product, so once it becomes familiar, it is easier to
read and faster to use.
Instead of qubits, physical systems with states modeled by three- or n-dimensional vector
spaces could be used as fundamental units of computation. Three-valued units are called qutrits,
and n-valued units are called qudits. Since qudits can be modeled using multiple qubits, a model
of quantum information based on qudits has the same computational power as one based on qubits.
For this reason we do not consider qudits further, just as in the classical case most people use a
bit-based model of information.
We now have a mathematical model with which to describe quantum bits. In addition, we need
a mathematical model for measuring devices and their interaction with quantum bits.
The interaction of a polaroid with a photon illustrates key properties of any interaction between
a measuring device and a quantum system. The mathematical description of the experiment can
be used to model all measurements of single qubits, whatever their physical instantiation. The
measurement of more complicated systems retains many of the features of single-qubit measure-
ment: the probabilistic outcomes and the effect measurement has on the state of the system. This
section considers only measurements of single-qubit systems. Chapter 4 discusses measurements
of more general quantum systems.
Quantum theory postulates that any device that measures a two-state quantum system must have
two preferred states whose representative vectors, {|u, |u⊥ }, form an orthonormal basis for the
associated vector space. Measurement of a state transforms the state into one of the measuring
device’s associated basis vectors |u or |u⊥ . The probability that the state is measured as basis
vector |u is the square of the magnitude of the amplitude of the component of the state in the
direction of the basis vector |u. For example, given a device for measuring the polarization of
2.3 Single-Qubit Measurement 17
photons with associated basis {|u, |u⊥ }, the state |v = a|u + b|u⊥ is measured as |u with
probability |a|2 and as |u⊥ with probability |b|2 .
This behavior of measurement is an axiom of quantum mechanics. It is not derivable from
other physical principles; rather, it is derived from the empirical observation of experiments with
measuring devices. If quantum mechanics is correct, all devices that measure single qubits must
behave in this way; all have associated bases, and the measurement outcome is always one of the
two basis vectors. For this reason, whenever anyone says “measure a qubit," they must specify
with respect to which basis the measurement takes place. Throughout the book, if we say “measure
a qubit" without further elaboration, we mean that the measurement is with respect to the standard
basis {|0, |1}.
Measurement of a quantum state changes the state. If a state |v = a|u + b|u⊥ is measured
as |u, then the state |v changes to |u. A second measurement with respect to the same basis will
return |u with probability 1. Thus, unless the original state happens to be one of the basis states,
a single measurement will change that state, making it impossible to determine the original state
from any sequence of measurements.
While the mathematics of measuring a qubit in the superposition state a|0 + b|1 with respect to
the standard basis is clear, measurement brings up questions as to the meaning of a superposition.
To begin with, the notion of superposition is basis-dependent; all states are superpositions with
respect to some bases and not with respect to others. For instance, a|0 + b|1 is a superposition
with respect to the basis {|0, |1} but not with respect to {a|0 + b|1, b|0 − a|1}.
Also, because the result of measuring a superposition is probabilistic, some people are tempted
to think of the state |v = a|0 + b|1 as a probabilistic mixture of |0 and |1. It is not. In particular,
it is not true that the state is really either |0 or |1 and that we just do not happen to know which.
Rather, |v is a definite state, which, when measured in certain bases, gives deterministic results,
while in others it gives random results: a photon with polarization | = √12 (|↑ + |→) behaves
deterministically when measured with respect to the Hadamard basis {|, |}, but it gives
random results when measured with respect to the standard basis {|↑, |→}. It is okay to think
of a superposition |v = a|0 + b|1 as in some sense being in both state |0 and state |1 at the
same time, as long as that statement is not taken too literally: states that are combinations of |0
and |1 in similar proportions but with different amplitudes, such as √12 (|0 + |1), √12 (|0 − |1)
and √12 (|0 + i|1), represent distinct states that behave differently in many situations.
Given that qubits can take on any one of infinitely many states, one might hope that a single
qubit could store lots of classical information. However, the properties of quantum measurement
severely restrict the amount of information that can be extracted from a qubit. Information about
a quantum bit can be obtained only by measurement, and any measurement results in one of only
two states, the two basis states associated with the measuring device; thus, a single measurement
yields at most a single classical bit of information. Because measurement changes the state, one
cannot make two measurements on the original state of a qubit. Furthermore, section 5.1.1 shows
that an unknown quantum state cannot be cloned, which means it is not possible to measure a
qubit’s state in two ways, even indirectly by copying the qubit’s state and measuring the copy.
18 2 Single-Qubit Quantum Systems
Thus, even though a quantum bit can be in infinitely many different superposition states, it is
possible to extract only a single classical bit’s worth of information from a single quantum bit.
