2 Pdfsam Iq PDF

Master de mathmatiques
UFR Mathmatiques
Quantum mechanics
Mathematical foundations
and applications
Lecture notes
Dimitri Petritis
Rennes (draft of 2 October 2013 at 13:45)
2003 2013 D Petritis
Contents
1 Physics, mathematics, and mathematical physics
2 Phase space, observables, measurements, and yes-no experiments
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 Classical physics as a probability theory with a dynamical law . .
2.2.1 Some reminders from probability theory . . . . . . . . . . .
2.2.2 Postulates for classical systems . . . . . . . . . . . . . . . . .
11
2.3 Classical probability does not suffice to describe Nature! . . . . . .
15
2.3.1 Bells inequalities . . . . . . . . . . . . . . . . . . . . . . . . .
17
2.3.2 The Orsay experiment . . . . . . . . . . . . . . . . . . . . . .
17
2.4 Quantum systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22
2.4.1 Postulates of quantum mechanics: the Hilbert space approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23
2.4.2 Interpretation of the basic postulates . . . . . . . . . . . . .
24
2.5 First consequences of quantum formalism . . . . . . . . . . . . . .
29
2.5.1 Irreducibility of quantum randomness and Heisenbergs uncertainty principle . . . . . . . . . . . . . . . . . . . . . . . . 29

2.5.2 Quantum explanation of the Orsay experiment . . . . . . .
i
30
3 Short resum of Hilbert spaces
31
3.1 Scalar products and Hilbert spaces . . . . . . . . . . . . . . . . . . .
31
3.2 Orthogonality and projection; direct and orthogonal sums . . . .
33
3.3 Duality and Frchet-Riesz theorem; adjoint . . . . . . . . . . . . .
36
3.4 Tensor product of Hilbert spaces . . . . . . . . . . . . . . . . . . . .
36
3.5 Orhonormal systems, ortonormal bases; criteria of completeness

of a basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
38
3.6 Diracs bra and ket notation . . . . . . . . . . . . . . . . . . . . . . .
38
3.7 Some more involved quantum mechanical phenomena . . . . . .
39
3.7.1 Composite systems and entanglement . . . . . . . . . . . .
39
3.7.2 Decoherence and quantum to classical transition . . . . . .
39
3.8 Rigged Hilbert spaces and generalised kets . . . . . . . . . . . . . .
39
4 Algebras of operators
41
4.1 Introduction and motivation . . . . . . . . . . . . . . . . . . . . . .
41
4.2 Algebra of operators . . . . . . . . . . . . . . . . . . . . . . . . . . .
42
4.3 Convergence of sequences of operators . . . . . . . . . . . . . . . .
45
4.4 Classes of operators in B(H) . . . . . . . . . . . . . . . . . . . . . .
45
4.4.1 Self-adjoint and positive operators . . . . . . . . . . . . . . .
46
4.4.2 Projections . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
46
4.4.3 Unitary operators . . . . . . . . . . . . . . . . . . . . . . . . .
47
4.4.4 Isometries and partial isometries . . . . . . . . . . . . . . . .
47
4.4.5 Normal operators . . . . . . . . . . . . . . . . . . . . . . . . .
49
4.5 States on algebras, GNS construction, representations . . . . . . .
49
5 Spectral theory in Banach algebras
51
5.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51
5.2 The spectrum of an operator acting on a Banach space . . . . . . .
53
5.3 The spectrum of an element of a Banach algebra . . . . . . . . . .
56
5.4 Relation between diagonalisability and the spectrum . . . . . . . .
58
5.5 Spectral measures and functional calculus . . . . . . . . . . . . . .
60
5.6 Some basic notions on unbounded operators . . . . . . . . . . . .
66
6 Propositional calculus and quantum formalism based on quantum logic 69

6.1 Lattice of propositions . . . . . . . . . . . . . . . . . . . . . . . . . .
70
6.2 Classical, fuzzy, and quantum logics; observables and states on

logics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
74
6.2.1 Logics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
74
6.2.2 Observables associated with a logic . . . . . . . . . . . . . .
75
6.2.3 States on a logic . . . . . . . . . . . . . . . . . . . . . . . . . .
77
6.3 Pure states, superposition principle, convex decomposition . . . .
80
6.4 Simultaneous observability . . . . . . . . . . . . . . . . . . . . . . .
82
6.5 Automorphisms and symmetries . . . . . . . . . . . . . . . . . . . .
84
7 Standard quantum logics
87
7.1 Observables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
87
7.2 States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
88
7.3 Symmetries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
92
8 States, effects, and the corresponding quantum formalism
95
8.1 States and effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
95
8.2 Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
95
8.3 General quantum transformations, complete positivity, Kraus theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
95
9 Quantum formalism based on the informational approach
97
10 Two illustrating examples
99
10.1 The harmonic oscillator . . . . . . . . . . . . . . . . . . . . . . . . .
99
10.1.1 The classical harmonic oscillator . . . . . . . . . . . . . . . .
99
10.1.2 Quantum harmonic oscillator . . . . . . . . . . . . . . . . . 103

10.1.3 Comparison of classical and quantum harmonic oscillators 107
10.2 Schrdingers equation in the general case, rigged Hilbert spaces . 108
10.3 Potential barriers, tunnel effect . . . . . . . . . . . . . . . . . . . . . 108
10.4 The hydrogen atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
11 Quantifying information: classical and quantum
109
11.1 Classical information, entropy, and irreversibility . . . . . . . . . . 109

12 Turing machines, algorithms, computing, and complexity classes
113
12.1 Deterministic Turing machines . . . . . . . . . . . . . . . . . . . . . 113

12.2 Computable functions and decidable predicates . . . . . . . . . . 116
12.3 Complexity classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
12.4 Non-deterministic Turing machines and the NP class . . . . . . . . 117
12.5 Probabilistic Turing machine and the BPP class . . . . . . . . . . . 118
12.6 Boolean circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
12.7 Composite quantum systems, tensor products, and entanglement 121

12.8 Quantum Turing machines . . . . . . . . . . . . . . . . . . . . . . . 121
13 Cryptology
125
13.1 An old idea: the Vernams code . . . . . . . . . . . . . . . . . . . . . 126

13.2 The classical cryptologic scheme RSA . . . . . . . . . . . . . . . . . 127
13.3 Quantum key distribution . . . . . . . . . . . . . . . . . . . . . . . . 129
13.3.1 The BB84 protocol . . . . . . . . . . . . . . . . . . . . . . . . 129
13.3.2 Simple eavesdropping strategies, disturbance and informational gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
13.3.3 Other cryptologic protocols . . . . . . . . . . . . . . . . . . . 133
14 Elements of quantum computing
135
14.1 Classical and quantum gates and circuits . . . . . . . . . . . . . . . 135

14.2 Approximate realisation . . . . . . . . . . . . . . . . . . . . . . . . . 136
14.3 Examples of quantum gates . . . . . . . . . . . . . . . . . . . . . . . 140
14.3.1 The Hadamard gate . . . . . . . . . . . . . . . . . . . . . . . 140
14.3.2 The phase gate . . . . . . . . . . . . . . . . . . . . . . . . . . 140
14.3.3 Controlled-NOT gate . . . . . . . . . . . . . . . . . . . . . . . 140
14.3.4 Controlled-phase gate . . . . . . . . . . . . . . . . . . . . . . 141
14.3.5 The quantum Toffoli gate . . . . . . . . . . . . . . . . . . . . 141
15 The Shors factoring algorithm
143
16 Error correcting codes, classical and quantum
145
References
146
Index
150
1
Physics, mathematics, and
mathematical physics
La mathmatique est une science exprimentale. Contrairement

en effet un contresens qui se rpand de nos jours (. . . ), les objets
mathmatiques prexistent leurs dfinitions ; celles-ci ont t
labores et prcises par des sicles dactivit scientifique et, si
elles se sont imposes, cest en raison de leur adquation aux objets
mathmatiques quelles modlisent.
Michel Demazure: Calcul diffrentiel, Presses de lcole Polytechnique, Palaiseau (1979).
Physics relies ultimately on experiment. Observation of many different experiments of similar type establishes a phenomenology revealing relations between the experimentally measured physical observables. A phenomenology,
even relying on false hypotheses, can still be useful if it predicts correctly quantitative relationships occurring in yet unrealised experiments1 The next step is
inductive: physical models are proposed satisfying the phenomenological relations. Then, new phenomenology is predicted, new experiments designed to
1
For instance the phenomenology prevailing in the Anticythera mechanism had useful predictive power although the underlying hypothesis of a geocentric solar system was false.
Chapter 1
verify it, and new models are proposed. When sufficient data are available, a
physical theory is proposed verifying all the models that have been developed
so far and all the phenomenological relations that have been established. The
theory can deductively predict the outcome for yet unrealised experiments. If
it is technically possible, the experiment is performed. Either the subsequent
phenomenology contradicts the theoretical predictions and the theory must
be rejected or it is in accordance with them and this precise experiment
serves as an additional validity check of the theory.2 Therefore, physical theories have not a definite status: they are accepted as long as no experiment
contradicts them!
It is a philosophical debate how mathematical theories emerge. Some scientists among them the author of these lines share the opinion expressed by
Michel Demazure (see quotation), claiming that Mathematics is as a matter of
fact an experimental science. Accepting, for the time being, this view, hypotheses for particular mathematical branches are the pendants of models. What
differentiates strongly mathematics from physics is that once the axioms are
stated, the proved theorems (phenomenology) need not be experimentally corroborated, they exist per se. The experimental nature of mathematics is hidden
in the mathematicians intuition that served to propose a given set of axioms
instead of another.
Mathematical physics is physics, i.e. its truth relies ultimately on experiment
but it is also mathematics, in the sense that physical theories are stated as a
set of axioms and the resulting physical phenomenology must derive both as
theorems and as experimental truth.
A general physical theory must describe all physical phenomena in the universe, extending from elementary particles to cosmological phenomena. Numerical values of the fundamental physical quantities, i.e. mass, length, and
time span vast ranges: 1031 kg M 1051 kg, 1015 m L 1027 m, 1023 s T
1017 s. Units used in measuring fundamental quantities, i.e. kilogramme (kg),
metre (m), and second (s) respectively, were introduced after the French Revolution so that everyday life quantities are expressed with reasonable numerical values (roughly in the range 103 103 .) The general theory believed to
describe the universe3 is called Quantum Field Theory; it contains two fundamental quantities, the speed of light in the vacuum, c = 2.99792458 108 m/s,
2
See an example of this procedure in le Monde of 20 September 2002 (reproduced on the

web page of the author http://perso.univ-rennes1.fr/dimitri.petritis/).
3
Strictly speaking, there remain unsolved theoretical difficulties in order to succesfully include gravitational phenomena.
/Users/dp/a/ens/mq/iq-intro.tex
2 Stabilised version of 16 September 2013
and the Planks constant = 1.05457 1034 Js. These constants have extraordinarily atypical numerical values. Everyday velocities are negligible compared
to c, everyday actions are overwhelmingly greater than . Therefore, everyday
phenomena can be thought as the c and 0 limits of quantum field
theory; the corresponding theory is called classical mechanics.
It turns out that considering solely the c limit of quantum field theory
gives rise to another physical theory called quantum mechanics; it describes
phenomena for which the action is comparable with . These phenomena are
important when dealing with atoms and molecules.
The other partial limit, 0, is physically important as well; it describes
phenomena involving velocities comparable with c. These phenomena lead to
another physical theory called special relativity.
Although quantum field theory is still mathematically incomplete, the theories obtained by the limiting processes described above, namely quantum mechanics, special relativity, and classical mechanics are mathematically closed,
i.e. they can be formulated in a purely axiomatic fashion and all the experimental observations made so far (within the range of validity of these theories)
are compatible with the derived theorems.
Among those three theories, quantum mechanics has a very particular status. It can be formulated in a totally axiomatic way; all its predictions have been
verified with unprecedented accuracy; not a single experiment has ever put the
theory in difficulty; quantum phenomena play a prominent role in the global
economy (a very conservative estimate is that 35% of the global wealth relies
on exploiting quantum phenomena). In spite of this tremendous success in the
predictive/explanatory power of quantum mechanics and its mathematically
closed form, its axiomatic setting is not very satisfactory; the theory looks as if
a conceptual building block were missing in the description of the theory.
The purpose of this course is twofold. Firstly, the mathematical foundations of quantum mechanics are presented. Algebra, analysis, probability, and
statistics are necessary to describe and interpret this theory. Its predictions are
often totally counter-intuitive. Hence it is interesting to study this theory that
provides a useful application of the mathematical tools, a source of inspiration4 for new developments for the underlying branches of mathematics, and
4
Recall that entire branches of mathematics have been developed on purpose, to give precise mathematical meaning to initially ill-defined mathematical objects introduced by
physicists to formulate and handle quantum theory. To mention but the few most prominent
examples of such mathematical theories: von Neumann algebras, spectral theory of operators,
Chapter 1
a description of unusual physical phenomena. All these phenomena are verified experimentally nowadays. Quantum mechanics intervenes in a decisive
manner in the explanation of vast classes of phenomena in other fundamental
sciences and in technology. Without being exhaustive, here are some examples
of such quantum phenomena:
atomic and molecular physics (e.g. stability of matter, physical properties
of matter), quantum optics (e.g. lasers),
on which rely chemistry (e.g. valence theory) and biology (e.g. photosynthesis, structure of DNA),
solid state physics (e.g. physics of semiconductors, transistors),
tunnel effect (e.g. atomic force microscope) and nanotechnology,
supraconductivity (e.g. used in magnetic levitation for ultra-fast trains)
and superfluidity
...
There is however another major technological breakthrough that is foreseen
with a tremendous socio-economical impact: if the integration of electronic
components continues at the present pace (see figure 1.1), within 1015 years,
only some tenths of silicium atoms will be required to store a single bit of information. Classical (Boolean) logics does not apply any longer to describe atomic
logical gates, quantum (orthocomplemented lattice) logics is needed instead.
Theoretical exploration of this new type of informatics has started and it is
proven [30] that some algorithmically complex problems, like the integer prime
factoring problem for which the best known algorithm [18] requires a time
is superpolynomial in the number of digits can be achieved in polynomial
time using quantum logic. The present time technology does not yet allow the
prime factoring of large integers but it demonstrates that there is no fundamental physical obstruction to its achievement for the rapidly improving computer technology. Should such a breakthrough occur, all our electronic transmissions, protected by classical cryptologic methods could become vulnerable.
The following table gives a very rough estimate of the time needed to factor an
n = 1000 digits numbers, assuming an operation per nanosecond for various
hypotheses of algorithmic complexity.
On the other hand, present day technology allows to securely and unbreakably cipher messages using quantum cryptologic protocols. Thus the second
purpose of this course is to present the applications of quantum mechanics
theory of distributions, non-commutative probabilities.
Figure 1.1: The evolution of the number of transistors on integrated circuits in

the period 19712011. (Source: Wikipedia, transistor count.)
into the rapidly developing field of quantum information, computing, communication, and cryptology.
Chapter 1
O (exp(n))
10417 yr
O (exp(n 1/3 (log n)2/3 ))

0.2 yr
O (n 3 )
1s
Table 1.1: A very rough estimate of the order of magnitude of the time needed
to factor an n-digit number, under the assumption of execution of the algorithm on a hypothetical computer performing an operation per nanosecond, as
a function of the time complexity of the used algorithm. When the cryptologic
protocol RSA has been proposed [24] in 1978, the best factoring algorithm had
a time complexity in O (exp(n)) (for comparison: age of the universe 1.5 1010
yr). The best known algorithm (the general number field sieve algorithm) reported in [19] requires time O (exp(n 1/3 log2 n)) to factor a n-digit number. The
Shors quantum factoring algorithm [30] requires a time of O (n 3 ).
2
Phase space, observables,
measurements, and yes-no
experiments
2.1 Introduction
As is the case in all experimental sciences, information on a physical system
is obtained through observation (also called measurement) of the values
within a prescribed set that can take the physical observables. The acquisition procedure of the information must be described operationally in terms
of
macroscopic instruments, and
prescriptions on their application on the observables of an objectified
physical system.
The biggest the set of observables whose values are known, the finest is
the knowledge about the physical system. Since crude physical observables
(e.g. number of particles, energy, velocity, etc.) can take values in various sets
7
2.1. Introduction
(N, R+ , R3 , etc.), to have a unified treatment for general systems, we reduce any
physical experiment into a series of measurements of a special class of observables, called yes-no experiments. This is very reminiscent of the approximation of any integrable random variable by a sequence of step functions. Therefore, ultimately, we can focus on observables taking values in the set {0, 1}.
To become quantifiable and theoretically exploitable, experimental observations must be performed under very precise conditions, known as the experimental protocol. Firstly, the objectified system must be carefully prepared
in an initial condition known as the state of the system. Mathematically, the
state incorporates all the a priori information we have on the system, it belongs
to some abstract space of states. Secondly, the system enters in contact with
a measuring apparatus, specifically designed to measure the values of a given
observable, returning the experimental data with values in some space (X, X );
this is precisely the measurement process.
The whole physics relies on the postulate of statistical reproducibility of
experiments: if the same measurement is performed a very large number of
times on a system prepared in the same given state, the experimentally observed data for a given observable are scattered around some mean value in
X with some fluctuations around the mean value. However, when the number of repetitions tends to infinity, the empirical distribution of the observed
data tends to some probability distribution on (X, X ). Thus, abstractly, a measurement is a black box transforming states into probability measures on some
space of observations. Mathematically, it will be shown that the measurement
corresponds to a transformation kernel.
Dealing with random variables, the natural question that arises is: what is
the appropriate (abstract) probability space, if any, on which random variables
entering a given problem can be defined? For sequences of classical random
variables, the answer is well known: such an abstract probability space exists,
provided that the sequence verifies the Kolmogorovs compatibility conditions;
in that case, there exists a canonical (minimal) realisation of the abstract probability space on which the whole sequence is defined. Elements of this probability space are called trajectories of the sequence. The physical analogue of
the minimal realisation of the abstract probability space is called phase space.
Elements of the phase space are called (pure) phases. The physical analogue
of a random variable is called observable.
It turns out that physical observables for classical systems are just random
variables so that the phase space for such systems is a genuine probability space,
/Users/dp/a/ens/mq/iq-phase.tex
while for quantum systems, observables are (generally non-commuting) Hermitean operators acting on an abstract Hilbert space that plays the rle of
quantum phase space.
2.2 Classical physics as a probability theory with a

dynamical law
2.2.1 Some reminders from probability theory
Let us start with the mathematical notion of a random variable.
Definition 2.2.1. (Random variable) Let (, F , P) be an abstract probability
space and (X, X ) a measurable space1 . A function X : X that is (F , X )measurable is called (a X-valued) random variable. The induced probability
measure P X on (X, X ) (i.e. P X (A) = P({ : X () A}, A X ) is called the
law (or distribution) of X .
Remark 2.2.2. We cannot refrain from stressing that in the above definition
of a random variable the pertinent property is that of a measurable function
X : X. Now, it is elementary to show that the datum of X is equivalent to
the datum of a deterministic Markovian kernel K X : X [0, 1] such that
K X (, A) = X () (A) = 1 X 1 (A) () = 1 A (X (). The kernel K X acts on its right to
bX = { f : X R, f bounded measurable} by
Z
bX 3 f 7 K X f () :=
K X (, d x) f (x) bF , ,
and on its left to M1 (F ) = {probability measures on F }, by