The quantum theory introduced so far is sufficient to describe a first application of quantum
information processing: a key distribution protocol that relies on quantum effects for its security
and for which there is no classical analog.
Keys—binary strings or numbers chosen randomly from a sufficiently large set—provide the
security for most cryptographic protocols, from encryption to authentication to secret sharing.
For this reason, the establishment of keys between the parties who wish to communicate is of
fundamental importance in cryptography. Two general classes of keys exist: symmetric keys and
public-private key pairs. Both types are used widely, often in conjunction, in a wide variety of
practical settings, from secure e-commerce transactions to private communication over public
networks.
Public-private key pairs consist of a public key, knowable by all, and a corresponding private
key whose secrecy must be carefully guarded by the owner. Symmetric keys consist of a single
key (or a pair of keys easily computable from one another) that are known to all of the legitimate
parties and no one else. In the symmetric key case, multiple parties are responsible for guarding
the security of the key.
Quantum key distribution protocols establish a symmetric key between two parties, who are
generally known in the cryptographic community as Alice and Bob. Quantum key distribution
protocols can be used securely anywhere classical key agreement protocols such as Diffie-Hellman
can be used. They perform the same task; however, the security of quantum key distribution rests
on fundamental properties of quantum mechanics, whereas classical key agreement protocols
rely on the computational intractability of a certain problem. For example, while Diffie-Hellman
remains secure against all known classical attacks, the problem on which it is based, the discrete
logarithm problem, is tractable on a quantum computer. Section 8.6.1 discusses Shor’s quantum
algorithm for the discrete log problem.
The earliest quantum key distribution protocol is known as BB84 after its inventors, Charles
Bennett and Gilles Brassard, and the year of the invention. The aim of the BB84 protocol is
to establish a secret key, a random sequence of bit values 0 and 1, known only to the two
parties, Alice and Bob, who may use this key to support a cryptographic task such as exchanging
secret messages or detecting tampering. The BB84 protocol enables Alice and Bob to be sure
that if they detect no problems while attempting to establish a key, then with high probability
it is secret. The protocol does not guarantee, however, that they will succeed in establishing a
private key.
SupposeAlice and Bob are connected by two public channels: an ordinary bidirectional classical
channel and a unidirectional quantum channel. The quantum channel allows Alice to send a
sequence of single qubits to Bob; in our case we suppose the qubits are encoded in the polarization
states of individual photons. Both channels can be observed by an eavesdropper Eve. This situation
2.4 A Quantum Key Distribution Protocol 19
classical channel
Eve
Figure 2.5
Alice and Bob wish to agree on a common key not known to Eve.
is illustrated in figure 2.5. To begin the process of establishing a private key, Alice uses quantum
or classical means to generate a random sequence of classical bit values. As we will see, a random
subset of this sequence will be the final private key. Alice then randomly encodes each bit of
this sequence in the polarization state of a photon by randomly choosing for each bit one of the
following two agreed-upon bases in which to encode it: the standard basis,
0 → |↑
1 → |→,
or the Hadamard basis,
She sends this sequence of photons to Bob through the quantum channel.
Bob measures the state of each photon he receives by randomly picking either basis. Over
the classical channel, Alice and Bob check that Bob has received a photon for every one Alice
has sent, and only then do Alice and Bob tell each other the bases they used for encoding and
decoding (measuring) each bit. When the choice of bases agree, Bob’s measured bit value agrees
with the bit value that Alice sent. When they chose different bases, the chance that Bob’s bit
matches Alice’s is only 50 percent. Without revealing the bit values themselves, which would
also reveal the values to Eve, there is no way for Alice and Bob to figure out which of these bit
values agree and which do not. So they simply discard all the bits on which their choice of bases
differed. An average of 50 percent of all bits transmitted remain. Then, depending on the level of
20 2 Single-Qubit Quantum Systems
assurance they require, Alice and Bob compare a certain number of bit values to check that no
eavesdropping has occurred. These bits will also be discarded, and only the remaining bits will
be used as their private key.
We describe one sort of attack that Eve can make and how quantum aspects of this protocol
guard against it. On the classical channel, Alice and Bob discuss only the choice of bases and
not the bit values themselves, so Eve cannot gain any information about the key from listening to
the classical channel alone. To gain information, Eve must intercept the photons transmitted by
Alice through the quantum channel. Eve must send photons to Bob before knowing the choice of
bases made by Alice and Bob, because they compare bases only after Bob has confirmed receipt
of the photons. If she sends different photons to Bob, Alice and Bob will detect that something is
wrong when they compare bit values, but if she sends the original photons to Bob without doing
anything, she gains no information.
To gain information, Eve makes a measurement before sending the photons to Bob. Instead of
using a polaroid to measure, she can use a calcite crystal and a photon detector; a beam of light
passing through a calcite crystal is split into two spatially separated beams, one polarized in the
direction of the crystal’s optic axis and the other polarized in the direction perpendicular to the
optic axis. A photon detector placed in one of the beams performs a quantum measurement:
the probability with which a photon ends up in one of the beams can be calculated just as described
in section 2.3.
Since Alice has not yet told Bob her sequence of bases, Eve does not know in which basis
to measure each bit. If she randomly measures the bits, she will measure using the wrong basis
approximately half of the time. (Exercise 2.10 examines the case in which Eve does not even know
which two bases to choose from.) When she uses the wrong basis to measure, the measurement
changes the polarization of the photon before it is resent to Bob. This change in the polarization
means that, even if Bob measures the photon in the same basis as Alice used to encode the bit, he
will get the correct bit value only half the time.
Overall, for each of the qubits Alice and Bob retain, if the qubit was measured by Eve before
she sent it to Bob, there will be a 25 percent chance that Bob measures a different bit value than
the one Alice sent. Thus, this attack on the quantum channel is bound to introduce a high error
rate that Alice and Bob detect by comparing a sufficient number of bits over the classical channel.
If these bits agree, they can confidently use the remaining bits as their private key. So, not only
is it likely that 25 percent of Eve’s version of the key is incorrect, but the fact that someone is
eavesdropping can be detected by Alice and Bob. Thus Alice and Bob run little risk of establishing
a compromised key; either they succeed in creating a private key or they detect that eavesdropping
has taken place.
Eve does not know in which basis to measure the qubits, a property crucial to the security of this
protocol, because Alice and Bob share information about which bases they used only after Bob has
received the photons; if Eve knew in which basis to measure the photons, her measurements would
not change the state, and she could obtain the bit values without Bob and Alice noticing anything
suspicious. A seemingly easy way for Eve to overcome this obstacle is for her to copy the qubit,
keeping a copy for herself while sending the original on to Bob. Then she can measure her copy
2.5 The State Space of a Single-Qubit System 21
later after learning the correct basis from listening in on the classical channel. Such a protocol is
defeated by an important property of quantum information. As we will show in section 5.1.1, the
no-cloning principle of quantum mechanics means that it is impossible to reliably copy quantum
information unless a basis in which it is encoded is known; all quantum copying machines are
basis dependent. Copying with the wrong machine not only does not produce an accurate copy,
but it also changes the original in much the same way measuring in the wrong basis does. So Bob
and Alice would detect attempts to copy with high probability.
The security of this protocol, like other pure key distribution protocols such as Diffie-Hellman,
is vulnerable to a man-in-the-middle attack in which Eve impersonates Bob to Alice and imper-
sonates Alice to Bob. To guard against such an attack, Alice and Bob need to combine it with an
authentication protocol, be it recognizing each other’s voices or a more mathematical authenti-
cation protocol.
More sophisticated versions of this protocol exist that support quantum key distribution through
noisy channels and stronger guarantees about the amount of information Eve can gain. In the noisy
case, Eve is able to gain some information initially, but techniques of quantum error correction
and privacy amplification can reduce the amount of information Eve gains to arbitrarily low levels
as well as compensate for the noise in the channels.
The state space of a classical or quantum physical system is the set of all possible states of the sys-
tem. Depending on which properties of the system are under consideration, a state of the system
consists of any combination of the positions, momenta, polarizations, spins, energy, and so on of
the particles in the system. When we are considering only polarization states of a single photon,
the state space is all possible polarizations. More generally, the state space for a single qubit, no
matter how it is realized, is the set of possible qubit values,
{a|0 + b|1},
where |a | + |b| = 1 and a|0 + b|1 and a |0 + b |1 are considered the same qubit value if
2 2
a|0 + b|1 = c(a |0 + b |1) for some modulus one complex number c.
This quotient space, a space obtained by identifying sets of equivalent vectors with a single point
in the space, is expressed with the compact notation used for quotient spaces:
So the quantum state space for a single-qubit system is in one-to-one correspondence with the
points of the complex projective space CP1 . We will make no further use of CP1 in this book,
but it is used in the quantum information processing literature.
Because the linearity of vector spaces makes them easier to work with than projective spaces
(we know how to add vectors and there is no corresponding way of adding points in projec-
tive spaces), we generally perform all calculations in the vector space corresponding to the
quantum state space. The multiplicity of representations of a single quantum state in this vec-
tor space representation, however, is a common source of confusion for newcomers to the
field.
A physically important quantity is the relative phase of a single-qubit state a|0 + b|1. The
relative phase (in the standard basis) of a superposition a|0 + b|1 is a measure of the angle in
the complex plane between the two complex numbers a and b. More precisely, the relative phase
is the modulus one complex number eiφ satisfying a/b = eiφ |a |/|b|. Two superpositions a|0 +
b|1 and a |0 + b |1 whose amplitudes have the same magnitudes but that differ in a relative
phase represent different states.