M1 (F ) 3 7 K X (A) :=
(d )K X (, A) M1 (X ), A X .
Note also that X = K X 1 X and P X = PK X . If X is fixed, on denoting Q A the random variable Q A () = 1 A (X ()), for A X , we remark that Q 2A = Q A (hence Q A
is a projection) and Q A Q B = 0 if A B = ;. Such random variables are called
questions.
1
The space X can be any Polish space (i.e. a metric, complete, and separable space.) We shall
only consider the case X = Rd , for some d , in this course.
2.2. Classical physics as a probability theory with a dynamical law

Example 2.2.3. Let X = {0, 1}, X be the algebra of subsets of X, and P X ({0}) =
P X ({1}) = 1/2 the law of a random variable X (the honest coin tossing). A possible realisation of (, F , P) is ([0, 1], B([0, 1]), ), where denotes the Lebesgue
measure, and a possible realisation of the random variable X is
X () =
0 if [0, 1/2[
1 if [1/2, 1].
Notice however that the above realisation of the probability space involves
the Borel -algebra over an uncountable set, quite complicated an object indeed. A much more economical realisation should be given by = {0, 1}, F =
X , and P({0}) = P({1}) = 1/2. In the latter case the random variable X should
read X () = : on this smaller probability space, the random variable is the
identity function. Such a realisation is minimal.
Exercise 2.2.4. (An elementary but important exercise; please solve it before
reading the sequel!) Generalise the above minimal construction to the case we
consider two random variables X i : X, for i = 1, 2. Are there some plausible
requirements on the joint distributions for such a construction to be possible?
The canonical construction of the minimal probability space carrying an
infinite family of random variables is also possible.
Definition 2.2.5. (Consistency) Let T be an infinite set (countable or uncountable) and for each t T denote by Rt a copy of the real line, indexed by t . Denote by RT = t T Rt and for n 1 by = (t 1 , . . . , t n ) a finite ordered set of distinct indices t i T, i = 1, . . . , n. Denote P a probability measure on (R , B(R ))
where R = Rt1 Rtn . We say that the family (P ), where runs through all
finite ordered subsets of T , is consistent, if
1. P(t1 ,...tn ) (A 1 A n ) = P(t(1) ,...t(n) (A (1) A (n) ), where is an arbitrary permutation of (1, . . . , n) and A i B(Rti ), and
2. P(t1 ,...tn ) (A 1 A n1 R) = P(t1 ,...tn1 ) (A 1 A n1 ).
Definition 2.2.6. Let T be a subset of R. A family of random variables X
(X t )t T is called a stochastic process with time domain T .
If T = N or Z, the process is called a discrete time process or random sequence, if T = [0, 1] or R or R+ , the process is a continuous time process.
The natural question that arises is whether there exists a probability space
(, F , P) carrying the whole process. In other words, if P X denotes the distribution of the process X , what are the conditions it must fulfil so that there
exists a probability space (, F , P) such that P(B ) = P({ : X () B } for all
B B(RT )? The answer is given by the following
Theorem 2.2.7 (Kolmogorovs existence). Suppose that for n 1, the family
P(X t1 ,...,X tn ) , with t 1 < . . . t n and t i T R, for i = 1, . . . , n, is a consistent family of
probability measures. Then, there are
1. a probability space (, F , P) and
2. a stochastic process X = (X t )t T such that P(X t1 ,...,X tn ) (] , x 1 ] ]
, x n] = P({ : X t1 () x 1 , . . . , X tn () x n }.
Proof. See, for instance, in [29, Theorem II.2.1, p. 247].
Remark 2.2.8. The canonical construction of the minimal probability space

is = RT , F = B(RT ) and for every t T , X t () = t . This minimal space is
also called space of trajectories of the random process and the realisation of X
coordinate method.
2.2.2 Postulates for classical systems

A rough definition of the notion of classical phase space is: the (minimal) space
on which all legitimate physical observables can be defined, or, equivalently,
all legitimate questions can receive a definite answer. From this conceptual
view, the classical phase space shares the same indeterminacy as the probability space. The only objects having physical relevance are the physical observables (as is the case for random variables in probability theory.) Therefore, the
same system can be minimally described by two different phase spaces, depending on the set of questions to be answered.
Example 2.2.9. (Die rolling) Let the physical system be a die and the complete
set of questions to be answered the set {Q 1 , . . . ,Q 6 }, where Q i , i = 1, . . . , 6 stands
for the question: When the die lies at equilibrium on the table, does the top
face read i ? An obvious choice for the phase space is = {1, . . . , 6}. The random
variable X corresponding to the physical observable value of the top face is
realised by X () = , and the questions by Q i = 1 {X =i } , for i = 1, . . . , 6.

Example 2.2.10. Consider the same physical system as in example 2.2.9 and
the same set of questions but think of the die as a solid body that can evolve
in the space. To completely describe its state, we need 3 coordinates for its
barycentre, 3 coordinates for the velocity of the barycentre, 3 coordinates for
the angular velocity, and 3 Euler angles for the unit exterior normal at the centre of face 6. Thus, = R9 S2 . Now the realisation X : {1, . . . , 6} is much
more involved (but still possible in principle) and the questions are again represented by Q i = 1 {X =i } , for i = 1, . . . , 6. Yet, obviously, the representation given
in example 2.2.9 is much simpler than the present one.
Exercise 2.2.11. Determine the phase space for a point mass in dimension 1
subject to the force exerted by a spring of elastic constant k.
Solution: Recall that a point mass m in dimension 1 obeys Newtons equation:
m
d 2x
(t ) = F (x(t )),
dt2
subject to the initial conditions x(0) = x 0 and x(0)

= v 0 , where x(t ) denotes
the position of the mass at instant t and F (y) denotes the force exerted by the
spring on the particle when it is at position y. It reads F (y) = k(y y 0 ) where y 0
is the equilibrium elongation of the spring. The kinetic energy, K , of the particle
is a quadratic form in the velocity
=
K (x)
and the potential energy, U , is given by
Z
U (x) =
m 2
x
2
F (y)d y.
x0
In order to conclude, we need the following

= K (x)+U
Theorem 2.2.12. The total energy H (x, x)

(x) is a constant of motion,
i.e. does not depend on t .
Proof.
d
U
+U (x)) = m x x +
(K (x)
(x)x
dt
x
x F (x))
= x(m
= 0.
p
H0
H
Figure 2.1: The phase space for a point mass in dimension one.
Hence the Newtons equation is equivalent to the system of first order differential equations, known as Hamiltons equations:
dp
dt
dq
dt
H
q
H
,
p
=
=
subject to the initial condition

q(0)
q0
=
,
p(0)
p0
p2
q = x, and H = 2m + U (q). Therefore, the phase space for the

where p = m x,
point mass in dimension one is R2 (one dimension for the position, q, and one
for the momentum p.) Moreover, this space is stratified according to constant
energy surfaces that are ellipses for the case of elastic spring, because potential
energy is quadratic in q (see figure 2.1.)
q(t )
If (t ) =
R2 represents the coordinate and momentum of the sysp(t )
tem at time t , the time evolution induced by the system of Hamiltons equations
2
can be thought
as
the flow on R , described by (t ) = T t (0), with initial conq0
dition (0) =
.
p0
Postulate 2.2.13. The phase space of a classical system is an abstract measurable space (, F ). The states of a classical system are the probability measures
on (, F ). Pure states correspond to Dirac masses. When two systems, respectively described by (1 , F1 ) and (2 , F2 ) are merged and considered as a single

system, their phase space is (1 2 , F1 F2 ), where F1 F2 is the sigma algebra
generated by F1 F2 .
Postulate 2.2.14. Any time evolution of an isolated classical system is implemented by an invertible measurable transformation T : leaving the states
invariant.
Postulate 2.2.15. To any physical observable of a classical system corresponds
a random variable X : X, where (X, X ) is a measurable space. Yes-no questions are special observables of the form Q : {0, 1}. Measurement of a classical observable X when the system is in state , corresponds in determining its
law under .
Remark 2.2.16. Questions are special kinds of random variables. They always
can be written as Q = 1 A X , where X : X R is a random variable, and
A X . Any random variable X is termed physical observable. Questions are
special types of physical observables. Since any question is an indicator, it verifies Q 2 = Q i.e. it is a projector. When dealing with a single random variable
(physical observable), the complete set of possible questions is in bijection with
the -algebra X of measurable subsets of X. If Q A = 1 A X and Q B = 1 B X are
two different questions and moreover A B = ; then Q A Q B = 0, i.e. questions
testing disjoint sets in the range of a random variable are orthogonal projectors.
Exercise 2.2.17. Let be a state on (, F ), X a X-valued random variable (X
R), and Q A the question 1 {X A} for some fixed A X . Compute (Q A ). What
happens if is a pure state? What happens if (, F ) is minimal for the random
variable X ?
Solution:
(Q A ) =
Z
Z
1 {X A} ()(d )
1 A (X ())(d )
If = 0 for some 0 .
0 (Q A ) =
1 {X A} ()0 (d )
= 1 A (X (0 )).
RIf the space is minimal for X , then X () = and we get respectively: (Q A ) =
1 A (X ())(d ) = (A) and 0 (Q A ) = 0 (A).
Exercise 2.2.18. What is the minimal phase space for a mechanical system
composed by N point particles in dimension 3?
2.3 Classical probability does not suffice to describe

Nature!
Quantum mechanics has been proposed as a theory explaining physical phenomena occurring mainly in microscopic systems. Nowadays quantum mechanics provides us with a formalism that has been successfully tested in all
known experimental situations; not a single prediction made by quantum theory has ever been falsified by an experiment! Nevertheless, the quantum formalism remains highly counter-intuitive and several physicists have advanced
the hypothesis that the theory is incomplete. The most prominent among those
physicists was Einstein2 who refused to admit3 the intrinsically stochastic nature of quantum mechanics in an influential paper [13] written with Podolsky4
and Rosen5 in 1935, based on a Gedankenexperiment 6 and known as the EPR
paradox7 .
When dealing with physical theories, we are confronted with two basic notions, locality and realism. Locality means that we can always take actions that
have consequences only within a small region of space. In physics, locality
stems from the finiteness of the speed of light. Since no interaction can propagate faster than light, no influence can be sensed in space points lying beyond
the wave front of light. Realism means that although experiments have always
random outputs, the observed randomness is nothing else than the reflection
of the imperfection of the measuring instruments. In a theory where realism
applies, there is no conceptual obstruction to think that there exists a state in
which the system can be perfectly described in principle. The observed randomness is only a reflection of our incomplete knowledge of the precise state
2
Albert Einstein, (Ulm) 1879 (Princeton) 1955. Developed the theory of special (1905) and
general (1913) relativity, pillars of modern theoretical physics but has been awarded the Nobel
Prize of Physics in 1921 not for this achievement but for his explanation of the photoelectric
effect. Published more than 300 scientific papers of extreme originality. Beyond his scientific contributions Einstein was a humanist worried about war; he has signed with the British
philosopher and mathematician Bertrand Russell the Russell-Einstein manifesto against nuclear weapons.
3
Cf. for instance his famous aphorism God does not play dice with the world (quoted, for
instance, from the conversations with Hermanns, in William Hermanns, Einstein and the Poet:
in search of the cosmic man, Branden Press, Brookline, MA (1983).
4
Boris Yakovlevich Podolsky, (naturalised) American physicist 1896 1966.
5
Nathan Rosen, American physicsist1909 1995.
6
Literally: thought experiment. A very powerful epistemological method developed by
Einstein of questioning the validity of the theoretical predictions.
7
See the Analyse item at http://bibnum.education.fr/physique/physique-quantique/leparadoxe-epr for the unconventional beginnings and fate of the EPR paper.
2.3. Classical probability does not suffice to describe Nature!

of the system. The role of the EPR paradox was to show the incompleteness of
the quantum theory.
Based on Einsteins refutations, Bohm8 proposed, in [9, 10], a formalism of
quantum mechanics postulating the existence of hidden variables (i.e. non
observed ones), allowing to describe the same phenomenology as quantum
mechanics without postulating an intrinsically stochastic nature of the theory.
Therefore, the hidden variables formalism intended to restore the realism of
quantum mechanics. The introduction of hidden variables did not predict any
new phenomenon beyond those predicted by standard quantum theory and
for many years, it was the root of a mainly philosophical controversy between
the tenants of standard quantum theory (the only one we shall present in the
sequel of this course) and the tenants of the hidden variables description.
A major conceptual step was performed by Bell9 who established in [7] that
if hidden variables existed then they should have predictable consequences
that could be experimentally tested. In spite of the fundamental interest such
an experiment would have in the conceptual foundations of the theory, it was
dismissed for several years as an uninteresting philosophical quest not worth
the efforts of respectable scientists10 . It was only thanks to the ingeniousness of
several groups of experimental physicists around the world (Clauser, Shimony,
Horne, Holt, Aspect, Dalibard, Roger who persevered in willing to experimentally test the hypothesis of hidden variables) that the existence of both local and
realistic physical theories has been refuted in the three seminal papers [4, 5, 3]
of the group at the Universit dOrsay. This experiment, of the utmost fundamental importance, refutes all physical theories that are both realistic and local. It established that quantum theory is local but does not satisfies realism;
it describes phenomena that when interpreted within classical probability
appear as non-local. This fact is sometimes wrongly termed quantum nonlocality in the literature. This term will never used in the sequel of this course.
Therefore, before presenting the nowadays accepted form of quantum mechanics, we spend some lines to describe Bell inequalities and explain in some
details the Orsay experiment. We follow the exposition of [21].
8
David Joseph Bohm, 1997 1992. American physicist who introduced the hidden variables
formalism in the quest of restoring realism of quantum mechanics.
9
John Stewart Bell, 1928 1990. A Northern-Irish physicist with a major contribution known
as Bells theorem, establishing the experimental consequences that should have hidden variables in quantum mechanics.
10
Readers so inclined to philosophical meditation are invited to consider the ravages
fashion-led or project-oriented research can cause to the advancement of science.
2.3.1 Bells inequalities

Proposition 2.3.1 (The three-variable Bells inequality). Let X 1 , X 2 , X 3 be an
arbitrary triple of {0, 1}-valued random variables defined on some probability
space (, F , P). Then
P(X 1 = 1, X 3 = 0) P(X 1 = 1, X 2 = 0) + P(X 2 = 1, X 3 = 0).
Proof:
P(X 1 = 1, X 3 = 0) = P(X 1 = 1, X 2 = 0, X 3 = 0) + P(X 1 = 1, X 2 = 1, X 3 = 0)
P(X 1 = 1, X 2 = 0) + P(X 2 = 1, X 3 = 0).
Proposition 2.3.2 (The four-variable Bells inequality). Let X 1 , X 2 , Y1 , Y2 be an

arbitrary quadruple of {0, 1}-valued random variables defined on some probability space (, F , P). Then
P(X 1 = Y1 ) P(X 1 = Y2 ) + P(X 2 = Y2 ) + P(X 2 = Y1 ).
Proof: The random variables being {0, 1}-valued, it is enough to check on all 16
possible realisations of the quadruple (X 1 (), X 2 (), Y1 (), Y2 ()) that
{X 1 = Y1 } {[X 1 = Y2 ] [X 2 = Y2 ] [X 2 = Y1 ]}.
2.3.2 The Orsay experiment

The idea behind the Orsay experiment is to associate precise physical quantities with {0, 1}-valued quantities. Classical theory of light assigns electromagnetic waves; the electric field oscillates in a plane perpendicular to the propagation direction known as polarisation. When monochromatic light emitted
from a random source (i.e. unpolarised) of some intensity I passes through a
polariser, the emerging beam is polarised in the direction of the polariser and
has intensity I /2. Now, it has been established that light beam is composed of a
Figure 2.2: When a photon passes through the first polariser oriented in
direction emerges polarised in that direction. When it encounters a second polariser oriented in direction passes through with probability
cos2 ( ). If the photon is initially already polarised in direction , nothing
changes if the first polariser is removed.
great number of elementary light quanta called photons; the corpuscular nature of light has been conjectured already by Gassendi11 and Newton12 and irrefutably confirmed by the photoelectric effect whose theoretical explanation
was given by Einstein. Therefore, the statement on intensities made above have
only a statistical meaning; if a photon passing through a polariser oriented in a
given direction , encounters a second polariser oriented in a direction , has
probability 21 cos2 ( ) to pass through (see figure 2.2). This is an experimental fact, in accordance with both quantum mechanical prescriptions and with
classical electromagnetic theory of light.
If the experiment is to be explained in terms of classical probability, with every polariser in direction [0, /2] is associated a random variable X {0, 1};
the random variables X are defined on a probability spaces (, F , P) where
represent the microscopic state of the photon. Now for the experimental
setting depicted in figure 2.2, the random variables X are correlated as
E(X X ) = P(X = 1, X = 1) =
1
cos2 ( ).
2
But now there is a problem because this correlation cannot be that of classical
random variables. Choosing in fact three polarisations 1 , 2 , and 3 , we have
P(X i = 1, X j = 0) = P(X i = 1) P(X i = 1, X j = 1) = 21 (1 cos2 (i j )) =
11
Pierre Gasendi, French philosopher, priest, scientist, and mathematician 1592 1655.
Sir Isaac Newton, English physicist and mathematician, 1642 1727. In his very influential
work, its Philosophi naturalis principia mathematica established the classical theory of universal gravitation and discovered what later became differential calculus to solve the equations
of motion.
12
Ca
PM1
PM2
Coincidence monitoring
Figure 2.3: Schematic view of the Orsay experiment [5]. A beam of calcium (Ca)
atoms is triggered by a laser. When thus excited, calcium atom emit simultaneously two photons (at different frequencies) in opposite directions and having
correlated polarisations. An ingenious system of optical switches is used whose
net effect can be described by the following equivalent description. The left polariser is oriented in one of the angles 1 or 2 , the right polariser into one of 1
or 2 . After passing through the polarisers, the photons are counted by photomultipliers (PM1) and (PM2) and only photons detected in synchronisation are
recorded. A careful design is made so that all photons travel on same optical
lengths and the choice of left and right polarisation is made after the photons
are emitted (so that any causal influence of the choice of orientations on the
manner the photons are emitted can be excluded).
1
2
2 sin (i
j ) for i , j {1, 2, 3}. The three-variable Bell inequality 2.3.1 reads
then
1
1
1
sin2 (1 3 ) sin2 (1 2 ) + sin2 (2 3 ).
2
2
2
The choice 1 = 0, 2 = /6, and 3 = /3 leads to the impossible inequality
3/8 1/8 + 1/8. Therefore, classical probability cannot describe this simple experiment.
On a second reading, this experiment is not very convincing because on
arranging polarisers on the optical table as described above, there is nothing
preventing conceptually the second random variable X to depend in fact on
both and . But then the correlation reads E(X X , ) = P(X = 1, X , = 1) =
1
cos2 ( ) and this can be satisfied by choosing, for instance, X and X ,
2
independent with P(X = 1) = 1/2 and P(X , = 1) = cos2 () which of course
can be easily conceived.
The irrefutable evidence of the impossibility of describing Nature with merely
classical probability is provided through the second experiment Aspect, Dal/Users/dp/a/ens/mq/iq-phase.tex

ibard and Roger performed in 1982, [3], schematically described in figure 2.3.
The same analysis can be made as in the previous experimental setting. Denoting by X the {0, 1}-valued random variable quantifying the passage of the
photon through the left polariser and Y through the right one, it is experimentally established in [3] that,
P(X = Y ) =
1
sin2 ( ),
2
for every choice of and . (Note incidentally that the same conclusion is obtained using the not yet presented quantum formalism). Now the choice
1 = 0, 2 = /3, 1 = /2, and 2 = /6, should read 1 1/4 + 1/4 + 1/4, manifestly violating the four variable Bells inequality 2.3.2.
To better grasp the significance of this experiment, it has been proposed,
see [21] for instance, to think of it as a card game between two players X and Y
who can pre-agree on any conceivable strategy in order to win the game. The
game is described in the following
Exercise 2.3.3. (The Orsay experiment as a card game [21]) The game is played
between players X , Y (see figure 2.3), and A who acts as an arbiter and as game
leader (A like . . . Aspect).
Description of the game
A disposes of a well shuffled deck of red and black cards (consider it as an
infinite sequence of i.i.d. {red, black}-valued random variables uniformly
distributed on {red, black} := {r, b}).
X and Y are free to use random resources (e.g. dice) if they wish.
Before the game starts, X and Y agree on given strategy (deterministic,
non-deterministic, or random) how to determine a {yes, no}-valued variable out of the colour of the card they will be presented. Once the game
starts, the players are not allowed any longer to communicate.
A picks two cards from the deck and presents the one to X and the other
to Y (mind that X and Y dont know each others card).
X and Y apply their own pre-agreed strategy to the colour the are presented with and simultaneously say yes or no.
After the announcement of the players, the cards are laid on the table.
Four different card pairs are possible (r r ), (r b), (br ), (bb), where the first
colour refers to the colour of the X s card and the second to the Y s one
(consider these colour pairs as boxes in a 22 board). If both players have
given the same answer then 1 is written in the corresponding box, else 0
is marked.
In the course of the game, the boxes get filled by sequences of 0s and 1s.
Let cc 0 , with c, c 0 {r, b}, be the limit of the empirical probability of 1s
in the box corresponding to colours (cc 0 ) when the game runs indefinitely. The players win the game if r r > r b + br + bb . The purpose
is to show that there exists no strategy (deterministic, non-deterministic,
or random) allowing the players to win the game.
Questions
1. Suppose that X and Y have agreed on the following strategy: X always
says yes, independently of the colour of the card presented to her/him
and Y answers the question is my card red?. Compute explicitly the
values of cc 0 for c, c 0 {r, b} and show that with this deterministic strategy, the numbers cc 0 satisfy the four variable Bells inequality r r r b +
br + bb .
2. In the above strategy, the decision making process is described through
the matrices D X and D Y with D X : {r, b} {0, 1} [0, 1] (and similarly for
D Y ), defined respectively by
0 1
0 1
DX =
and D Y =
,
0 1
1 0
interpreted as meaning P(Answer of X is a | card colour is c) = D X (c, a)
for c {r, b} and a {0, 1} (and similarly for Y ). Hence the previously
described strategies are termed deterministic strategies. Determine all
deterministic strategies and show that for all of them, Bells inequality is
verified.
3. Propose a plausible parametrisation of the space of all strategies (deterministic and random) and show that this space is convex. Is it a simplex?
Show that the previous deterministic strategies are extremal points of this
space.
4. Conclude that in general any strategy can be written as a convex combination of deterministic strategies. Is this decomposition unique? How
such a convex combination is related to hidden variables? Conclude that
no classical strategy exists allowing to win this game.
(Some hints concerning convexity and convex decomposition of stochastic matrices needed for the the two last questions can be found in the chapter Markov
chains on finite state spaces of the lecture notes [22]).
2.4. Quantum systems

This exercise shows that any system described in terms of classical probabilities, even augmented to incorporate hidden variables, cannot win this game.
But the Orsay experiment proved that Nature wins! Therefore, the quantum
strategy used by Nature is strictly more powerful than any classical strategy.
Probably the first person to really understand this power was Feynman13 (see
[15, chap. 7 and 15] for instance), who first conjectured the computational power
of quantum mechanics and was the first who proposed to use atoms as quantum computers.
2.4 Quantum systems

Quantum mechanics emerged thanks to various experimental and theoretical
advances, due mainly to Bohr14 , Planck15 , Schrdinger16 , Heisenberg17 , Dirac18 ,
and many others. Various formulations of quantum mechanics are possible.
We start from the most straightforward one, historically introduced by von Neumann19 [35], based on the Hilbert space formalism. Later on, a more general
13
Richard Phillips Feynman, 1918 1988. American physicist with major contributions in
quantum mechanics, especially known for his work in the path integral formulation of quantum mechanics. Has been awarded the Nobel prize in Physics in 1965. Feynman was an unparalleled populariser of physics; he left co-authored with Leighton and Sands as legacy
to generations of young physicists the seminal 3-volume textbook known as The Feynman
lecture on Physics.
14
Niels Henrik David Bohr, 1865 1962,. Danish physicist with foundational contributions in
quantum mechanics.
15
Max Karl Ernst Ludwig Planck, 1858 1947. German physicist; has been awarded the Nobel
Prize in Physics in 1918. One of the originators of quantum theory.
16
Erwin Rudolf Josef Alexander Schrdinger, Austrian physicist. Developed quantum theory.
17
Werner Karl Heisenberg, 1901 1976. German physicist with major contributions in quantum mechanics. The uncertainty relations are named after him.
18
Paul Adrien Maurice Dirac, 1902 1984. English physicist with major contributions in
quantum electrodynamics. Predicted the existence of antimatter. Has been awarded the Nobel Prize in Physics in 1933. For the needs of quantum mechanics he introduced the notion
of -function and its derivatives, known nowadays as Dirac measure and Dirac distributions
respectively after they have been given a rigorous mathematical status by Laurent Schwartz.
Another happy achievement of Dirac was the so-called Diracs notation in terms of bras and
kets (it will presented in 3.6).
19
Jnos / Johann / John von Neumann, 1903 1954, mathematician, physicist, and computer
scientist (before this last term has been coined by lack of . . . any computer), born as Margittai
Neumann Jnos Lajos in Hungary, achieved his PhD under the direction of Fejr Lipot at the
Institute of theoretical physics of the university of Budapest. Moved to Italy then Switzerland
where he completed studies as chemical engineer at the Eidegenssische technische Hohcschule
Zrich. Exerced as Privatdozent in Gttingen, Berlin and Hamburg but finaly emigrated in 1930
formulation [32], based on quantum logics and C -algebras, will be given; this
formulation has the advantage of allowing a unified treatment for both classical
and quantum systems. Another possible formulation is provided by the informational formulation of quantum mechanics [11, 12]. The reader may wonder why so many formulations have been proposed so far. The answer is that
although all the formulations are totally satisfactory from the computational
point of view, their predictive power, and their adaptedness to explain diverse
experiments, none of the existing ones is philosophically and epistemologically
satisfactory. Quantum mechanics is a partial theory, describing fragile systems,
i.e. systems that eventually leave the quantum realm to enter the classical one;
such a fragility demands for a unified treatment of classical and quantum formalism, not guaranteed by any of the existing formalisms.
2.4.1 Postulates of quantum mechanics: the Hilbert space approach

For the time being, we proclaim that a quantum system verifies the following
postulates.
Postulate 2.4.1. The phase space of a quantum mechanical system is a complex
Hilbert space H. Unit vectors20 of H correspond to pure quantum states. When
two systems, respectively described by Hilbert spaces H1 and H2 , are considered
as a single system, the composite system is described by the Hilbert space H1 H2
where denotes the tensor product.
Postulate 2.4.2. Any time evolution of an isolated quantum system is described
by a unitary operator acting on H. Conversely, any unitary operator acting on H
corresponds to a possible time evolution of the system.
Postulate 2.4.3. With every physical observable, O X , of a quantum system is
associated a self-adjoint operator X acting on the phase space H of the system.
Yes-no questions are special self-adjoint operators that are projections. Measurement of an observable represented by the self-adjoint operator X for a quantum
to the United States (Institute of Adanced Studies - Princeton) to escape from the prosecutions
against Jews that had started in Germany. (Source the biography of von Neumann: [6]).
Among his major achievements are the mathematical foundations of quantum mechanics, a
class of operator algebras (known nowadays as von Neumann algebras), the construction of the
first computer in the world, the introduction of the theory of games, . . . . (See the impressive
sum of his collected works [36, 37, 37, 38, 39, 40, 41]).
20
Strictly speaking, equivalence classes of unit vectors differing by a global phase, called rays.
For the sake of simplicity we stick to unit vectors in this introductory section.

system being in the pure state described by the unit vector corresponds to the
spectral measure on the real line induced by | X .
These axioms will be revisited later. For the time being, it is instructing to
illustrate the implications of these axioms on a very simple non-trivial quantum
system and try to interpret their significance.
2.4.2 Interpretation of the basic postulates

In this subsection we study a quantum system whose phase space H = C2 . This
is the simplest non-trivial situation that might occur and could describe, for
instance, the internal degrees of freedom of an atom having two states. Notice
however that in general, even for very simple finite systems, the phase space is
not necessarily finite-dimensional.
Interpretation of postulate 2.4.1

Every
f H can

be decomposed into f = f 1 1 + f 2 2 with f 1 , f 2 C and 1 =
1
0
f
and 2 =
. If k f k 6= 0, denote by = k f k the corresponding normalised
0
1
vector21 .
Now = 1 1 + 2 2 with |1 |2 + |2 |2 = 1 is a pure state. The numbers |1 |2
and |2 |2 are non-negative reals summing up to 1; therefore, they are interpreted as a probability on the finite set of coordinates {1, 2}. Consequently, the
complex numbers 1 = 1 | and 2 = 2 | are complex probability amplitudes, their squared modulus represents the probability that a system in a pure
state is in the pure state 1 or 2 .
Some precisions are needed to explain the tensor product construction appearing in composite systems.
Definition 2.4.4. Let V,W be two vector spaces. Their (algebraic) tensor product is a vector space, denoted V W with a bilinear tensor mapping defined
by
V W 3 (x, y) 7 (x, y) := x y V W
21
General (unnormalised) vectors of H are denoted by small Latin letters f , g , h, etc.; normalised vectors, representing rays, by small Greek letters , , , etc.
satisfying the following universality property: let Z be an arbitrary vector space

and : V W Z an arbitrary bilinear map. Then there exists a unique linear
map T : V W Z such that = T .
Remark 2.4.5. The meaning of this definition for finite dimensional spaces
is the following: if (e 1 , . . . , e dimV ) and ( f 1 , . . . , f dimW ) are bases of V and W , then
P
P
(e i f j )1i dimV ;1 j dimW is a basis of V W , i.e. if v = i v i e i and w = j w j f j ,
P
then (v, w) = v w = i , j v i w j e i f j . On the other hand, bilinearity of reads
P
(v, w) = i , j v i w j (e i , f j ) Z . The unique linear map T is then defined by
its action on basis vectors by T (e i f j ) = (e i , f j ). Then obviously, (v, w) =
T ((x, y)) by construction.
In case of infinite-dimensional vector spaces some care must be paid on
defining the decomposition of vectors on the respective bases. When these
spaces are (infinite-dimensional separable) Hilbert spaces, their tensor product will be defined in 3.4.
Corollary 2.4.6. dim(V W ) = dimV dimW .
Definition 2.4.7. Let H1 Hn , for n 2. The pure state is called
entagled if it cannot be written as a tensor product = 1 n , with i
Hi , for all i = 1, . . . , n.
Example 2.4.8. Let n = 2 and H1 = H2 = H = C2 . If we denote by (e 0 , e 1 ) a basis
of C2 , a basis of H2 is given by (e 0 e 0 , e 0 e 1 , e 1 e 0 , e 1 e 1 ). An arbitrary vector
H2 is decomposed as
= 0 e 0 e 0 + 1 e 0 e 1 + 2 e 1 e 0 + 3 e 1 e 1 .
If 2 = 3 = 0 while 1 1 6= 0, then = 0 e 0 e 0 + 1 e 0 e 1 = e 0 (0 e 0 +
1 e 1 ) and the state can still be written as a tensor product. If 1 = 2 = 0 while
0 3 6= 0 then the state cannot be written as a tensor product. Such states are
called entangled22 .
The entanglement is a distinctive property of quantum mechanics without
classical analogue. It is exploited from all the informational applications of
quantum mechanics (quantum computing, quantum cryptography, quantum
communication, etc.)
22
The notion was introduced by Schrdinger himself who named this property Verschrnkung in German and translated into English (by Schrdinger himself) as entanglement.
The term is commonly translated intrication in French, although the author of these lines
prefers the term enchevtrement.

A unitary operator on H is a 2 2 matrix U , verifying UU = U U = I . If is a
pure state, then = U verifies kk2 = U |U = |U U = kk2 . Therefore quantum evolution preserves pure states. Moreover, due to the unitarity of
U , we have = U , and since U is again unitary, it corresponds to a possible
time evolution (as a matter of fact to the time reversed evolution of the one corresponding to U .) This shows that time evolution of isolated quantum systems
is reversible.

This axiom has the most counter-intuitive consequences.
Recall that any linear
R
operator X admits a spectral decomposition X = spec(X ) P (d ): If X is selfadjoint, then spec(X ) R. Letus illustrate with a very simple example: chose
1
2i
for X the matrix X =
. We compute easily
2i 2
Eigenvalues
Eigenvectors
u()
i
1
p
5
2
2i
1
p
5
1
3
2
Projectors
P ({})
1 2i
1
5 2i
4
4
2i
1
5 2i
1
Hence
X
P ({})
X
{3,2}
1
= (3)
5
1
2i
2i
4
1
+2
5
4
2i
2i
1
The operators P ({3}) and P ({2}) are self-adjoint (hence they correspond to
observables) and are projectors to mutually orthogonal subspaces. They play
the role of yes-no questions for a quantum system (recall remark 2.2.16.)
Now, let H be a pure phase; since u(3) and u(2) are two orthonormal vectors of H (hence also pure phases), they serve as basis to decompose
= 3 u(3) + 2 u(2), with kk2 = |3 |2 + |2 |2 = 1. Thus any pure state ,

with probability | | u(3) |2 is in the pure state u(3) and with probability
| | u(2) |2 is in the pure state u(2).
Compute further
| X =
X
,0 ,00
00 0 u() | P (0 )u(00 )
spec(X )
| |2 .
Yet (| |2 )spec(X ) can be interpreted as a probability on the set of the spectral

values. Hence, the scalar product | X is the expectation of the spectral
values with respect to the decomposition of on the basis of eigenvectors. It is
worth noticing that expectation of a classical random variable X taking values
in a finite set {x 1 , . . . , x n } with probabilities p 1 , . . . , p n respectively, is
EX
=
=
=
n
X
xi p i
i =1
n
X
p
p i xi p i
i =1
n
X
p
p i exp(i i )x i p i exp(i i ),
i =1
with
i R, i = 1, . . . , n. Hence, classically,
EX =
| X with =
p arbitrary
p 1 exp(i 1 )
x1
0
..
..
, verifying kk = 1 and with X =
. We have more
.
.
p
p n exp(i n )
0
xn
over seen that classical probability is equivalent to classical physics; thanks to
the previous lines, it turns out that that it is also equivalent to quantum physics
involving solely diagonal self-adjoint operators as observables. The full flavour
of quantum physics is obtained only when the observables are represented by
non-diagonal self-adjoint operators.
Consider now,
f = P ({})
u() | u()
=
0
if spec(X )
otherwise.
The vector f is in general unnormalised; the corresponding normalised state

P ({})
= kP ({})k , well defined when spec(X ), has a very particular interpretation. Suppose we ask the question: does the physical observable O X takes the

value 3? The answer, as in the classical case, is a probabilistic one: P({O X =
3}) = |3 |2 = f 3 | f 3 = kP ({3})k2 . What is new, is that once we have
asked this question, the state is projected on the eigenspace P ({3})H and is
represented by the state 3 . This means that asking a question on the system
changes its state! This is a totally new phenomenon without classical counterpart. Asking questions about a quantum system corresponds to a quantum measurement. Hence, the measurement irreversibly changes (projects)
the state of the system.
Summarising the interpretation of the three axioms, we have learnt that
quantum mechanics has a probabilistic interpretation, generalising the

classical probability theory to a quantum (non-Abelian) one,
quantum evolution is reversible,
quantum measurement is irreversible.
Were only to consider this generalisation of probability theory to a noncommutative setting and to explore its implications for explaining quantum
physical phenomena, should the enterprise be already a fascinating one. But
there is even much more fascination about it: there has been demonstrated
lately that quantum phenomena can serve to cipher messages in an unbreakable way and these theoretical predictions have already been exemplified by
currently working pre-industrial prototypes23 .
In a more speculative perspective, it is even thought that in the near future there will be manufactured computers capable of performing large scale
computations using quantum algorithms24 . Should such a construction be realised, a vast family of problems in the (classical) complexity class of exponential time could be solved in polynomial time on a quantum computer.
23
See the article [14], articles in Le Monde (they can be found on the website of this course),
the website www.idquantique.com of the company commercialising quantum cryptologic and
teleporting devices, etc.
24
Contrary to the quantum transmitters and cryptologic devices that are already available
(within pre-industrial technologies), the prototypes of quantum computers that have been
manufactured so far have still extremely limited scale capabilities.
2.5 First consequences of quantum formalism

2.5.1 Irreducibility of quantum randomness and Heisenbergs
uncertainty principle
The first and more spectacular direct consequence of the quantum mechanical
formalism is the so called Heisenbergs uncertainty principle establishing the
conceptual and practical impossibility of considering systems with arbitrarily
small randomness. Spectral decomposition allows computation of the expectation of an operator X , in a pure state, , by
E X = | X =
X
spec(X )
| |2
and when the operator X is self-adjoint, the spectrum is real and the expectation is then a real number. What makes quantum probability different from
classical one, is (among other things) the impossibility of simultaneous diagonalisation of two non-commuting operators. Following the probabilistic interpretation, denote by Var (X ) = E (X 2 ) (E (X ))2 .
Theorem 2.5.1 (Heisenbergs uncertainty). Let X , Y be two bounded self-adjoint
operators on a Hilbert space H and suppose a fixed pure state is given. Then
q
Var (X )Var (Y )
| | [X , Y ] |
.
2
Proof. First notice that (i [X , Y ]) = i [X , Y ] thus the commutator is skew-adjoint.