The physically meaningful relative phase and the physically meaningless global phase should
not be confused. While multiplication with a unit constant does not change a quantum state vector,
relative phases in a superposition do represent distinct quantum states: even though |v1 ∼ eiφ |v1 ,
the vectors √12 (eiφ |v1 + |v2 ) and √12 (|v1 + |v2 ) do not represent the same state. We must always
be cognizant of the ∼ equivalence when we interpret the results of our computations as quantum
states.
A few single-qubit states will be referred to often enough that we give them special labels:
√
|+ = 1/ 2(|0 + |1) (2.1)
√
|− = 1/ 2(|0 − |1) (2.2)
√
|i = 1/ 2(|0 + i|1) (2.3)
√
|−i = 1/ 2(|0 − i|1). (2.4)
The basis {|+, |−} is referred to as the Hadamard basis. We sometimes use the notation {|,
|} for the Hadamard basis when discussing photon polarization.
Some authors omit normalization factors, allowing vectors of any length to represent a state
where two vectors represent the same state if they differ by any complex factor. We will explicitly
write the normalizations factors, both because then the amplitudes have a more direct relation to
the measurement probabilities and because keeping track of the normalization factor provides a
check that helps avoid errors.
2.5 The State Space of a Single-Qubit System 23
Extended Complex Plane C ∪ {∞} A correspondence between the set of all complex numbers and
single-qubit states is given by
a|0 + b|1 → b/a = α
The preceding mapping is not defined for the state with a = 0 and b = 1. To make this corre-
spondence one-to-one we need to add a single point, which we label ∞, to the complex plane and
define ∞ ↔ |1. For example, we have
|0 → 0
|1 → ∞
|+ → + 1
|− → − 1
|i → i
|−i → − i.
We now describe another useful model, related to but different from the previous one.
Bloch Sphere Starting with the previous representation, we can map each state, represented by the
complex number α = s + it, onto the unit sphere in three real dimensions, the points (x, y, z) ∈ C
satisfying |x|2 + |y|2 + |z|2 = 1, via the standard stereographic projection
1 − |α |
2
2s 2t
(s, t) → , , ,
|α |2 + 1 |α |2 + 1 |α |2 + 1
24 2 Single-Qubit Quantum Systems
–i i
Figure 2.6
Location of certain single-qubit states on the surface of the Bloch sphere.
further requiring that ∞ → (0, 0, −1). Figure 2.6 illustrates the following correspondences:
|0 → (0, 0, 1)
|− → (−1, 0, 0)
|i → (0, 1, 0)
|−i → (0, −1, 0).
We have given three representations of the quantum state space for a single-qubit system.
1. Vectors written in ket notation: a|0 + b|1 with complex coefficients a and b, subject to
|a |2 + |b|2 = 1, where a and b are unique up to a unit complex factor. Because of this factor, the
global phase, this representation is not one-to-one.
2. Extended complex plane: a single complex number α ∈ C or ∞. This representation is one-
to-one.
3. Bloch sphere: points (x, y, z) on the unit sphere. This representation is also one-to-one.
As we will see in section 10.1, the points in the interior of the sphere also have meaning for
quantum information processing. For historical reasons, the entire ball, including the interior, is
called the Bloch sphere, instead of just the states on the surface, which truly form a sphere. For
2.6 References 25
this reason, we refer to the state space of a single-qubit system as the surface of the Bloch sphere
(figure 2.6).
One of the advantages of the Bloch sphere representation is that it is easy to read off all possible
bases from the model; orthogonal states correspond to antipodal points of the Bloch sphere. In
particular, every diameter of the Bloch sphere corresponds to a basis for the single-qubit state
space.
The illustration we gave in figure 2.4 differs from the Bloch sphere representation of single-
qubit quantum states in that the angles are half that of those in the Bloch sphere representation:
in particular, the angle between two states in figure 2.4 has the usual relation to the inner product,
whereas in the Bloch sphere representation the angle is twice that of the angle in the inner product
formula.
2.6 References
The early essays of Feynman and Manin can be found in [119, 120, 121] and [202, 203] respec-
tively. The bra / ket notation was first introduced by Dirac in 1958 [103]. It is found in most
quantum mechanics textbooks and is used in virtually all papers on quantum computing.
26 2 Single-Qubit Quantum Systems
More information about linear algebra, in particular proofs of facts stated here, can be found in
any linear algebra text, including Strang’s Linear Algebra and Its Applications [265] and Hoffman
and Kunze’s Linear Algebra [152], or in a book on mathematics for physicists such as Bamberg
and Sternberg’s A Course in Mathematics for Students of Physics [30].