Without loss of generality, we can assume that E X = E Y = 0 (otherwise consider X E X and similarly for Y .) Now, | X Y = + i , with , R.
Hence, | [X , Y ] = 2i and obviously
0 42 = | | [X , Y ] |2
4| | X Y |2
4 | X 2 | Y 2 ,
the last inequality being Cauchy-Schwarz.
This is a typically quantum phenomenon without classical counterpart. In

fact, given two arbitrary classical random variables X , Y on a measurable space
(, F ), there exists always states (i.e. probability measures) on (, F ) such that
Var(X )Var(Y ) = 0 (for instance chose P(d ) = 0 (d ).
2.5. First consequences of quantum formalism
2.5.2 Quantum explanation of the Orsay experiment
3
Short resum of Hilbert spaces
For the sake of completeness, some standard results on Hilbert spaces are reminded in this chapter. Most of the proofs in this chapter are omitted because
they are considered as exercises; they can be found in the classical textbooks
[1, 16, 23, 25] which are strongly recommended for further reading.
3.1 Scalar products and Hilbert spaces

Hilbert spaces have many distinct features. They are C-vector spaces (hence
are algebraic objects) equipped with a Hermitean scalar product (hence angles
and geometry follow) from which a Hilbert norm can be defined (they thus become analytic objects) for which they are complete s (hence a unit vector has
P
components on an orthonormal basis (e n ) reading n |n |2 = 1; therefore the
components can be interpreted as probability amplitudes).
Definition 3.1.1. Let X be a C-vector space, x, y, z X, and , C. A form
s : X X C is a a scalar product if it verifies
it is linear with respect to the second argument: s(x, y + z) = s(x, y) +
s(x, z),
31
3.1. Scalar products and Hilbert spaces

it is Hermitean: s(x, y) = s(y, x),
it is positive: s(x, x) 0, and
it is definite: s(x, x) = 0 x = 0.
The scalar product s is denoted usually | and the pair (X, | ) is called a
pre-Hilbert space.
Lemma 3.1.2. Let (X, | ) be a pre-Hilbert space and x, y X. The following
hold:
Cauchy-Schwarz-Buniakovski inequality: | x | y | x | x y | y ,
p
The scalar product defines uniquely a norm kxk = x | x , called the Hilbert
norm, and is compatible with the Hilbert norm, i.e.
x |y =
3
1X
i k kx + i k yk2 .
4 k=0
The function | : X X C is continuous.

The scalar product of a pre-Hilbert space (X, | ) induces a Hilbert norm
p
kxk = x | x turning (X, k k) into a normed space; furthermore, the Hilbert
norm defines a distance d (x, y) = kx yk, turning thus (X, d ) into a metric
space. We can therefore define the notion of a fundamental (Cauchy) sequence
on it as being a sequence x = (x n )nN of vectors x n X such that for every > 0
there exists N = N () N such that for m, n N , we have d (x n , x m ) < . Nevertheless, nothing imposes that any fundamental sequence converges within
X. If such is the case, the metric space (X, d ) (or the normed space (X, k k) is
termed complete or Banach space. A pre-Hilbert space that is complete for the
metric induced by its Hilbert norm is called a Hilbert space.
, d) is called a completion of a
We recall that a complete metric space (X
such that
metric space (X, d ) if there exists an isometric embedding : X X
. An arbitrary normed space (not necessarily comthe image (X) is dense in X
plete) can be completed via a standard procedure we recall here briefly. Let
CS(X) be the set of Cauchy sequences on X, an equivalence relation on CS(X)
defined by
x y lim kx n y n k = 0,
n
= CS(X)/ the set of equivalence classes of . The space X

acquires
and X
naturally a vector space structure and through the definition
3 [x] 7 k[x]k = lim kx n k
X
n
/Users/dp/a/ens/mq/iq-hilbe.tex
32
lud on 2 October 2013 at 13:45
that can be shown to be independent of the representative x of [x] becomes a complete normed space. On identifying elements of X with constant
we establish a canonical embedding : X X
. One can show that
sequences X
the completion is unique up to isomorphisms (see [16, 1.6, pp. 1721] for the
details).
Exercise 3.1.3. The following are classical examples.
1. The finite-dimensional vector space X = Cd with the ordinary scalar prodP
uct x | y = dn=1 x n y n is obviously a Hilbert space.
P
2. The space X = `p (N) = {x : N C; nN |x n |p < } is a complete normed
P
space for all p 1, with norm kxkp = ( nN |x n |p )1/p . In the particular
case p = 2 it becomes a Hilbert space with a scalar product, compatible
P
with k k2 , defined by x | y = nN x n y n . It is the infinite-dimensional
generalisation of the previous example.
3. The space X = L p ([a,
R b], ) with p 1 can be equipped with a norm k kp
defined by kxkp = ( [a,b] |x(t )|p (d t ))1/p . Then (X, kkp ) is a Banach space.
For
R p = 2, it becomes also a Hilbert space for the scalar product x | y =
[a,b] x(t )y(t )(d t ).
4. The space X = C k ([a, b]) can be equipped with a scalar product x | y =
Pk R
(j)
(j)
j =0 [a,b] x (t )y (t )d t . The corresponding norm is denoted k kW k,2
but the normed space (X, kkW k,2 ) is not complete. Its completion is called
Sobolev space W k,2 ([a, b]) and is a Hilbert space. In particular, W 0,2 = L 2 .
Definition 3.1.4. Let X : H1 H2 be a linear operator between Hilbert spaces
H1 and H2 . Its norm is defined by
kX k := kX kH1 ,H2 =
sup
hH1 :khkH1
kX hkH2
khkH1
The definition holds also in the special case when H2 = C (in which case the
operator is a form). When kX k < the operator is termed bounded. If an
operator is unbounded it can only be defined on its domain Dom(X ).
3.2 Orthogonality and projection; direct and orthogonal sums

In this section H will always denote a Hilbert space equipped with a scalar product | and corresponding Hilbert norm k k.
33
3.2. Orthogonality and projection; direct and orthogonal sums

Definition 3.2.1.
1. Two vectors g , h H are orthogonal if g | h = 0.
2. Two subsets A, B H are called orthogonal if a A and b B , we have
a | b = 0.
3. If A H, its orthogonal complement is defined as
A := {h H : a A, a | h = 0} .
Theorem 3.2.2. If A H, then A is a Hilbert subspace (closed vector subspace)
of H.
Theorem 3.2.3.
1. Let K H be a closed convex subset of H. Then

h H, !k 0 K : kh k 0 k = inf kh kk.
kK
2. Let K be a Hilbert subspace of H. Then

(a) h H, !k 0 K : kh k 0 k = infkK kh kk, and
(b) the vector k 0 is the unique element of K such that (h k 0 ) K.
Exercise 3.2.4. Let L 1 (, F , P; R) denote the vector space of integrable random variables over some probability space (, F , P) and G a sub--algebra of
2
F . Denote by X = L 2 (, F , P; R) and
R Y = L (, G , P; R). On X a sesquilinear
form s can be defined by s(X , Y ) = X ()Y ()P(d ) that can be turned into a
scalar product by considering the space L 2 instead of L 2 .
1. Use theorem 3.2.3 to establish that for every X X there exists a Y Y
(unique up to modifications differing from X on P-negligible sets), such
that X Y Z for all Z Y.
2. Using the previous result, with Z = 1 G for an arbitrary G G , to establish
that Y is a version of the (classical) conditional expectation E(X |G ).
3. Use the density of L 2 into L 1 and the monotone convergence theorem
to establish that for every random variable X L 1 (, F , P; R), there exists a random variable Y L 1 (, G , P; R) verifying
G G ,
Z
G
X ()P(d ) =
Z
G
Y ()P(d ).
Definition 3.2.5. Let X be a vector space and V, W vector subspaces of X.

1. If for every x X, there exists a unique v V (and consequently a unique
w W) such that x = v + w, we say that X is the direct sum of V and W
and write X = V W.
34
2. If X = V W, define a linear operator P X V by

X 3 x = v + w 7 P x = P (v + w) := v V.
Remark 3.2.6. The decomposition X = V+W is unique if and only if VW = {0};

we write then X = V W. If P is the operator defined in 3.2.5, obviously P 2 x =
P 2 (v + w) = P v = v = P x. Hence P 2 = P . Additionally im P = V and ker P = W.
Definition 3.2.7. A projection on a vector space X is a linear operator P : X X
satisfying the condition P 2 = P .
There exists a bijection between projections and decompositions in direct
sums as stated in the following
Theorem 3.2.8. Let X be a vector space.
1. If a linear operator P is a projection on X, then X = im P ker P .
2. If V and W are vector subspaces of X such that X = V W, there exists a
projection P on X such that im P = V and ker P = W.
When the vector space X is a Hilbert space (hence equipped with a scalar
product) and V a Hilbert subspace of X, we can consider the case W = V . Obviously then we can decompose the space into the direct sum X = V V . The
projection operator is then defined analogously, thanks to the theorem 3.2.8,
but now we have for x = v + w and x 0 = v 0 + w 0 that
P x | x 0 = v | v 0 + w 0 = v | v 0 = v | P x 0 = x | P x 0 .
We have thus:
Definition 3.2.9. A linear operator P : H H is an orhtoprojection on H if
P is a projection (i.e. P 2 = P ) and for every pair h, h 0 H, we have Ph | h 0 =
h | Ph 0 (i.e.1 P = P ).
Again we can establish a bijection between orthoprojections and decompositions into orthogonal Hilbert subspaces as shown in the next
Theorem 3.2.10.
1. If P is an orthoprojection on H, then im P is closed and
H = im P ker P .
2. If V is a Hilbert subspace of H then there exists an orthoprojection P on H
such that im P = V and ker P = V .
1
See ?? below.
35
3.3. Duality and Frchet-Riesz theorem; adjoint

Exercise 3.2.11. Let H = L 2 (R).
1. If V is the Hilbert subspace of even square integrable functions, then V is
the Hilbert subspace of odd square integrable functions, then P,Q defined
by
h(x) h(x)
h(x) + h(x)
, and Qh(x) =
Ph(x) =
2
2
are orthoproejections on H.
2. If A B(R) and V = {h H : h = 1 A h}, then V is a vector subspace of H not
necessarily closed. Nevertheless, an orthoprojection P can be associated
with the decomposition H = V V , where V is the Hilbert subspace of
functions with support contained in A.
Exercise 3.2.12. An orthoprojection P 6= 0 on H has norm kP k = 1.
Complter: exemples de mesures trangres et de somme infinie dnombrable despaces.
3.3 Duality and Frchet-Riesz theorem; adjoint

Complter.
3.4 Tensor product of Hilbert spaces

The notion of algebraic tensor product is given in 2.4.4 Here we extend this notion to hold in the case where the factor spaces of the tensor product are not
mere vector spaces but (infinite dimensional) separable Hilbert spaces. Intuitively, we interpret the tensor map : G H L as a kind of product (g , h) =
g h L . We wish to equip L with a scalar product rendering it a pre-Hilbert
space that will be ultimately be completed for the Hilbert norm to a Hilbert
space L. This construction has been carried on in [42, 3.4, pp. 4749] and
more extensively in [17, II.4, pp. 4954] that, although based on the ideas of
[42] is more direct. When the factors of the tensor product are Banach spaces,
this construction is carried on in [27]. The construction proposed here follows
the [23, pp. 4954] account of [42].
Let us now explain the main steps of the construction.
36
Let G, H be separable Hilbert spaces. For every G and H, construct

the coniugate bilinear form defined by
G H 3 (g , h) 7 (g , h) := g | G h | H ,
and consider the linear manifold of the finite linear combination of such forms
(
L :=
n
X
)
i
c i , G, H, i = 1, . . . , n, n N .
i =1
Obviously L is a vector space. A sesquilinear form | L is defined by its action on simple tensor products in L , by
| 0 0 L := | 0 G | 0 H
and extended by linearity on all elements of L .
Lemma 3.4.1. The sesquilinear form | L is
1. well defined, and

2. positive defined.
Proof.
1. No particular hypothesis has been made on the vectors i and i

entering in the decomposition of forms in L ; therefore, the decomposition of any bilinear form L into elementary tensor product forms
is not necessarily unique. To establish well definiteness of | 0 L , we
must show that the result is independent of the representation used in
and 0 . It is enough to show that if denotes the zero form, then
P
| L = 0 for any L . Let = iM=1 c i i i be an arbitrary form.
Then
M
M
X
X
c i i i | =
c i (i , i ) = 0,
| L =
i =1
i =1
establishing thus well definiteness.

P
2. Let again = iM=1 c i i i and Complter.
37
3.5. Orhonormal systems, ortonormal bases; criteria of completeness of a basis
3.5 Orhonormal systems, ortonormal bases; criteria of completeness of a basis

3.6 Diracs bra and ket notation
Diracs notation is a very convenient shorthand notation for dealing with Hilbert
spaces, scalar products, tensor products and forms in quantum mechanics.
Usual notation
Diracs notation
Orthonormal basis (e 1 , . . . , e n )
P
= i i e i
P
| = i i
H = { f : H C, linear}
: H H
: 7 f (() = |
| = f ()
X = X
|X = X | = X |
X u(i ) = i u(i )
P ({i }) projector
P
X = i i P ({i })
Tensor product
n symbols, eg. {e 1 , . . . , e n }
|e 1 , . . . |e n
P
| = i i |e i
P
| = i i
H = { f : H C, linear}
: H H
: | 7 |
| = ||
X = X
|X |
X |i = i |u(i )
|u( ) u(i )|
P i
X = i i |u(i ) u(i )|
| | = |
Exercise 3.6.1. Let (e n )nN be an orthonormal basis of a Hilbert space H.
1. What is the interpretation of | e n e n | for some n?

2. If and are unit vectors of H, what is the interpretation of | |?
P
s
s
3. What is the interpretation of the identity nN | e n e n | = I (where = denotes the strong limit of the partial sums)?
4. Let H = L 2 (T) and (e n ) the basis of trigonometric polynomials e n (t ) =
exp(i nt ). Derive the Parseval formula using the Dirac formalism.
Complter avec dcomposition matricielle, produit tensoriels.

38
3.7 Some more involved quantum mechanical phenomena

3.7.1 Composite systems and entanglement
3.7.2 Decoherence and quantum to classical transition
3.8 Rigged Hilbert spaces and generalised kets
39
3.8. Rigged Hilbert spaces and generalised kets
40
4
Algebras of operators
4.1 Introduction and motivation

Let V = Cn , with n N. Elementary linear algebra establishes that the set of
linear mappings L(V) = {T : V V : T linear } is a C-vector space of (complex)
dimension n 2 , isomorphic to Mn (C), the space of n n matrices with complex
coefficients. Moreover, if S, T L(V), the maps S and T can be composed,
their composition T S being represented by the corresponding matrix product. Thus, on the vector space L(V), is defined an internal multiplication
L(V) L(V) 3 (T, S) 7 T S L(V)

turning this vector space into an algebra.
When the underlying vector space V is of infinite dimension, caution must
be paid on defining linear maps. In general, linear mappings T : V V, called
(linear) operators, are defined only on some proper subset of V denoted Dom(T )
and called the domain1 of T . When V is a normed space, there is a natural way
to define a norm on L(V). We denote by B(V) the vector space of bounded lin1
The set Dom(T ) is generally a linear manifold, i.e. algebraically a vector subspace of V
which is not necessarily topologically closed.
41
4.2. Algebra of operators

ear operators on V, i.e. linear maps T : V V such that kT k < (equivalently,
verifying Dom(T ) = V.) When H is a Hilbert space, bounded linear operators
on H, whose set is denoted by B(H), with operator norm kT k = sup{kT xk, x
H, kxk 1}, share the properties of linear operators defined on more algebraic
setting. Sometimes it is more efficient to work with explicit representations of
operators in B(H) (that play the rle of matrices in the infinite dimensional
setting) and some others with abstract algebraic setting.
Since all operators encountered in quantum mechanics are linear, we drop
henceforth the adjective linear.
4.2 Algebra of operators

Definition 4.2.1. An algebra is a set A endowed with three operations:
1. a scalar multiplication C A 3 (, a) 7 a A,
2. a vector addition A A 3 (a, b) 7 a + b A, and
3. a vector multiplication A A 3 (a, b) 7 ab A,
such that A is a vector space with respect to scalar multiplication and vector
addition and a ring (not necessarily commutative) with respect to vector addition and vector multiplication. Moreover, (ab) = (a)b = a(b) for all C
and all a, b A. The algebra is called commutative if ab = ba, for all a, b A;
it is called unital if there exists (a necessarily unique) element e A (often also
written 1 or 1 A ) such that ae = ea = a for all a A;
A linear map from an algebra A1 to an algebra A2 is a homomorphism if it
is a ring homomorphism for the underlying rings, it is an isomorphism if it is a
bijective homomorphism.
Definition 4.2.2. An involution on an algebra A is a map A 3 a 7 a A that
verifies
1. (a + b) = a + b ,
2. (ab) = b a , and
/Users/dp/a/ens/mq/iq-algop.tex
42
lud on 1 October 2013
3. (a ) = a.
Involution is also called adjoint operation and a the adjoint of a. An involutive algebra is termed a -algebra.
An element a A is said normal if aa = a a, an isometry if a a = 1 , unitary if both a and a are isometries, self-adjoint or Hermitean if a = a . On
denoting h : A1 A2 a homomorphism between two -algebras, we call it a
-homomorphism if it preserves adjoints, i.e. h(a ) = h(a) .
A normed (respectively Banach) algebra A is an algebra equipped with a
norm map k k : A R+ that is a normed (respectively Banach) vector space for
the norm and verifies kabk kakkbk for all a, b A. A is normed (respectively
Banach) -algebra if it has an involution verifying ka k = kak for all a A.
Theorem 4.2.3. Let T : H1 H2 be a linear map between two Hilbert spaces H1
and H2 . Then the following are equivalent:
1. kT k = sup{kT f kH2 , f H1 , k f kH1 1} < ,
2. T is continuous,
3. T is continuous at one point of H1 .
Proof: Analogous to the proof of the theorem ?? for linear functional. (Please
complete the proof!)
Notation 4.2.4. We denote by B(H1 , H2 ) the algebra of bounded operators with

respect to the aforementioned norm:
B(H1 , H2 ) = {T L(H1 , H2 ) : kT k < }.

When H1 = H2 = H, we write simply B(H).
Proposition 4.2.5. Let H1 and H2 be two Hilbert spaces and T B(H1 , H2 ).
Then, there exists a unique bounded operator T : H2 H1 such that
T g | f = g | T f for all f H1 , g H2 .
Proof: For each g H2 , the map H1 3 f 7 g | T f H2 C is a continuous (why?)
linear form. By Riesz-Frchet theorem ??, there exists a unique h H1 such
that h | f H1 = g | T f H2 , for all f H1 . Let T : H2 H1 be defined by the
assignment T g = h; it is obviously linear and easily checked to be bounded
(exercise!)
43
4.2. Algebra of operators

Proposition 4.2.6. For all T B(H1 , H2 ),
1. kT k = kT k,
2. kT T k = kT k2 .
Proof:
1. By Cauchy-Schwarz inequality, for all f H2 , g H1 ,
| f | T g H2 | k f kH2 kT g kH2
kT kB(H1 ,H2 ) kg kH1 k f kH2
so that
kT k sup{| f | T g | : kg k 1, k f k 1}.
Conversely, we may assume that kT k 6= 0, and therefore choose some
]0, kT k/2[. Choose now g H1 with kg k 1, such that kT g k kT k and
Tg
f = kT g k H2 , k f k = 1. For this particular choice of f and g :
| f | T g H2 | kT g k kT k .
Hence,
sup{| f | T g | : kg k 1, k f k 1} kT k .
Since is arbitrary, we get kT k = sup{| f | T g H2 | : g H1 , f H2 , kg k
1, k f k 1}. As f | T g = T f | g for all f and g , we get kT k = kT k
2. B(H1 , H2 ) being a normed algebra, kT T k kT kkT k = kT k2 . Conversely,
kT k2 sup{|T f k : f H1 , k f k 1}
= sup{| T f | T f | : f H1 , k f k 1}
= sup{| f | T T f | : f H1 , k f k 1}
kT T k.
Definition 4.2.7. A C -algebra A is an involutive Banach algebra verifying additionally

ka ak = kak2 , for all a A.
44
Example 4.2.8. Let X be a compact Hausdorff2 space and A = { f : X C | f continuous}

C (X) Define
1. C A 3 (, f ) 7 f A by ( f )(x) = f (x), x X,
2. A A 3 ( f , g ) 7 f + g A by ( f + g )(x) = f (x) + g (x), x X,
3. A A 3 ( f , g ) 7 f g A by ( f g )(x) = f (x)g (x), x X,
4. A 3 f 7 f A by f (x) = f (x), x X,
Then A is a unital (specify the unit!) C -algebra for the norm k f k = supxX | f (x)|.
(Prove it!) The algebra A is moreover commutative.
Example 4.2.9. Let H1 and H2 be two Hilbert spaces. Then B(H1 , H2 ) is a unital
C -algebra. In general, this algebra is not commutative.
This example has also a converse, given in theorem 4.5.2, below.
Example 4.2.10. Let X be a compact Hausdorff space. Then A = C (X) := { f :
X C, continuous} equipped with the uniform norm and pointwise multiplication is a unital Banach commutative algebra; further equipped with an involution defined by complex conjugation, becomes a B -algebra.
Example 4.2.11. A = `1 (Z), with complter.
Example 4.2.12. A = L 1 (R), with complter.
The previous example must not induce the reader to erroneously conclude
that non-unital algebras have natural approximate identities since complter.
4.3 Convergence of sequences of operators

4.4 Classes of operators in B(H)
We shall see that any C -algebra can be faithfully represented on some Hilbert
space H; the different classes of abstract elements of the algebra, introduced
2
Recall that a topological space is called Hausdorff when every two distinct of its points
posses disjoint neighbourhoods.
45
4.4. Classes of operators in B(H)

in the previous section, have a counterpart in the context of this representation. But additionally, B(H) is a very special C -algebra because is closed for
the weak operator topology (defined in 4.3). This fact endows B(H) with a
very rich family of projections allowing to generate3 back the unital C -algebra
B(H).
4.4.1 Self-adjoint and positive operators

Definition 4.4.1. An operator T B(H) is called self-adjoint or Hermitean4 if
T = T . The set of Hermitean operators on H is denoted by Bh (H).
Exercise 4.4.2. The operator T B(H) is self-adjoint if and only if f | T f R
for all f H. (Hint: use the polarisation equality ??.)
Exercise 4.4.3. If T B(H) is self-adjoint then kT k = sup{ f | T f , f H, k f k
1}.
Definition 4.4.4. An operator T B(H) is called positive if f | T f 0 for all
f H. Such an operator is necessarily self-adjoint. We denote by B+ (H) the set
of positive operators.
Exercise 4.4.5. Show that T B+ (H) if and only if there exists S B(H) such
that T = S S.
4.4.2 Projections
Definition 4.4.6. Let P, P 1 , P 2 B(H).
1. P is a projection if P 2 = P .
2. P is an orthoprojection if is a projection satisfying further P = P .
3. Two orthoprojections P 1 , P 2 B(H) are orthogonal, denoted P 1 P 2 of
their images are orthogonal subspaces of H (equivalently P 1 P 2 = 0).
3
In some general unital C -algebras there are only two trivial projections 0 and 1. Therefore
the situation arising in B(H) is far from being a general property of C -algebras.
4
Strictly speaking, the term Hermitean is more general; it applies also to unbounded operators and it means self-adjoint on a dense domain. The two terms coincide for bounded
operators.
46
Projections are necessarily positive (why?). The set of orhoprojections is denoted by P(H). All projections considered henceforth will be orthoprojections.
Exercise 4.4.7. (A very important one!) Let (P n ) be a sequence of orthoprojections. We have already shown that there is a bijection between P(H) and the set
of closed subspaces of H and orthoprojections, given by P(H) 3 P 7 P (H) H,
with P (H) closed.
1. Show that that P(H) is partially ordered, i.e. P 1 P 2 if P 1 (H) subspace of
P 2 (H) (equivalently P 1 P 2 = P 1 .)
2. For general orthoprojections P 1 and P 2 , is P 1 P 2 an orthoprojection?
3. Show that P 1 and P 2 have a least upper bound.
4. Is Q = P 1 + . . . + P n an orthoprojection?
5. Is Q = P 2 P 1 an orthoprojection?
6. Show that a monotone sequence of orthoprojections converges strongly
towards an orthoprojection.
4.4.3 Unitary operators

Definition 4.4.8. An operator U B(H) is unitary if U U = UU = 1 . The set
of unitary operators is denoted by U(H) = {U B(H) : U U = UU = 1 } (it is in
fact a group; for H = Cn it is the Lie group denoted by U (n).)
4.4.4 Isometries and partial isometries

Definition 4.4.9. An operator T B(H1 , H2 ) is an isometry if T T = 1 (or
equivalently kT f k = k f k, for all f H1 .)
Exercise 4.4.10. Let H = `2 (N) and for x = (x 1 , x 2 , x 3 , . . .) H, define the left and
right shifts by
Lx = (x 2 , x 3 , . . .) H,
and
R x = (0, x 1 , x 2 , x 3 , . . .) H.
47
4.4. Classes of operators in B(H)

1. Show that R = L.
2. Show that R is an isometry.
3. Determine RanR.
This exercise demonstrates that, in infinite dimensional spaces, isometries

are not necessarily surjective.
Theorem 4.4.11. For T B(H1 , H2 ), the five following conditions are equivalent:
1. (T T )2 = T T ,
2. (T T )2 = T T ,
3. T T T = T ,
4. T T T = T ,
5. there exist closed subspaces E 1 H1 and E 2 H2 such that T = I S P
where P : H1 E 1 is a projection, S : E 1 E 2 an isometry, and I : E 2 H2
the inclusion map.
If one (hence all) condition holds then T T is the projection H1 E 1 and T T
is the projection H2 E 2 . In this situation T is called a partial isometry with
initial space E 1 , initial projection T T , final space E 2 , and final projection T T .
Proof. Exercise! (See [2] or [26].)
Exercise 4.4.12. Let (, F , P) be a probability space and T : a measure

preserving transformation i.e. P(T 1 B ) = P(B ) for all B F . On the Hilbert
space H = L 2 (, F , P) define U : H H by U f () = f (T 1 ).
1. Show that U is a partial isometry.

2. Under which condition is U surjective (hence unitary)?
48
4.4.5 Normal operators

Definition 4.4.13. An operator T B(H) is normal if T T = T T (or equivalently if kT f k = kT f k for all f H.)
Exercise 4.4.14. A vector f H \ {0} is called an eigenvector corresponding to
an eigenvalue of an operator T B(H) if T f = f for some C. Show that if
T is normal and f 1 , f 2 are eigenvectors corresponding to different eigenvalues
then f 1 f 2 . (The proof goes as for the finite dimensional case.)
Exercise 4.4.15. Let M be the multiplication operator on L 2 [0, 1] defined by
M f (t ) = t f (t ), t [0, 1]. Show that
1. M is self-adjoint (hence normal),
2. M has no eigenvectors.
Exercise 4.4.16. Choose some z C with |z| < 1 and consider `2 (N) given
by = (1, z, z 2 , z 3 , . . .). Let L and R be the left and right shifts defined in exercise
4.4.10.
1. Show that R is not normal,
2. compute R ,
3. conclude that R has uncountably many eigenvalues.
4.5 States on algebras, GNS construction, representations

Paragraphe incomplet.
Definition 4.5.1. Let A be an involutive Banach algebra. A representation on
a Hilbert space H of A is a -homomorphism of A into B(H), i.e. a linear map
: A B(H) such that
1. (ab) = (a)(b), a, b A,
2. (a ) = (a) , a A,
49
4.5. States on algebras, GNS construction, representations

The space H is called the representation space. We write (, H), or H if necessary. Two representations (1 , H1 ) and (2 , H2 ) are said to be unitarily equivalent if there exists an isometry U : H1 H2 such that for all a A, it holds
U 1 (a)U = 2 (a). If moreover for every non zero element of A, (a) 6= 0, then
the representation is called faithful.
Theorem 4.5.2 (Gelfand-Namark). If A is an arbitrary C -algebra, there exists
a Hilbert space H and a linear mapping : A B(H) that is a faithful representation of A.
Proof: It can be found in [17, theorem 4.5.6, page 281].
50
5
Spectral theory in Banach algebras
5.1 Motivation
In linear algebra one often encounters systems of linear equations of the type
Tf =g
(5.1)
with f , g Cn and T = (t i , j )i , j =1,...,n a n n matrix with complex coefficients. Elementary linear algebra establishes that this system of equations has solutions
provided that the map f 7 T f is surjective and the solution is unique provided
that this map is injective. Thus the system has a unique solution for each g Cn
provided that the map is bijective, or equivalently the matrix T is invertible.
This happens precisely when det T 6= 0. However, this criterion of invertibility is of limited practical use even for the elementary (finite-dimensional) case
because det is too complicated an object to be efficiently computed for large
n. For infinite dimensional cases, this criterion becomes totally useless since
there is no infinite dimensional analogue of det that discriminates between invertible and non-invertible operators T (see exercise 5.1.1 below!)
Another general issue connected with the system (5.1) is that of eigenvalues. For every C, denote by V = { f Cn : T f = f }. For most choices of ,
51
5.1. Motivation
the subspace V is the trivial subspace {0}; this subspace is not trivial only when
T 1 is not injective (i.e. ker(T 1 ) 6= {0}.) On defining the spectrum of T by
spec(T ) = { C : T 1 is not invertible },

one easily shows that spec(T ) 6= ; and cardspec(T ) n (why?) Not always the
family (V )spec(T ) spans the whole space Cn . When it does, on decomposing
g = g (1) + . . . + g (k) where g ( j ) V j and spec(T ) = {1 , . . . , k }, the solution of
(5.1) is given by
g (k)
g (1)
+...+
.
f =
1
k
(Notice that i 6= 0, for all i = 1, . . . , k; why?) When the family (V )spec(T ) does
not span Cn , the problem is more involved but the rle of the spectrum remains
fundamental.
A final issue involving the spectrum of T is the functional calculus associated with T . If p R[t ], this polynomial can be naturally extended on B(H).
In fact, if p(t ) = a n t n + . . . a 0 is the expression of the polynomial p; the expression p(T ) = a n T n + . . . a 0 1 is well defined for all T B(H). Moreover, if T
Bh (H) then p(T ) Bh (H). Suppose now that T Bh (H), m = infk f k=1 f | T f ,
M = supk f k=1 f | T f , and p(t ) 0 for all t [m, M ]; then p(T ) B+ (H). Now
every f C [m, M ] can be uniformly approximated by polynomials, i.e. there
is a sequence (p l )l N , with p l R[t ] such that for all > 0, there exists n 0 N
such that for l n 0 , maxt [m,M ] | f (t ) p l (t )| < . It is natural then to define
f (T ) = liml p l (T ). However, the computations involved in the right hand side
of this equation can be very complicated. Suppose henceforth that H = Cn
and T
n n matrix that is diagonalisable, i.e. T = U DU with
is a Hermitean
..
D =
and U unitary. Then p l (T ) = Up l (D)U and letting l
.
n
we get f (T ) = U f (D)U . Thus, if T is diagonalisable, the computation of f (T ) is
equivalent to the knowledge of f (t ) for t spec(T ). For the infinite dimensional
case, the problem is more involved but again the spectrum remains fundamental.
The rest of this chapter, based on [2], is devoted to the appropriate generalisation of the spectrum for infinite dimensional operators.
Exercise 5.1.1. (Infinite-dimensional determinant) Let H = `2 (N) and (t n )nN
be a fixed numerical sequence. Suppose that there exist constants K 1 , K 2 > 0
such that 0 < K 1 t n K 2 < for all n N. For every x `2 (N) define (T x)n =
t n x n , n N.
/Users/dp/a/ens/mq/iq-sptba.tex
52
1. Show that T B(H).

2. Exhibit a bounded operator S on H such that ST = T S = 1 .
3. Assume henceforth that (t n )nN is a monotone sequence. Let n (T ) =
t t n . Show that n (T ) converges to a non-zero limit (T ) if and only if
P1
n (1 t n ) < .
4. Any plausible generalisation, , of det in the infinite dimensional setting
should verify (1 ) = 1, (AB ) = (A)(B ), and if T is diagonal (T ) = (T ).
n
Choosing t n = n+1
, for n N, conclude that although T is diagonal and
invertible, nevetheless has (T ) = 0.
5.2 The spectrum of an operator acting on a Banach

space
Let V be a C-Banach space. Denote by B(V) the set of bounded operators T :
V V. This space is itself a unital Banach algebra for the induced operator
norm.
Exercise 5.2.1. If X and Y are metric spaces and d X and d Y denote their respective metrics
1. verify that
d p ((x 1 , y 1 ), (x 2 , y 2 )) = (d X (x 1 , x 2 )p + d Y (y 1 , y 2 )p )1/p ,
with p [1, [ and
d ((x 1 , y 1 ), (x 2 , y 2 )) = max(d X (x 1 , x 2 ), d Y (y 1 , y 2 ))
are metrics on X Y; (the corresponding metric space (X Y, d p ), p
[1, ] is denoted1 X Y)
2. show that the sequence (x n , y n )n in X Y converges to a point (, )
X Y with respect to any of the metrics d p if and only if d X (x n , ) 0 and
d Y (y n , ) 0.
1
more precisely X `p Y.
53
5.2. The spectrum of an operator acting on a Banach space

Exercise 5.2.2. Let X and Y be metric spaces and f : X Y be a continuous
map. We denote by
( f ) = {(x, f (x)) : x X}
the graph of f . Show that ( f ) is closed (i.e. if (x n )n is a sequence in X and
if there exists (x, y) X Y such that x n x and f (x n ) y, then necessarily
y = f (x).)
Exercise 5.2.3. (The closed graph theorem) Suppose X and Y are Banach spaces
and T : X Y a linear map having closed graph. Show that T is continuous.
Theorem 5.2.4. For every T B(V), the following are equivalent:
1. for every y V there is a unique x V such that T x = y,
2. there is an operator S B(V) such that ST = T S = 1 .
Proof: Only the part 1 2 is not trivial to show. Condition 1 implies that T is
invertible; call S its inverse. The only thing to show is the boundedness of S. As
a subset of V V, the graph of S is related to the graph of T . In fact
(S) = {(y, Sy) : y V} = {(T x, x), x V}.
Now T is bounded, hence continuous, so that that the set {(T x, x), x V} is
closed (see exercise 5.2.2.) Thus the graph of S is closed, and by the closed
graph theorem (see exercise 5.2.3), S is continuous hence bounded.
Definition 5.2.5. Let T B(V) where V is a Banach space.

1. T is called invertible if there exists an operator S B(V) such that ST =
TS =1.
2. The spectrum of T , denoted by spec(T ), is defined by
spec(T ) = { C : T 1 is not invertible}.

3. The resolvent set of T , denoted by Res(T ), is defined by
Res(T ) = C \ spec(T ).
Notice that in finite dimension, invertibility of an operator R reduces essentially to injectivity of R since surjectivity of R can be trivially verified if we
reduce the space V into Ran(R). In infinite dimension, several things can go
54
wrong: of course injectivity may fail as in finite dimension; but a new phenomenon can appear when Ran(R) is not closed: in this latter case, Ran(R) can
further be dense in V or fail to be dense in V . All these situations may occur
and correspond to different types of sub-spectra.
Definition 5.2.6. Let T B(V) where V is a Banach space.
1. The point spectrum of T is defined by
spec(T ) = { C : T 1 is not injective}.

p
Every specp (T ) is called an eigenvalue of T .