The BB84 quantum key distribution protocol was developed by Charles Bennett and Gilles
Brassard [42, 43, 45] building on work of Stephen Wiesner [284]. A related protocol was shown
to be unconditionally secure by Lo and Chau [198]. Their proof was later simplified by Shor
and Preskill [255] and extended to BB84. Another proof was given by Mayers [206]. The BB84
protocol was first demonstrated experimentally by Bennett et al. in 1992 over 30 cm of free space
[37]. Since then, several groups have demonstrated this protocol and other quantum key distri-
bution protocols over 100 km of fiber optic cable. Bienfang et al. [51] demonstrated quantum
key distribution over 23 km of free space at night, and Hughes et al. have achieved distances of
10 km through free space in daylight [156]. See the ARDA roadmap [157], the QIPC strategic
report [295], and Gisin et al. [130] for detailed overviews of implementation efforts and the
challenges involved. The companies id Quantique, MagiQ, and SmartQuantum currently sell
quantum cryptographic systems implementing the BB84 protocol. Other quantum key distribu-
tion protocols exist. Exercise 2.11 develops the B92 protocol, and section 3.4 describes Ekert’s
entanglement-based quantum key distribution protocol.
While we explain all quantum mechanics needed for the topics covered in this book, the reader
may be interested in books on quantum mechanics. Countless books on quantum mechanics are
available. Greenstein and Zajonc [140] give a readable high-level exposition of quantum mechan-
ics, including descriptions of many experiments. The third volume of the Feynman Lectures on
Physics [122] is accessible to a large audience. A classical explanation of the polarization exper-
iment is given in the first volume. Shankar’s textbook [247] defines much more of the notation
and mathematics required for performing calculations than do the previously mentioned books,
and it is quite readable as well. Other textbooks, such as Liboff [194], may be more appropriate
for readers with a physics background.
2.7 Exercises
Exercise 2.1. Let the direction |v of polaroid B’s preferred axis be given as a function of θ ,
|v = cos θ|→ + sin θ|↑, and suppose that the polaroids A and C remain horizontally and
vertically polarized as in the experiment of Section 2.1.1. What fraction of photons reach the
screen? Assume that each photon generated by the laser pointer has random polarization.
Exercise 2.2. Which pairs of expressions for quantum states represent the same state? For those
pairs that represent different states, describe a measurement for which the probabilities of the two
outcomes differ for the two states and give these probabilities.
a. |0 and −|0
b. |1 and i|1
2.7 Exercises 27
c. √1
2
(|0 + |1) and √1
2
(−|0 + i|1)
d. √1
2
(|0 + |1) and √1
2
(|0 − |1)
e. √1
2
(|0 − |1) and √1
2
(|1 − |0)
f. √1
2
(|0 + i|1) and √1
2
(i|1 − |0)
g. √1
2
(|+ + |−) and |0
h. √1
2
(|i − |−i) and |1
i. √1
2
(|i + |−i) and √1
2
(|− + |+)
j. √1
2
|0 + eiπ/4 |1 and √1
2
e−iπ/4 |0 + |1
Exercise 2.3. Which states are superpositions with respect to the standard basis, and which are
not? For each state that is a superposition, give a basis with respect to which it is not a superposition.
a. |+
b. √1 (|+ + |−)
2
c. √1 (|+ − |−)
2
√
d. 2
3
|+ − 12 |−)
e. √1 (|i − |−i)
2
f. √1 (|0 − |1)
2
Exercise 2.4. Which of the states in 2.3 are superpositions with respect to the Hadamard basis,
and which are not?
Exercise 2.5. Give the set of all values of θ for which the following pairs of states are equivalent.
a. |1 and √1 |+ + eiθ |−
2
b. √1
2
|i + eiθ |−i and √12 |−i + e−iθ |i
√ √
c. 12 |0 − 23 |1 and eiθ 12 |0 − 23 |1
Exercise 2.6. For each pair consisting of a state and a measurement basis, describe the possible
measurement outcomes and give the probability for each outcome.
√
a. 2
3
|0 − 12 |1, {|0, |1}
28 2 Single-Qubit Quantum Systems
√
b. 2
3
|1 − 12 |0, {|0, |1}
c. |−i, {|0, |1}
d. |0, {|+, |−}
e. √1
2
(|0 − |1), {|i, |−i}
f. |1, {|i, |−i}
√ √
g. |+, { 12 |0 + 2
3
|1, 23 |0 − 12 |1}
Exercise 2.7. For each of the following states, describe all orthonormal bases that include that
state.
a. √1
2
(|0 + i|1)
b. 1+i
2
|0 − 1−i
2
|1
c. √1
2
|0 + eiπ/6 |1
√
d. 1
2
|+ − i 3
2
|−
Exercise 2.8. Alice is confused. She understands that |1 and −|1 represent the same state. But
she does not understand why that does not imply that √1 (|0 + |1) and √1 (|0 − |1) would be
2 2
the same state. Can you help her out?