2. The continuous spectrum, specc (T ), of T is defined as the set of complex
values such that T 1 is injective but not surjective and Ran(T 1 )
is dense in V.
3. The residual spectrum, specr (T ), of T is defined as the complex values
such that T 1 is injective but not surjective and Ran(T 1 ) is not
dense in V.
Example 5.2.7. Let V be a finite dimensional Banach space and T : V V a
linear transformation (hence bounded.) Since dim ker(T 1 ) + dim Ran(T
1 ) = dim V, it follows that T 1 is injective if and only if Ran(T 1 ) = V.
Therefore specr (T ) = ;. Further, if T 1 is injective, then it has an inverse on
V. Since any linear transformation of a finite dimensional space is continuous,
it follows that (T 1 )1 is continuous, hence specc (T ) = ;. Therefore, in finite
dimension we always have spec(T ) = specp (T ).
Exercise 5.2.8. Let V = `2 (N) and consider the right shift, R, on V.
1. Show that R 1 is injective for all C. Conclude that specp (R) = ;.
2. Show that for || > 1, Ran(R 1 ) = V. Conclude that all C with || > 1
belong to Res(R).
3. For || < 1, show that Ran(R1 ) is orthogonal to the vector = (1, , 2 , . . .).
Show that for || < 1, Ran(R 1 ) = {y V : y }. Conclude that all C
with || < 1 belong to specr (R).
4. The case || = 1 is the most difficult. Try to show that Ran(R 1 ) is dense
in V so that the unit circle coincides with specc (R).
55
5.3. The spectrum of an element of a Banach algebra
5.3 The spectrum of an element of a Banach algebra

In the previous section we studied spectra of bounded operators acting on Banach spaces. They form a Banach algebra with unit. Spectral theory can be
established also abstractly on Banach algebras. Before stating spectral properties, it is instructive to give some more examples.
Example 5.3.1. Let C c (R) be the set of continuous functions on R which vanish
outside a bounded interval; it is a normed vector space (with respect to the L 1
norm for instance; its completion is the Banach space L 1 (R, ), where stands
for the Lebesgue measure.) A product can be defined by the convolution
f ? g (x) =
Z
R
f (y)g (x y)(d y)
turning this space into a commutative Banach algebra. This algebra is not unital (this can be seen by solving the equation f ? f = f in L 1 ), but it has an approximate unit (i.e. a sequence ( f n )n of integrable functions with k f n k = 1 for
all n and such that for all g L 1 (R), kg ? f n g k 0. (Give an explicit example
of such an approximate unit!)
Example 5.3.2. The algebra Mn (C) is a unital non-commutative algebra. There
are many norms that turn it into a finite-dimensional Banach algebra, for instance:
1. kAk =
Pn
i , j =1 |a i , j |
.
2. kAk = supkxk1 kAxk
kxk
Definition 5.3.3. Let A be a unital Banach algebra. (We can always assume
that k1 k = 1, may be after re-norming the elements of A.) An element a A is
called invertible if there is an element b A such that ab = ba = 1 . The set of
all invertible elements of A is denoted by GL(A) and called the general linear
group of invertible elements of A.
Theorem 5.3.4. Let A be a unital Banach algebra. If a A and kak < 1 then
1 a is invertible and
X
(1 a)1 =
an .
n=0
Moreover,
k(1 a)1 k
56
1
1 kak
and
k1 (1 a)1 k
kak
.
1 kak
Proof: Since ka n k kakn for all n, we can define b A as the sum of the absoP
PN
n
n
lutely convergent series b =
n=0 a . Moreover, b(1 a) = (1 a)b = limN n=0 b =
limN (1 b N +1 ) = 1 . Hence 1 a is invertible and (1 a)1 = b. The first maP
1
jorisation holds because kbk
kakn = 1kak
. The second one follows from
P nn=0
remarking that 1 b = n=1 a = ab, hence k1 bk kakkbk.
Exercise 5.3.5.
1. Prove that GL(A) is an open set in A and that the map1
ping a 7 a is continuous on GL(A).
2. Justify the term general linear group of invertible elements, i.e. show
that GL(A) is a topological group in the relative norm topology.
Definition 5.3.6. Let A be a unital Banach algebra. For every a A, the spectrum of a is the set
spec(a) = { C : a 1 6 GL(A)}.
In the rest of this section, A will be a unital algebra and we shall write a
instead of a 1 .
Proposition 5.3.7. For every a A, the set spec(a) is a closed subset of the disk
{ C : || kak}.
Proof: Consider the resolvent set
Res(a) = { C : a GL(A)} = C \ spec(a).

Since the set GL(A) is open (see exercise 5.3.5) and the map C 3 7 a A
continuous, the set Res(a) is open hence the set spec(a) is closed. Moreover,
if || > kak, on writing a = ()[1 a/] and remarking that ka/k < 1, we
conclude that a GL(A).
Theorem 5.3.8. For every a A, the set spec(a) is non-empty.

Proof: Fix some 0 Res(a). Since Res(a) is open, there is a small neighbourhood V of 0 contained in Res(a). The A-valued function 7 (a )1 is well
defined for all V0 . Moreover, for , 0 Res(a),
(a )1 (a 0 )1 = (a )1 [(a 0 ) (a )](a 0 )1
= ( 0 )(a )1 (a 0 )1 .
57
5.4. Relation between diagonalisability and the spectrum

Thus
1
[(a ) (a 0 )] = (a 0 )2 .
0 0
lim
Assume now that spec(a) = ; and choose an arbitrary bounded linear functional : A C. Then, the scalar function f : C C defined by 7 f () =
((a )1 ) is defined on the whole C. By linearity, the function f has everywhere a complex derivative, satisfying f 0 () = ((a )2 ). Thus f is an entire
function. Notice moreover that f is bounded and for || > kak, by theorem
5.3.4,
k(a )1 k =
k(1 a/)1 k
||
1
||(1 kak/||)
1
.
|| kak
Thus lim f () = 0 and since this function is bounded and entire, by Liouvilles theorem (see [?] for instance), it is constant, hence f () = 0 for all C
and every linear functional . The Hahn-Banach theorem implies then that
(a )1 = 0 for all C. But this is absurd because (a ) is invertible and
1 6= 0 in A.
Definition 5.3.9. For every a A, the spectral radius of a is defined by r (a) =
sup{|| : spec(a)}.
Exercise 5.3.10.
1. Let p R[t ] and a A. Show that p(spec(a)) spec(p(a)).
(Hint: if spec(a), the map 0 7 p(0 ) p() is a polynomial vanishing
at 0 = . Conclude that p(a) p() cannot be invertible.)
2. For every a A show that r (a) = limn ka n k1/n .
5.4 Relation between diagonalisability and the spectrum

Motivated again by elementary linear algebra, we recall that a self-adjoint n
n matrix
T can be
diagonalised, i.e. it is possible to find a diagonal matrix

d1
..
D =
and a unitary matrix U such that T = U DU ; we have then
.
dn
58
spec(T ) = {d1 , . . . , dn }. We shall generalise this result to infinite dimensional

spaces.
An orthonormal basis for H is a sequence E = (e 1 , e 2 , . . .) of mutually orthogonal unit vectors of H such that2 spanE = H. On fixing such a basis, we define a
unitary operator U : `2 (N) H by
X
Uf =
fi ei
i N
for f = ( f 1 , f 2 , . . .). Specifying a particular orthonormal basis in H is equivalent to specifying a particular unitary operator U . Suppose now that T B(H)
is a normal operator and admits the basis vectors of E as eigenvectors, i.e.
Te k = t k r k , t k C, k N. Then t = (t k )k ` (N) and U T U = M where M
is the multiplication operator defined by (M f )k = (U T U f )k = (U 1 T U f )k =
P
(U 1 T i f i e i )k = f k t k . Thus an operator T on H is diagonalisable in a given
basis E if the unitary operator associated with E implements an equivalence
between T and a multiplication operator M acting on `2 (N). This notion is still
inadequate since it involves only normal operators with pure point spectrum;
it can nevertheless be appropriately generalised.
Definition 5.4.1. An operator T acting on a Hilbert space H is said diagonalisable if there exist a (necessarily separable) -finite measure space (, F , ), a
function m L (, F , ), and a unitary operator U : L 2 (, F , ) H such that
U Mm = T U
where M m denotes the multiplication operator by m, defined by M m f () =
m() f (), for all and all f L 2 (, F , )
Example 5.4.2. Let H = L 2 ([0, 1]) and T : H H defined by T f (t ) = t f (t ), for
t [0, 1] and f H. This operator is diagonalisable since it is already a multiplication operator.
Notice that a diagonalisable operator is always normal because the multiplication operator is normal. The following theorem asserts the converse.
Theorem 5.4.3. Every normal operator acting on a Hilbert space is diagonalisable.
Proof: Long but without any particular difficulty; it can be found in [2], pp. 52
55.
Le reste du chapitre doit tre re-crit.

2
Recall that H is always considered separable.
59
5.5. Spectral measures and functional calculus
5.5 Spectral measures and functional calculus

Start again from some heuristic ideas. Let (, F , ) be a probability space and
f : R a bounded measurable function. Standard integration theory states
that f can be approximated by simple functions. More precisely, for every > 0,
there exists a finite family (E i )i of disjoint measurable sets E i F and a finite
P
family of real numbers (i )i such that | f () i i 1 E i ()| < for all . It is
instructive to recall the main idea of the proof of this elementary result.
Ij
!1
f (I j)
Figure 5.1: The approximation of a bounded measurable function by simple

functions
Let m = inf f (), M = sup f (), and subdivide the interval [m, M ] into a
finite family of disjoint intervals (I j ) j , with |I j | < (see figure 5.1.) For each j ,
select an arbitrary j I j ; in the subset f 1 (I j ) F , the values of f lie within
from j . Therefore, we get the desired result by setting E j = f 1 (I j ). If for
every Borel set B B(R), we define P (B ) = 1 f 1 (B ) (this is a function-valued set
60
function!), the approximation result can be rewritten as

| f ()
j P (I j )()| < , .
P
Now, P is set function
(a measure actually) and the sum j j P (I j ) tends (in
R
some sense3 ) to P (d ). More precisely, the function f is equivalent to a
deterministic stochastic kernel K := K f from (, F ) to (R, B(R)), defined by the
formula
B(R) 3 (, B ) 7 K (, B ) = f () (B ) = 1 f 1 (B ) ().
The action of the kernel on is

Z
(K )(B ) :=
(d )K f (, B ) = f (B ),
where f denotes the law of f . Conversely, the kernel acts on measurable positive functions g defined on the real axis by
Z
f
(K g )() := K f (, d x)g (x) = g ( f ()).
R
In particular, if g = id then
f
(K id)() =
Z
R
K f (, d x)x = f ().
Summarising the heuristics developed so far: the approximability of a realvalued, bounded,

R measurable function f by simple functions can be expressed
by writing f = P (d ), where P is the function-valued measure
B(R) 3 B 7 P (B ) = 1 f 1 (B )
with the following properties:
1. P is idempotent: i.e. P (B )2 = P (B ) for all B B(R),
2. P is multiplicative: i.e. P (B C ) = P (B )P (C ) for all B,C B(R),
3. P is supported by Ran( f ): i.e. P (B ) 0 for all B B(R) such that B
Ran( f ) = ;.
3
As a matter of fact, it is possible to construct a descent theory of integration in which

P (d ) acquires a precise meaning.
61

The measure P reflects the properties of f ; it is called the spectral measure of
f.
In the non-commutative setting, the analogue of a bounded, real-valued,
measurable function is a bounded Hermitean operator on H. Idempotence,
characterising indicators in the commutative case, is verified by projections belonging to P(H). Hence, we are seeking approximations of bounded Hermitean
operators by complex finite combinations of projections. Now we can turn into
precise definitions.
Definition 5.5.1. Let (X, F ) be a measurable space and H a Hilbert space. A
function P : F P(H) is called a spectral measure on (X, F ) if
1. P (X) = 1 ,
2. if (F n )nN is a sequence of disjoint elements in F , then P (tnN F n ) =
nN P (F n ).
Example 5.5.2. Let (X, F , ) be a probability space and H = L 2 (X, F , ). Then

the mapping F 3 F 7 P (F ) P(H), defined by P (F ) f = 1 F f for all f H, is a
spectral measure.
Exercise 5.5.3. If P is a spectral measure on (X, F ), then P (;) = 0 and P is
finitely disjointly additive.
Theorem 5.5.4. Let (X, F ) be a measurable space and H a Hilbert space. If P
is a finitely disjointly additive function F P(H) such that P (X) = 1 then (for
F,G F )
1. P is monotone: F G P (F ) P (G),
2. P is subtractive: F G P (G \ F ) = P (G) P (F ),
3. P is modular: P (F G) + P (F G) = P (F ) + P (G),
4. P is multiplicative: P (F G) = P (F )P (G).
Proof: The statements 1 and 2 are immediate by noticing that F 1 F 2 F 2 =
F 1 t (F 2 \ F 1 ).
3) Since F G = (F \ G) t (F G) t (G \ F ) we have: P (F G) + P (F G) = [P (F \
G) + P (F G)] + [P (G \ F ) + P (G F )] = P (F ) + P (G).
4) By 1)
P (F G) P (F ) P (F G).
62
()
Multiplying the first inequality of (*) by P (F G), we get P (F G) P (F )P (F

G) and since P (F ) 1 , the right hand side of the latter inequality is bounded
further by P (F G). Hence P (F )P (F G) = P (F G). Similarly, multiplying the
second inequality of (*) by P (F ) and since again P (F G) 1 , we get P (F )P (F
G) = P (F ). Adding the thus obtained equalities, we get:
P (F )[P (F G) + P (F G)] = P (F G) + P (F )
and we conclude by modularity.
Exercise 5.5.5. Show that for all F,G F ,

1. P (F ) is an orthoprojection, and
2. we have [P (F ), P (G)] = 0.
Theorem 5.5.6. Let (X, F ) be a measurable space and H a Hilbert space. A map
P : F P(H) is a spectral measure if and only if
1. P (X) = 1 , and
2. for all f , g H, the set function f ,g : F C, defined by
f ,g (F ) = f | P (F )g , F F ,
is countably additive.
Proof:
(): If P is a spectral measure, then statements 1 and 2 hold trivially.
(): Suppose, conversely, that 1 and 2 hold. If F G = ; then f | P (F G)g =
f | P (F )g + f | P (G)g = f | [P (F ) + P (G)]g , hence P is finitely additive (hence multiplicative). Let now (F n )n be a sequence of disjoint sets
in F . Multiplicativity of P implies (P (F n ))n is a sequence of orthogonal projections and hence (P (F n )g )n a sequence of orthogonal vectors
for any g H. Let F = n F n . Hence, for all f , g H, we have: f | P (F )g =
P
f | n P (F n )g , due to the countable additivity property of f ,g . We are
P
P
tempted to conclude that P (F ) = n P (F n ). Yet, it may happen that n P (F n )
does not make any sense because weak convergence does not imply conP
P
vergence in the operator norm. However, n kP (F n )g k2 = n g | P (F n )g =
g | P (F )g = kP (F )g k2 . It follows that the sequence (P ( f n )g )n is summable.
P
If we write n P ( f n )g = T g , it defines a bounded operator T coinciding
with P (F ).
63
Notation 5.5.7. Let (X, F ) be a measurable space and F : X C. We denote by

kF k sup{|F (x)| : x X}, and B(X) = {F : X C | measurable, kF k < }.
Henceforth, the Hilbert space H will be fixed and B(H) (respectively P(H))
will denote as usual the set of bounded operators (respectively projections) on
H.
Theorem 5.5.8. Let (X, F ) be a measurable space and H a Hilbert space. If P is
a spectral measure on (X, F ) and F B(X), then there exists a unique operator
TF B(H) such that
Z
f | TF g = F (x) f | P (d x)g ,
X
for all f , g H. We write TF =
X F (x)P (d x).
Proof: The boundedness of F implies that the right hand Rside of the integral
gives rise to a well-defined sesquilinear
R functional ( f , 2g ) = X F (x)2 f | P (d x)g ,
for f , g H. Moreover, |( f , f )| X |F (x)|kP (d x) f k kF kk f k , hence the
functional is bounded. Existence and uniqueness of TF follows from the
Riesz-Frchet theorem.
Theorem 5.5.9 (Spectral decomposition theorem). If T Bh (H) then there exists a spectral measure on (C, B(C)), supported by spec(T ) R, such that
Z
T=
P ().
spec(T )
Proof: Let p R[t ] and f , g H be two arbitrary vectors. Denote by L f ,g (p) =

f | p(T )g . Then |L f ,g (p)| kp(T )kk f kkg k and since p(T ) B(H) we have
also kp(T )k = sup{|p()| : spec(T )} (exercise!). Since spec(T ) is a bounded
set, kp(T )k < for all p R[t ]. Hence the linear functional L f ,g is a bounded
linear functional on R[t ]. By Riesz-Frchet theorem, there exists consequently
a unique complex measure f ,g , supported by spec(T ), such that
Z
L f ,g (p) f | p(T )g =
spec(T )
p() f ,g (d ),
for all p R[t ], verifying | f ,g (B )| k f kkg k, for all B B(C). Using the uniqueness of f ,g , it is immediate to show that for every B B(C), S B ( f , g ) = f ,g (B )
is a sesquilinear form. Now, |S B ( f , g )| = | f ,g (B )| k f kkg k, for all B . Hence the
64
sesquilinear form is bounded; therefore, there exists an operator P (B ) Bh (H)

such that S B ( f , g ) = f | P (B )g for all f , g H. Recall that neither f ,g , nor
S B , Rnor P depend on the initially chosen polynomial p. Choosing p 0 () = 1, we
get Rspec(T ) f | P (d )g = f | P (spec(T ))g = f | g and choosing p 1 () = , we
get spec(T ) f | P (d )g = f | T g , for all f , g H. To complete the proof, it remains to show that P is a projection-valued measure. It is enough to show the
multiplicativity property. For any fixed pair f , g H and Rany fixed real polynomial q, introduce the auxiliary complex measure (B ) = B q() f | P (d )g ,
with B B(C). For every real polynomial p, we have
Z
p()(d ) =
p()q() f | P (d )g
= f | P (p(T )q(T )g
= q(T ) f | p(T )g
Z
=
p() q(T ) f | P (d )g .
Therefore,
(B ) =
q()1 B () f | P (d )g
= q(T ) f | P (B )g
= f | q(T )P (B )g
Z
=
q() f | P (d )P (B )g .
Since q is arbitrary,
Z
f | P (B C )g =
f | P (d )P (B )g
= f | P (B )P (C )g ,
and since f , g H are arbitrary, we get P (B C ) = P (B )P (C ).
Theorem 5.5.10. If T is a normal operator in B(H), then there exists a necessarily unique complex spectral measure on (C, B(C)), supported by spec(T ), such
that
Z
T=
spec(T )
P (d ).
Proof: Exercise! (Hint: T = T1 + i T2 with T1 , T2 Bh (H).)

65
5.6. Some basic notions on unbounded operators
5.6 Some basic notions on unbounded operators

The operators arising in quantum mechanics are very often unbounded.
Definition 5.6.1. Let H be a Hilbert space. An operator on H, possibly unbounded, is a pair (Dom(T ), T ) where Dom(T ) H is a linear manifold and
T : Dom(T ) H is a linear map. The set of operators on H is denoted L(H).
The graph of an operator T L(H) is the linear sub-manifold of H H of the
form
(T ) = {(( f , T f ) H H : f Dom(T )}.
The operator T is closed if (T ) is closed. The operator T is closable if there
exists T L(H) such that (T ) = (T ) in H H. Such an operator is unique and
is called the closure of T . An operator T is said densely defined if Dom(T ) = H.
If T1 , T2 L(H) with Dom(T1 ) Dom(T2 ) and T1 f = T2 f for all f Dom(T1 ),
then T2 is called an extension of T1 and T1 the restriction of T2 on Dom(T1 ); we
write T1 T2 . If T is bounded on its domain and Dom(T ) = H, then T can be
extended by continuity on the whole space.
The definitions of null space and range are also modified for unbounded
operators:
ker(T ) = { f Dom(T ) : T f = 0}
Ran(T ) = {T f H : f Dom(T )}.