Exercise 2.9. In the BB84 protocol, how many bits do Alice and Bob need to compare to have a
90 percent chance of detecting Eve’s presence?
Exercise 2.10. Analyze Eve’s success in eavesdropping on the BB84 protocol if she does not
even know which two bases to choose from and so chooses a basis at random at each step.
a. On average, what percentage of bit values of the final key will Eve know for sure after listening
to Alice and Bob’s conversation on the public channel?
b. On average, what percentage of bits in her string are correct?
c. How many bits do Alice and Bob need to compare to have a 90 percent chance of detecting
Eve’s presence?
Exercise 2.11. B92 quantum key distribution protocol. In 1992 Bennett proposed the following
quantum key distribution protocol. Instead of encoding each bit in either the standard basis or the
Hadamard basis as is done in the BB84 protocol, Alice encodes her random string x as follows
0 → |0
1
1 → |+ = √ (|0 + |1)
2
and sends them to Bob. Bob generates a random bit string y. If yi = 0 he measures the i th qubit
in the Hadamard basis {|+, |−}, if yi = 1 he measures in the standard basis {|0, |1}. In this
protocol, instead of telling Alice over the public classical channel which basis he used to measure
2.7 Exercises 29
z z
y y
x
Figure 2.7
Bloch sphere representation of single-qubit quantum states.
each qubit, he tells her the results of his measurements. If his measurement resulted in |+ or |0
Bob sends 0; if his measurement indicates the state is |1 or |−, he sends 1. Alice and Bob discard
all bits from strings x and y for which Bob’s bit value from measurement yielded 0, obtaining
strings x and y . Alice uses x as the secret key and Bob uses y . Then, depending on the security
level they desire, they compare a number of bits to detect tampering. They discard these check
bits from their key.
a. Show that if Bob receives exactly the states Alice sends, then the strings x and y are identical
strings.
b. Why didn’t Alice and Bob decide to keep the bits of x and y for which Bob’s bit value from
measurement was 0?
c. What if an eavesdropper Eve measures each bit in either the standard basis or the Hadamard
basis to obtain a bit string z and forwards the measured qubits to Bob? On average, how many
bits of Alice and Bob’s key does she know for sure after listening in on the public classical? If
Alice and Bob compare s bit values of their strings x and y , how likely are they to detect Eve’s
presence?
Exercise 2.13. Relate the four parametrizations of the state space of a single qubit to each other:
Give formulas for
a. vectors in ket notation
b. elements of the extended complex plane
c. spherical coordinates for the Bloch sphere (see exercise 2.12)
in terms of the x, y, and z coordinates of the Bloch sphere.
Exercise 2.14.
a. Show that antipodal points on the surface of the Block sphere represent orthogonal states.
b. Show that any two orthogonal states correspond to antipodal points.
3 Multiple-Qubit Systems
The first glimpse into why encoding information in quantum states might support more efficient
computation comes when examining systems of more than one qubit. Unlike classical systems,
the state space of a quantum system grows exponentially with the number of particles. Thus, when
we encode computational information in quantum states of a system of n particles, there are vastly
more possible computation states available than when classical states are used to encode the infor-
mation. The extent to which these large state spaces corresponding to small amounts of physical
space can be used to speed up computation will be the subject of much of the rest of this book.
The enormous difference in dimension between classical and quantum state spaces is due to
a difference in the way the spaces combine. Imagine a macroscopic physical system consisting
of several components. The state of this classical system can be completely characterized by
describing the state of each of its component pieces separately. A surprising and unintuitive as-
pect of quantum systems is that often the state of a system cannot be described in terms of the
states of its component pieces. States that cannot be so described are called entangled states.
Entangled states are a critical ingredient of quantum computation.
Entangled states are a uniquely quantum phenomenon; they have no classical counterpart. Most
states in a multiple-qubit system are entangled states; they are what fills the vast quantum state
spaces. The impossibility of efficiently simulating the behavior of entangled states on classical
computers suggested to Feynman, Manin, and others that it might be possible to use these quan-
tum behaviors to compute more efficiently, leading to the development of the field of quantum
computation.
The first few sections of this chapter will be fairly abstract as we develop the mathematical
formalism to discuss multiple-qubit systems. We will try to make this material more concrete
by including many examples. Section 3.1 formally describes the difference between the way
quantum and classical state spaces combine, the difference between the direct sum of two or more
vector spaces and the tensor product of a set of vector spaces. Section 3.1 then explores some
of the implications of this difference, including the exponential increase in the dimension of a
quantum state space with the number of particles. Section 3.2 formally defines entangled states
and begins to describe their uniquely quantum behavior. As a first illustration of the usefulness
of this behavior, section 3.4 discusses a second quantum key distribution scheme.