The operator T is invertible if ker(T ) = {0} and its inverse, T 1 is the operator
defined on Dom(T 1 ) = Ran(T ) by T 1 (T f ) = f for all f Dom(T ).
If T1 , T2 L(H), then T1 +T2 is defined on Dom(T1 +T2 ) = Dom(T1 ) Dom(T2 )
by (T1 +T2 ) f = T1 f +T2 f . Similarly, the product T1 T2 is defined on Dom(T1 T2 ) =
{ f Dom(T2 ) : T2 Dom(T1 )} by (T1 T2 ) f = T2 (T1 f ).
Definition 5.6.2. Suppose that T is densely defined. Then T is the adjoint
operator with Dom(T ) = {g H : sup | g | T f | < , f Dom(T ), k f k = 1};
since Dom(T ) = H, by Riesz theorem, there exists a unique g H such that
g | f = g | T f for all f Dom(T ). We define then T g = g .
Example 5.6.3. (The position operator) Let (, F , ) be any separable, -finite
measure space, H = L 2 (, F , ; C), and f R H measurable. Let T L(H) be the
operator defined by Dom(T ) = {g H : (1 + | f |2 )|g |2 d < } and T g () =
f ()g () for g Dom(T ) and . Then T is closed, densely defined, with
66
Dom(T ) = Dom(T ) and T g () = f ()g (). When = R, F = B(R), and is

the Lebesque measure, we say that T is the position operator; it is obviously
self-adjoint.
Example 5.6.4. (The momentum operator) Let H = L 2 (R). A function u : R R
is called absolutely continuous, (a.c.) if there exists a function v : R R such
that
Z
b
u(b) u(a) =
v(x)d x, for all a < b.
In such a case, we write u 0 = v, u 0 is called the derivative of u. The function v is

determined almost everywhere. Define now T L(H) on
Z
Dom(T ) = { f H : f a.c., (| f |2 + | f 0 |2 )d x < }
by T f = f 0 . Then T is a closed, densely defined operator with T = i T . The
operator i T is called the momentum operator.
Exercise 5.6.5. Let q be the position operator, p the momentum operator.
Show that [q, p] i 1 .
Exercise 5.6.6. (Heisenbergs uncertainty principle) Denote by S(R) the Schwartz
space of indefinitely differentiable functions
of rapid decrease. If f S(R), deR
note by f its Fourier transform f () = R f (x) exp(i x)d x. Let p : S(R) S(R)
be defined by p f = i f 0 and q : S(R) S(R) by q f (x) = x f (x), for all x R.
Show that [q, p] = i 1 . If | denotes the L 2 scalar product on S(R), show that
| f | f | 2kp f k2 kq f k2 .
Conclude that for any f S(R),
k f k2 4kx f kL 2 (R) k fkL 2 (R ) .
Below are depicted the graphs of pairs | f (x)|2 and | f()|2 , chosen among a
class of Gaussian functions, for different values of some parameter. How do you
interpret these results?
67
5.6. Some basic notions on unbounded operators
0.8
1.5
0.6
0.4
0.5
0.2
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
1 6
xi
0.5
0.4
0.3
0.2
0.1
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
1 10
10
xi
16
0.12
14
0.1
12
10
0.08
8
0.06
0.04
2
0.02
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
1 40
30
20
10
10
20
30
40
xi
100
0.02
80
0.016
0.018
0.014
60
0.012
0.01
40
0.008
0.006
20
0.004
0.002
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
1 300
200
100
100
200
300
xi
68
6
Propositional calculus and quantum
formalism based on quantum logic
Phenomenology is an essential step in constructing physical theories. Phenomenological results are of the following type: if a physical system is subject
to conditions A, B,C , . . ., then the effects X , Y , Z , . . . are observed. We further introduced yes-no experiments consisting in measuring questions in given states.
However, there may exist questions that depend on other questions and hold
independently of the state in which they are measured. More precisely, suppose for instance that Q A denotes the question: does the physical particle lie
in A, for some A B(R3 )? Let now B A be another Borel set in R3 . Whenever Q A is true (i.e. for every state for which Q A is true) Q B is necessarily true.
This remark defines a natural order relation in the set of questions. Considering
questions on given physical system more abstractly, as a logical propositions,
it is interesting to study first the abstract properties of a partially ordered set of
propositions. This abstract setting allows the statement of the basic axioms for
classical or quantum systems on an equal footing.
69
6.1. Lattice of propositions
6.1 Lattice of propositions

Let be a set of propositions and for any two propositions a and b, denote by
a b the implication whenever a is true, it follows that b is true
Definition 6.1.1. The pair (, ) is a partially ordered set (poset) if the relation
is a partial order (i.e. a reflexive, transitive, and antisymmetric binary operation). For a, b , we say that u is a least upper bound if
1. a u and b u,
2. if a v and b v for some v , then u v.
If a least upper bound of two elements a and b exists, then it is unique and
denoted by sup(a, b) ,
Definition 6.1.2. A lattice is a set with two binary operations, denoted respectively by (join) and (meet), and two constants 0 and 1 , satisfying, for all a, b, c the following properties:
1. idempotence: a a = a = a a,
2. commutativity: a b = b a and a b = v a,
3. associativity: a (b c) = (a b) c and a (b c) = (a b) c,
4. identity: a 1 = a and a 0 = a,
5. absorption: a (a b) = a = a (a b).
Theorem 6.1.3. Let (, ) be a poset. Suppose that
1. has a least element 0 and a greatest element 1, i.e. for all a , we have
0 a 1,
2. any two elements a, b have a least upper bound in , denoted by a b,
and a greatest lower bound in , denoted by a b. Then (, , , 0, 1) is a
lattice.
Conversely, if (, , , 0, 1) is a lattice, then, on defining a b whenever a b = b,
the pair (, ) is a poset verifying properties 1 and 2 of definition 6.1.1
/Users/dp/a/ens/mq/iq-proca.tex
70
lud on 17 February 2013
{1, 2, 3}
{1, 2}
{2, 3}
{1, 3}
{1}
{3}
{2}
{}
Figure 6.1: The Hasse diagram of the lattice of subsets of the set {1, 2, 3}.
Proof: : Exercise!
Definition 6.1.4. A lattice (, , , 0, 1) is called distributive if it verifies, for all

a, b, c ,
a (b c) = (a b) (a c),
and
a (b c) = (a b) (a c).
Remark 6.1.5. A finite lattice (or finite poset) can be represented by its Hasse
diagram in the plane. The points of the lattice are represented by points in
the plane arranged so that if a b then the representative of b lies higher in
the plane than the representative of a. We join the representatives of a and b
by a segment when b covers a, i.e. when a b but there is no c such that
a < c < b.
Example 6.1.6. Let S be a finite set and P (S) the collection of its subsets. Then
(P (S), ) is a poset, equivalent to the lattice (P (S), , , ;, S), called the lattice
of subsets of S. This lattice is distributive. For the particular choice S = {1, 2, 3}
its Hasse diagram is depicted in figure 6.1.
Exercise 6.1.7. Let V = R2 (viewed as a R-vector space) and E 1 , E 2 , E 3 be three
distinct one-dimensional subspaces of V. Denote by the order relation be a
vector subspace of. Show that there is a finite set S of vector subspaces of V
containing E 1 , E 2 , and E 3 such that (S, ) is a lattice. Is this lattice distributive?
71
6.1. Lattice of propositions

In any lattice , a complement of a is an element a 0 such that aa 0 =
0 and a a 0 = 1. Complements may fail to exist and they may be not unique.
However, in a distributive lattice, any element has at most one complement.
Definition 6.1.8. A Boolean algebra is a complemented distributive lattice (i.e.
a distributive lattice in which any element has a necessarily unique complement.)
When the lattice is infinite, one can consider infinite subsets F . When
both aF a and aF a exist (in ) for any countable subset F , the lattice is
called -complete. A Boolean -algebra is a Boolean algebra that is -complete.
Definition 6.1.9. A lattice is called modular of it satisfies the modularity
condition:
a c b , a (b c) = (a b) c.
If is a modular and complemented lattice then satisfies, for every complement a 0 of a, the modularity condition,
a b b = a (a 0 b).
If the complement of a is an orthocomplement, then the complemented modular lattice is called orthomodular.
Example 6.1.10. The Dilworth lattice, whose Hasse diagram is depicted in figure 6.2, is a complemented modular but not distributive.
Exercise 6.1.11. Show that a Boolean algebra is always modular.
Definition 6.1.12. An atom in a lattice is a minimal non-zero element, i.e. a
is an atom if a 6= 0 and if x < a for some x then x = 0. A lattice is atomic if
every point is the join of a finite number of atoms.
Definition 6.1.13. A homomorphism from a complemented lattice 1 into a
complemented lattice 2 is a map h : 1 2 such that
1. h(01 ) = 02 and h(11 ) = 12 ,
2. h(a 0 ) = h(a)0 for all a 1 ,
3. h(a b) = h(a) h(b) and h(a b) = h(a) h(b), for all a, b 1
An isomorphism is a lattice homomorphism that is bijective. If the condition 3
above holds also for countable joins and meets, h is called a -homomorphism.
If 1 = 2 a lattice isomorphism is called lattice automorphism.
72
Figure 6.2: The Hasse diagram of the Dilworth lattice.

Theorem 6.1.14. Let be a Boolean -algebra. Then there exist an abstract set
X, a -algebra, X , of subsets of X and a -homomorphism h : X .
Proof: It is first given in [?] and later reproduced in [33].
This theorem serves to extend the notion of measurability, defined for maps
between measurable spaces, to maps defined on abstract Boolean -algebras.
Recall that if X is an arbitrary set of points equipped with a Boolean -algebra of
subsets X , and Y a complete separable metric space equipped with its Borel algebra B(Y), a map f : X Y is called measurable if for all B B(Y), f 1 (B )
X.
Definition 6.1.15. Let be an abstract Boolean -algebra and (Y, B(Y)) a
complete separable metric space equipped with its Borel -algebra. A Y-valued
classical observable associated with is a -homomorphism h : B(Y) . If
Y = R, the observable is called real-valued.
The careful reader will have certainly remarked that the previous definition
is compatible with axiom 2.2.15. As a matter of fact, with every real random
variable X on an abstract measurable space (, F ) is associated a family of
propositions Q BX = 1 {X B } , for B B(R). The aforementioned -homomorphism
h : B(R) F , stemming from X () through the spectral measure K X (, B ), is
73
6.2. Classical, fuzzy, and quantum logics; observables and states on logics
given by
h(B ) = { : Q BX () = 1} = X 1 (B ) F .
Notice that this does not hold for quantum systems where some more general
notion is needed.
6.2 Classical, fuzzy, and quantum logics; observables

and states on logics
6.2.1 Logics
Definition 6.2.1. Let (, ) be a poset (hence a lattice). By an orthocomplementation on is meant a mapping : 3 a 7 a , satisfying for a, b :
1. is injective,
2. a b b a ,
3. (a ) = a,
4. a a = 0.
A lattice with an orthocomplementation operation is called orthocomplemented.
We remark that from condition 2 it follows that 0 = 1 and 1 = 0. From
condition 3 it follows that is also surjective. Finally, conditions 1, 2, and 3
imply that a a = 1.
Definition 6.2.2. An orthocomplemented lattice, , is said to be a logic if
1. for any countable sequence (a n )nN of elements of , both nN a n and
nN exist in ,
2. if a 1 , a 2 and a 1 a 2 , then there exists b , such that b a 1 and
b a1 = a2 .
Without loss of generality, we can always assume that an orthocomplemented
lattice verifies orthomodularity for a = a 0 . Remark also that the element whose
74
existence is postulated in item 2 of the previous definition is unique and equal

in fact to a 1 a 2 . In fact, if b a 1 is such that b a 1 = a 2 , then necessarily,
a 1 b and b a 1 = a 2 . Using orthomodularity, a 1 (b a 1 ) = b and substituting the left hand side parenthesis by a 2 , we get the dual of the required
equality. Dualising, we conclude.
The element a is called the orthogonal complement of a in . If a b
and b a , then a and b are said orthogonal and we write a b.
Exercise 6.2.3. Assume that (, ) is a poset (hence a lattice) that is orthocomplemented. Let a, b be such that a < b. Denote by
[a, b] = {c : a c b}.
Show that
1. [0, b] becomes a lattice in which countable joins and meets exist and
whose zero element is 0 and unit element is b,
2. if we define, for x [0, b], x 0 = x b, then the operation 0 : [0, b]
[0, b] is an orthocomplementation,
3. conclude that [0, b] is a logic.
Example 6.2.4. Any Boolean -algebra is a logic provided we define, for any
element a, its orthocomplement to be its complement a 0 . Boolean -algebras
are called classical logics.
Example 6.2.5. Let H be a C-Hilbert space. Let be the collection of all Hilbert
subspaces of H. If is meant to denote be a Hilbert subspace of and the orthogonal complementation in the Hilbert space sense, then is a logic, called
standard quantum logic.
Postulate 6.2.6. In any physical system (classical or quantum), the set of all
experimentally verifiable propositions is a logic (classical or standard quantum).
6.2.2 Observables associated with a logic

Suppose that is the logic of verifiable propositions of a physical system and
let X be any real physical quantity relative to this system. Denoting x(B ) the
proposition the numerical results of the observation of X lie in B , it is natural
and harmless to consider that B B(R); obviously then, x is a mapping x :
75
B(R) . We regard to physical quantities X and X 0 as identical whenever
the corresponding maps x, x 0 : B(R) are the same. If f : R R is a Borel
function, we mean by X 0 = f X a physical quantity taking value f (r ) whenever
X takes value r . The corresponding map is given by B(R) 3 B : x 0 7 x 0 (B ) =
x( f 1 (B )) . Hence we are led naturally to the following
Definition 6.2.7. Let be a logic. A real observable associated with is a mapping x : B(R) verifying:
1. x(;) = 0 and x(R) = 1,
2. if B 1 , B 2 B(R) with B 1 B 2 = ; then x(B 1 ) x(B 2 ),
3. if (B n )nN is a sequence of mutually disjoint Borel sets, then x(nN B n ) =
nN x(B n ).
We write O () for the set of all real observables associated with .
Exercise 6.2.8. Let be a logic and x O (). Show that for any sequence of
Borel sets (B n )nN we have
x(nN B n ) = nN x(B n )
and
x(nN B n ) = nN x(B n ).
Definition 6.2.9. Let be a logic and O () the set of its associated observables.
A real number is called a strict value of an observable x O (), if x({}) 6= 0.
The observable x O () is called discrete if there exists a countable set C =
{c 1 , c 2 , . . .} such that x(C ) = 1; it is called constant if there exists c R such that
x({c}) = 1. It is called bounded if there exists a compact Borel set K such that
x(K ) = 1.
Definition 6.2.10. We call spectrum of x O () the closed set defined by
spec(x) = C closed :x(C )=1C .

The numbers spec(x) are called spectral values of x.
Any strict value is a spectral value; the converse is not necessarily true.
Exercise 6.2.11. Show that spec(x) if and only if any open set U containing
verifies x(U ) 6= 0.
76
If (a n )nN is a partition of unity, i.e. a family of mutually orthogonal propositions in such that nN a n = 1, there exists a unique discrete observable
admitting as spectral values a given discrete subset {c 1 , c 2 , . . .} of the reals. In
fact, it is enough to define for all n N, x({c n }) = a n and for any B B(R),
x(B ) = n:cn B a n . Notice however that discrete observables do not exhaust
all the physics of quantum mechanics; important physical phenomena involve
continuous observables.
6.2.3 States on a logic

We have seen that to every classical system is attached a measurable space
(, F ) (its phase space); observables are random variables and states are probability measures that may degenerate to Dirac masses on particular points of
the phase space. This description is incompatible with the experimental observation for quantum systems. For the latter, the Heisenbergs uncertainty principle stipulates that no matter how carefully the system is prepared, there always
exist observables whose values are distributed according to some non-trivial
probability distribution.
Definition 6.2.12. Let be a logic and O () its set of associated observables.
A state function is a mapping : O () 3 x 7 x M1+ (R, B(R)).
For every Borel function f : R R, for every observable x, and every Borel
set B on the line, we have:
f x (B ) = x ( f 1 (B )).
Denoting by o the zero observable and 0 the zero of R, we have that o = 0 .
In fact, suppose that f : R R is the identically zero map. Then f o = o and
R if 0 B
f 1 (B ) =
; otherwise.
Hence, if 0 B , then o (B ) = f o (B ) = o ( f 1 (B )) = 1, because o is a probability on R; if 0 6 B then similarly o (B ) = 0. Therefore, in all circumstances,
o (B ) = 0 (B ).
If x O () is any observable and B B(R) is such that x(B ) = 0 , then
x (B ) = 0. In fact, for this B , we have 1 B x = o and x (B ) = o ({1}) = 0 ({1}) = 0.
This implies that if x is discrete, the measure x is supported by the set of the
strict values of x.
77
Definition 6.2.13. An observable q O () is a question if q({0, 1}) = 1. A question is the necessarily discrete. If q({1}) = a , then q is the only question such
that q({1}) = a; we call it question associated with the proposition a and denote
by q a if necessary.
Definition 6.2.14. Let be a logic. A function p : [0, 1] satisfying
1. p(0) = 0 and p(1) = 1,
2. if (a n )nN is a sequence of mutually orthogonal propositions of , and
P
a = nN a n , then p(a) = nN p(a n )
is called state (or probability measure) on the logic . The set of states on is
denoted by S ().
The concept of probability measure on a logic coincides with a classical
probability measure when the logic is a Boolean -algebra. For non distributive logics however, the associated probability measures are genuine generalisations of the classical probabilities. For standard quantum logics, the associated states are called quantum probabilities.
Theorem 6.2.15. Let p S (), where is a logic.
1. On defining a map p : O () M1+ (R, B(R)), by the formula: for every x
p
O () and for every B B(R), x (B ) = p(x(B )), then p is a state function.
2. Conversely, if is an arbitrary state function, then for every x O (), then
there exists a unique probability measure p S () such that for every x
O () and for every B B(R), x (B ) = p(x(B )).
Proof:
p
1. The map x : B(R) [0, 1] is certainly a -additive, non-negative map.

p
Moreover, x (R) = p(1) = 1, hence it is a probability. If f : R R is a Borel
function,
p
f x (B ) = p( f x(B )) = p(x( f 1 (B ))) = x (( f 1 (B )).