32 3 Multiple-Qubit Systems
In classical physics, the possible states of a system of n objects, whose individual states can
be described by a vector in a two-dimensional vector space, can be described by vectors in a
vector space of 2n dimensions. Classical state spaces combine through the direct sum. However,
the combined state space of n quantum systems, each with states modeled by two-dimensional
vectors, is much larger. The vector spaces associated with the quantum systems combine through
the tensor product, resulting in a vector space of 2n dimensions. We begin by reviewing the formal
definition of a direct sum as well as of the tensor product in order to compare the two and the
difference in size between the resulting spaces.
The vector spaces V and W embed in V ⊕ W in the obvious canonical way, and the images are
orthogonal under the standard inner product.
Suppose that the state of each of three classical objects O1 , O2 , and O3 is fully described by two
parameters, the position xi and the momentum pi . Then the state of the system can be described
by the direct sum of the states of the individual objects:
⎛ ⎞
x1
⎜ p ⎟
⎜ ⎜
1 ⎟
⎟
x1 x2 x3 ⎜ x2 ⎟
⊕ ⊕ =⎜ ⎟.
p1 p2 p3 ⎜ p2 ⎟
⎜ ⎟
⎝ x3 ⎠
p3
More generally, the state space of n such classical objects has dimension 2n. Thus the size of the
state space grows linearly with the number of objects.
3.1 Quantum State Spaces 33
for some vi ∈ V and wi ∈ W . Due to the relations defining the tensor product, such a represen-
tation is not unique. Furthermore, while all elements of V ⊗ W can be written
α1 (|α1 ⊗ |β1 ) + α2 (|α2 ⊗ |β1 ) + · · · + αnm (|αn ⊗ |βm ),
Example 3.1.1 Let V and W be two-dimensional vector spaces with orthonormal bases A =
{|α1 , |α2 } and B = {|β1 , |β2 } respectively. Let |v = a1 |α1 + a2 |α2 and |w = b1 |β1 +
b2 |β2 be elements of V and W . Then
|v ⊗ |w = a1 b1 |α1 ⊗ |β1 + a1 b2 |α1 ⊗ |β2 + a2 b1 |α2 ⊗ |β1 + a2 b2 |α2 ⊗ |β2 .
If V and W are vector spaces corresponding to a qubit, each with standard basis {|0, |1},
then V ⊗ W has {|0 ⊗ |0, |0 ⊗ |1, |1 ⊗ |0, |1 ⊗ |1} as basis. The tensor product of two
single-qubit states a1 |0 + b1 |1 and a2 |0 + b2 |1 is a1 a2 |0 ⊗ |0 + a1 b2 |0 ⊗ |1 + a2 b1 |1 ⊗
|0 + a2 b2 |1 ⊗ |1.
To write examples in the more familiar matrix notation for vectors, we must choose an ordering
for the basis of the tensor product space. For example, we can choose the dictionary ordering
{|α1 |β1 , |α1 |β2 , |α2 |β1 , |α2 |β2 }.
Example 3.1.2 With the dictionary ordering of the basis for the tensor product space, the tensor
product of the unit vectors with matrix representation |v = √1 (1, −2)†
5
and |w = √1 (−1, 3)†
10
is the unit vector |v ⊗ |w = 1
√
5 2
(−1, 3, 2, −6) .
†
34 3 Multiple-Qubit Systems
If V and W are inner product spaces, then V ⊗ W can be given an inner product by taking the
product of the inner products on V and W ; the inner product of |v1 ⊗ |w1 and |v2 ⊗ |w2 is
given by
The tensor product of two unit vectors is a unit vector, and given orthonormal bases {|αi } for
V and {|βi } for W , the basis {|αi ⊗ |βj } for V ⊗ W is also orthonormal. The tensor product
V ⊗ W has dimension dim(V ) × dim(W ), so the tensor product of n two-dimensional vector
spaces has 2n dimensions.
Most elements |w ∈ V ⊗ W cannot be written as the tensor product of a vector in V and a
vector in W (though they are all linear combinations of such elements). This observation is of
crucial importance to quantum computation. States of V ⊗ W that cannot be written as the tensor
product of a vector in V and a vector in W are called entangled states. As we will see, for most
quantum states of an n-qubit system, in particular for all entangled states, it is not meaningful to
talk about the state of a single qubit of the system.
A tensor product structure also underlies probability theory. While the tensor product structure
there is rarely mentioned, a common source of confusion is a tendency to try to impose a direct
sum structure on what is actually a tensor product structure. Readers may find it useful to read
section A.1, which discusses the tensor product structure inherent in probability theory, which
illustrates the use of tensor product in another, more familiar, context. Readers may also wish to
do exercises A.1 through A.4.
The subscripts are often dropped, since the corresponding qubit is clear from position. The
convention that adjacency of kets means the tensor product enables us to write this basis more
compactly:
3.1 Quantum State Spaces 35
{|0 · · · |0|0,
|0 · · · |0|1,
|0 · · · |1|0,
..
.,
|1 · · · |1|1}.
Since the tensor product space corresponding to an n-qubit system occurs so frequently
throughout quantum information processing, an even more compact and readable notation uses
|bn−1 . . . b0 to represent |bn−1 ⊗ · · · ⊗ |b0 . In this notation the standard basis for an n-qubit
system can be written
{|0 · · · 00, |0 · · · 01, |0 · · · 10, . . . , |1 · · · 11}.
Finally, since decimal notation is more compact than binary notation, we will represent the state
|bn−1 . . . b0 more compactly as |x, where bi are the digits of the binary representation for the
decimal number x. In this notation, the standard basis for an n-qubit system is written
{|0, |1, |2, . . . , |2n − 1}.
Täällä näki Kustaa ensi kerran nuoren tytön, joka teki häneen
voimakkaan vaikutuksen. Se oli miltei vielä lapsi, joka loi häneen
ihmettelevät silmänsä, mutta hän tunsi, että hän iätpäivät väsymättä
olisi voinut niitä katsella. Sitte tuli hänelle muuta ajattelemista, mutta
jälkeenpäinkin hän monasti näki edessään nuo kirkkaat lapsensilmät.
20.
Aarent Pietarinpojan luona oli koolla suuri seura, sillä sinne olivat
kerääntyneet kaikki eri joukot; oli päätetty yhteisvoimin tehdä
rynnäkkö Jotnisteenin luostaria vastaan, joka oli läänin voudin Olavi
Olavinpojan hallussa. Kunhan vaan saataisiin valloitetuksi edes yksi
ainoa paikka, niin sitte jo olisi helppo ryhtyä uusiin yrityksiin.
Hänellä oli tapana sanoa, ettei missään linnassa ole niin komeaa
naisväkeä ja ettei nainen missään osaa syleillä niin lämpimästi.
— Kyllä, johtaja!
— Jos voidaan?
— Epäilettekö?
— En!
— Vai niin, koska minä olen tiellä, niin kyllä menen, mutta ensin
minä Ruotsin tulevalle kuningattarelle pyydän kertoa, että Björkestan
pitäjässä Vestmanlannissa…
— Ole vaiti!
— Rakas Niilo, eihän tuo juttu liikuta minua eikä sinua. Anna
hänen kertoa se loppuun, niin hän menee.
— Ette toki mahtane luulla, että minä uskoisin teitä? sanoi Richissa
ylpeästi ja nousi.
Grym läksi.
— Täydellisesti.
— Yhtaikaako?
— Ja jollei…
— Niin et milloinkaan, vastasi taalainjunkkari.
21.
KUSTAAN AIKALAISET.
Mutta etsiminen oli ollut turhaa. Hän oli kuin taikakeinojen avulla
kokonaan hävinnyt.
— Ceciliako? Cecilia!
— Ei, ei… odotappa… hänen piti hakea sinua ja hän läksi Peder
Grymin kanssa.
Hän oli vasta saanut Ingierd rouvalta runsaasti rahaa, joten siitä
luostarillekin riitti kaunis lahjoitus, ja toivossa että lahjoja tulisi lisää,
joutui Cecilia nunnien hellän hoidon esineeksi, vieläpä hän
muutamilta sai osakseen ystävyyttäkin.
Mies luostariin!
Kun hän muutaman päivän perästä palasi takaisin, sai hän kuulla,
että sisar Cecilian tila niin oli pahentunut, että hänen täytyi olla
vuoteen omana. Erityinen rukous oli kyllä pidetty, mutta se ei sillä
kertaa ollut vaikuttanut mitään.
Hän tarjosi sitte sairaalle lääkkeen ja Cecilia joi. Taas koetti Ingiald
hänen valtasuontaan, mutta ennenkuin hän päästi käden, lausui hän
hiljaa, miltei kuiskaten:
Silloin sai hän kuulla, että sisar Cecilia on parempi ja että hän
varmaankin haluaa häntä tavata, vaikkei hän olekkaan sanonut
mitään.
Olihan Cecilia vieras, joka vaan oleskeli luostarissa, piti siis kaikin
tavoin koettaa parantaa häntä.
— Elääkö isäni?
ebookbell.com