Hence p is a state function.
78
2. Let be a state function. If a and q a O () the question associated

with proposition a, then q a is a probability measure on B(R). Since q a is
a question, q a ({0, 1}) = 1. Define p(a) = q a ({1}). Obviously, for all a ,
p(a) is well defined and is taking values in [0, 1]. It remains to show that
p is a probability measure on , that is to say verify -additivity and normalisation. For 0 , q 0 ({1}) = 0. Hence q ({1}) = 0 = p(0). Similarly,
0
we show that = p(1) = 1. This shows normalisation.
Let (a n )nN be a sequence of mutually orthogonal elements of , and denote by a = nN a n . Let x O () be the discrete observable defined by
x({0}) = a and x({n}) = a n , for n = 1, 2, . . .. Then, 1 {n} x({1}) = x({n}) =
a n . Hence q an = 1 {n} x and p(a n ) = x ({n}). Since x is a probabilP
ity measure, n p(a n ) = x ({1, 2, 3, . . .}) = x (N). Similarly, 1 N x = q a
because 1 N x({1}) = x(N) = nN x({n}) = nN a n = a. Hence, finally,
P
p(a) = n p(a n ) establishing thus -additivity of p. Finally, for x O ()
and B B(R),
x (B ) = 1 B x ({1}) = q x(B ) ({1}) = p(x(B )).
If p S () and x O (), the map B(R) 3 B 7 p(x(B )) [0, 1] defines a

probability measure on B(R). It is called the probability distribution induced
on the space of its values by the observable x when the system is in state p and
p
is denoted x . The expected value of x in state p is
Z
p
Ep (x) = t x (d t )
R
and for a Borel function f : R R, we have

Z
p
Ep ( f x) = f (t ) x (d t )
R
(provided the above integrals exist.) If Ep (x 2 ) < , the variance of x in p is

Varp (x) = Ep (x 2 ) (Ep (x))2 .
Postulate 6.2.16. The phase space of a physical system described by the logic .
States of the system are given by S ().
Postulate 6.2.17. Observables of a physical system described by the logic are
O ().
Postulate 6.2.18. Measuring whether the values of a physical observable x
O () lie in B B(R) when the system is prepared in state p S () means deterp
mining x (B ).
79
6.3. Pure states, superposition principle, convex decomposition
6.3 Pure states, superposition principle, convex decomposition

Proposition 6.3.1. Let S () be the set of states on the logic . Let (p n )nN be
P
a sequence in S () and (c n )nN a sequence in R+ such that nN c n = 1. Then
P
P
p = nN c n p n , defined by p(a) = nN c n p n (a) for all a , is a state.
Proof: Exercise!
Corollary 6.3.2. For any logic , the set S () is convex.

P
Remark 6.3.3. Notice that if p = nN c n p n as above, for every x O (), we
P
p
p
have that x = nN c n x n . In fact, for all B B(R),
p
x (B ) = p(x(B )) =
X
nN
c n p n (x(B )) =
X
nN
c n x n (B ).
This decomposition has the following interpretation: the sequence (c n )nN defines a classical probability on N meaning that in the sum defining p, each p n is
chosen with probability c n . Therefore, for each integrable observable x O (),
P
the expectation Ep (x) = nN c n Ep n (x) consists in two averages: a classical average on the choice of p n and a (may be) quantum average Ep n (x).
Exercise 6.3.4. Give a plausible definition of the notion of integrable observable used in the previous remark and then prove the claimed equality: Ep (x) =
P
nN c n Ep n (x)
Definition 6.3.5. A state p S () is said to be pure if the equation p = cp 1 +
(1c)p 2 , for p 1 , p 2 S () and c [0, 1] implies p = p 1 = p 2 . We write S p () for
the set of pure states of . Obviously S p () = Extr S ().
Definition 6.3.6. Let D S () and p 0 S (). We say that p 0 is a superposition of states in D if for a ,
p D, p(a) = 0 p 0 (a) = 0.
P
It is an exercise to show that the state p = nN c n p n defined in the proposition ?? is a superposition of states in D = {p 1 , p 2 , . . .}. In the case is a Boolean
-algebra, the next theorem 6.3.7 shows that this is in fact the only kind of possible superposition. This implies, in particular, the unicity of the decomposition of a classical state into extremal (pure) states. If is a standard quantum
logic, unicity of the decomposition does not hold any longer!
80
Theorem 6.3.7. Let be a Boolean -algebra of subsets of a space X. Suppose

that
1. is separable1 ,
2. for all a X, {a} .
For any a X and any A X, let a be the state defined by
a (A) =
1 if a A
0 otherwise.
Then, (a )aX is precisely the set of all pure states in . If D S p () and p 0

S p (), then p 0 is a superposition of states in D if and only if p 0 D.
Proof: Denote {A 1 , A 2 , . . .} a denumerable collection of subsets of X generating
. Purity of a is trivially verified. Suppose that p is a pure state. If for some
A 0 we have 0 < p 0 (A) < 1, then, on putting for A
p 1 (A) =
and
p 2 (A) =
1
p(A A 0 )
p(A 0 )
1
p(A A c0 ),
1 p(A 0 )
()
()
we get p(A) = p(A 0 )p 1 (A) + (1 p(A 0 ))p 2 (A). Yet, applying (*) and (**) to A 0 ,
we get p 1 (A 0 ) = 1 and p 2 (A 0 ) = 0, hence p 1 6= p 2 . This is in contradiction with
the assumed purity of p. Therefore, we conclude that for all A , we have
p(A) {0, 1}. Replacing A n by A cn if necessary, we can assume without loss of
generality that p(A n ) = 1 for all the sets of the collection generating . Let B =
n A n . Then p(B ) = 1 and consequently B cannot be empty. Now B cannot
contain more than one point either. In fact, the collection of all sets C such
that either B C or B C = ; is a -algebra containing all the sets A n , n N.
Hence, it coincides with . As singletons are members of , the set B must be
a singleton, i.e. B = {a} for some a X. Put then p = a . Finally, let p 0 be a
superposition of states in D (all its elements are pure states). If p 0 = a0 but
p 0 6 D, then p({a 0 }) = 0 for all p D but p 0 ({a 0 }) 6= 0, a contradiction.
i.e. there is a countable collection of subsets A n X, n N, generating by complementation, intersections, and unions.
81
6.4. Simultaneous observability
6.4 Simultaneous observability

In quantum systems, the Heisenbergs uncertainty principle, already shown in
chapter 2, there are observables that cannot be simultaneously observed with
arbitrary precision.
Definition 6.4.1. Let a, b . Propositions a and b are said to be simultaneously verifiable, denoted by a b, if there exists elements a 1 , b 1 , c such that
1. a 1 , b 1 , c are mutually orthogonal and,
2. a = a 1 c and b = b 1 c hold.
Observables x, y O () are simultaneously observable if for all B B(R), x(B )
y(B ). For A, B , we write A B if for all a A and all b B we have a b.
Lemma 6.4.2. Let a, b . The following are equivalent:
1. a b,
2. a (a b) b,
3. b (a b) a,
4. there exist x O () and A, B B(R) such that x(A) = a and x(B ) = b,
5. there exists a Boolean sub-algebra of containing a and b.
Proof:
1 2:
a b a = a 1 c and b = b 1 c
c a and c b
c a b.
From the definition 6.2.2 (logic), it follows that there exists d such that
c d and c d = a b.
Now d c d = a b a and d c (since d d .) Hence, d a c = a 1
(see remark immediately following the definition 6.2.2.) Similarly, d
b 1 d b 1 q 1 = 0. Therefore d = 0 and consequently c = a b. It
follows a 1 = a (a b) . Yet, a 1 c and a 1 b 1 so that a 1 (b 1 c) = b.
Summarising, a (a b) b.
82
1 3: By symmetry.
2 1: Since a (a b) b, on writing a 1 = a (a b) , b 1 = b (a b) , and
c = a b, we find a = a 1 c and b = b 1 c. Since a 1 b, it follows that a 1
b 1 and a 1 c, while, by definition, c b 1 which proves the implication.
Henceforth, the equivalence 1 2 3 is established.
1 4: If a = a 1 c, b = b 1 c and a 1 , b 1 , c mutually orthogonal, write d = a 1
b 1 c and define x to be the discrete observable such that x({0}) = a 1 ,
x({1}) = b 1 , x({2}) = c, and x({3}) = d . Then x({0, 2}) = a and x({1, 2}) = b.
4 5: x(A (A B )c ) = a (a b) and x(B (A B )c ) = b (a b) . On writing
a 1 = a (a b) , a 2 = a b, a 3 = b (a b) , and a 4 = (a b) , we see
that (a i )i =1,...,4 are mutually orthogonal and a 1 a 2 a 3 a 4 = 1. If
A = {a i 1 . . . a i k : k 4; 1 i 1 . . . i k 4},
it is easily verified that A is Boolean sub-algebra of . Since a, b A , this
proves the implication.
5 2: Let A be a Boolean sub-algebra of containing a and b. Now, [a (a
b) ] b = 0. As a, b, a (a b) , b A , it follows that
a (a b) = [(a (a b) ) b]
[(a (a b) ) b ]
= [(a (a b) ) b ]
b.
Therefore a (a b) b.
The significance of this lemma is that if two propositions are simultaneously

verifiable, we can operate on them as if they were classical.
Theorem 6.4.3. Let be any logic and (x )D a family of observables. Suppose that x x 0 for all , 0 D. Then there exist a space X, a -algebra X
of subsets of X, a family of measurable functions g : X R, D, and a homomorphism : X such that (g l1 (B )) = x (B ) for all D and all
b B(R). Suppose further that either is separable or D is countable. Then, for
all D, there exist a x O () and a measurable function f : R R such that
x = f x.
83
6.5. Automorphisms and symmetries

The proof of this theorem is omitted. Notice that it allows to construct functions of several observables that are simultaneously observable. This latter result is also stated without proof.
Theorem 6.4.4. Let be any logic and (x 1 , . . . , x n ) a family of observables that
are simultaneously observable. Then there exists a -homomorphism : B(Rn )
such that for all B B(R) and all i = 1, . . . , n,
x i (B ) = (1
i (B )),
()
where i : Rn R is the projection (t 1 , . . . , t n ) = t i , i = 1, . . . , n. If g is a Borel

function on Rn , then g (x 1 , . . . , x n )(B ) = (g 1 (B )) is an observable. If g 1 , . . . , g k
are real valued Borel functions on Rn and y i = g i (x 1 , . . . , x n ), then y 1 , . . . , y k are
simultaneously observable and for any real valued Borel function h on Rk , we
have h(y 1 , . . . , y k ) = h(g 1 , . . . , g k )(x 1 , . . . , x n ) where, for t = (t 1 , . . . , t n ), h(g 1 , . . . , g k )(t) =
h(g 1 (t), . . . , g k (t)).
An immediate consequence of this theorem is that if p is a probability meap
sure on , then x1 ,...,xn (B ) = p((B )), for B B(Rn ), is the joint probability
distribution of (x 1 , . . . , x n ) in state p.
6.5 Automorphisms and symmetries

Let be a logic. The set Aut(), of automorphisms of , acquires as usual a
group structure; they induce naturally automorphisms on S (), called convex
automorphisms.
Let, in fact, Aut() and p S (). If we define to be the induced
action of on p, by (p)(a)
= p(1 (a)), for all a , then is a convex automorphism of S ().
Definition 6.5.1. A map : S () S () is a convex automorphisms if
1. is bijective and
2. if (c n )nN is a sequence of non-negative reals such that
(p n )nN is a sequence of states in S (), then
(
X
nN
cn p n ) =
84
X
nN
nN c n
= 1 and
c n (p n ).
The set of convex automorphisms of S () is denoted Aut(S ()).

Lemma 6.5.2. Let Aut(). Then the induced automorphism on S () is
convex.
P
Proof: Bijectivity of follows immediately from the bijectivity of . If p = nN c n p n
P
S () (with the notation of definition 6.5.1), then (p)(a)

= p(1 (a)) = nN c n p n (1 (a)) =
P
n )(a) for all a .
nN c n (p
Remark 6.5.3. It is obvious that convex automorphisms map pure states of

S p () into pure states.
Dynamics, i.e. time evolution of a system described by a logic can be
defined in the following manner. For each t R, there exists a unique map
D(t ) : S () S () having the following interpretation: if p S () is the state
of the system at time t 0 , then D(t )(p) will represent the state of the system at
time t + t 0 .
Definition 6.5.4. Let G be a locally compact topological group. By a representation of G into Aut(S ()), we mean a map : G Aut(S ()) such that
1. (g 1 g 2 ) = (g 1 )(g 2 ) for all g 1 , g 2 G,
2. for each a and each p S (), the mapping g 7 (g )(p)(a) is B(G)measurable.
Postulate 6.5.5. Time evolution of an isolated physical system described by a
logic , is implemented by a map R 3 t 7 D(t ) Aut(S ()). This map provides
a representation of the Abelian group (R, +) into Aut(S ()). More generally, any
physical symmetry, implemented by the action of a locally compact topological
group G, induces a representation into Aut(S ()).
P
Here is an interpretation and/or justification of this axiom. If p = nN c n p n
represents the initial state of the system, we can realise this state as follows.
First chose an integer n N with probability c n and prepare the system at state
p n . Let the system evolve under the dynamics. Then at time t it will be at state
p n0 = D(t )(p n ) with probability c n . Assuming now that D(t ) is a convex autoP
morphism means that D(t )(p) = nN c n D(t )(p n ), i.e. at time t , the system is in
state p n0 = D(t )(p n ) with probability c n , exactly the result we obtained with the
first procedure.
To further exploit the notions of logic, states, observables, and convex automorphisms, we must specialise the physical system.
85
6.5. Automorphisms and symmetries
86
7
Standard quantum logics
We recall that a standard quantum logic was defined in chapter 5 to be the set
of Hilbert subspaces of C-Hilbert space H. For every Hilbert subspace M ,
we denote by P M the orthogonal projection to M . If x O (), then B 7 P x(B ) ,
for B B(R), is a projection-valued measure on B(R). Conversely, for every
projection-valued measure P on B(R), there exists an observable x O () such
that P (B ) = P x(B ) , for all B B(R). We identify henceforth Hilbert subspaces
with the orthogonal projectors mapping the whole space on them (recall exercise 4.4.7.)
7.1 Observables
Lemma 7.1.1. Let M 1 , M 2 . Then propositions associated with M 1 and M 2
are simultaneously verifiable if and only if [P M1 , P M2 ] = 0.
Proof:
(): Propositions M 1 and M 2 are simultaneously verifiable if there exist
mutually orthogonal elements N1 , N2 , N such that M i = Ni N , for
87
7.2. States
i = 1, 2. Then P Mi = P Ni + P N and the commutativity of the projectors
follows immediately.
(): If [P M1 , P M2 ] = 0, let P = P M1 P M2 . Then P is a projection. Define
Q i = P Mi P , for i = 1, 2; it is easily verified that Q i are projections and
PQ i = Q i P = 0. Therefore Q 1Q 2 = Q 2Q 1 = 0. If we define Ni = Q i (H),
for i = 1, 2 and N = P (H), then N1 , N2 , N are mutually orthogonal and
M i = Ni N which proves that M 1 M 2 .
Theorem 7.1.2. Let be a standard logic with associated Hilbert space H. For
any x O (), denote X the self-adjoint (not necessarily bounded) operator on H
with spectral measure given by the mapping B(R) 3 B 7 P x(B ) . Then
1. the map x 7 X is a bijection between O () and self-adjoint operators on
H,
2. the observable x is bounded if and only if X Bh (H),
3. two bounded observables x 1 and x 2 are simultaneously observable if and
only if the corresponding bounded operators X 1 and X 2 commute,
4. if x is a bounded observable and Q R[t ], then the operator associated
with Q x is Q(X ),
5. more generally, if x 1 , . . . , x r are bounded observables any two of them being
simultaneously observable, and Q R[t 1 , . . . , t 2 ], then the observable Q
(x 1 , . . . , x r ) has associated operator Q(X 1 , . . . , X r ).
Proof: Assertions 14 are simple exercises based on the spectral theorem for
self-adjoint operators. Assertion 5 is a direct consequence of theorem 6.4.4.
7.2 States
In chapter 2, we defined (pure) quantum states to be unit vectors of H. In chapter 6, states have been defined as probability measures on a logic. We first show
that in fact rays correspond to states viewed as probability measures on .
/Users/dp/a/ens/mq/iq-stqlo.tex
88
Unit vectors of H are called rays. Let H, with kk = 1 be a ray and denote
by p : [0, 1] the map defined by
3 M 7 p (M ) = | P M = kP M k2 .
We have: p (1) p (H) = 1, p (0) p ({0}) = 0, and if (M n )nN is a sequence of
mutually orthogonal Hilbert subspaces of H and M = nN M n , then
X
X
p (M ) = kP M k2 =
| P Mn =
p (M n ).
nN
nN
Hence p S (). If c C, with |c| = 1, then p c = p .

Theorem 7.2.1. Let H be a Hilbert space, (n )nN an orthonormal basis in it an
T B+ (H). We define the trace of T by
X
tr(T ) =
n | T n [0, +].
nN
Then for all T, T1 , T2 B+ (H) the trace has the following properties
1. is independent of the chosen basis,
2. tr(T1 + T2 ) = tr(T1 ) + tr(T2 ),
3. tr(T ) = tr(T ) for all 0,
4. tr(U T U ) = tr(T ), for all U U(H).
Proof: (To be filled in a later version.)
Definition 7.2.2. Let T B(H). The operator T is called trace-class operator if

tr(|T |) < . The family of trace-class operators is denoted by T1 (H).
Lemma 7.2.3. The space T1 (H) is a two-sided ideal of B(H) and tr(T B ) = tr(BT )
for all B B(H).
Proof: This will be shown in several steps.
1. Every B B(H) can be decomposed as a linear combination of four unitary operators. In fact, writing B = 21 (B + B ) 2i [i (B B )], the operator B is decomposed into a sum of two self-adjoint operators. p
Now, if
A Bh (H), we can w.l.o.g. assume that kAk 1 and thence A I A 2
are unitary. We conclude that B = c 1U1 + . . . + c 4U4 with Ui U(H) and
c C.
89
7.2. States
2. We show then that T1 (H) is a vector space. In fact, for every C, due to
the fact that |A| = |||T |, it follows that if A T1 (H) then A T1 (H) as
well.
For T1 , T2 T1 (H), denote by U ,V,W the partial isometries arising into
the polar decompositions T1 +T2 = U |T1 +T2 |, T1 = C |T1 |, and T2 = W |T2 |.
Then,
X
X
e n |U (T1 + T2 )e n
e n | |T1 + T2 |e n =
n
n
X
X
e n |U V |T1 |e n + e n |U W |T2 |e n
=
n
n
X
X
| e n |U V |T1 |e n | + | e n |U W |T2 |e n |.
n
Now,
X
X
1
1
e n |U V |T1 |e n =
|T1 | 2 V U e n | |T1 | 2 e n
n
k|T1 | 2 V U e n kk|T1 | 2 e n k Cauchy-Scwharz on H
X
X
1
1
( k|T1 | 2 V U e n k2 )1/2 ( k|T1 | 2 e n k2 )1/2 Cauchy-Scwharz on `2 (N
n
We conclude that
X
X
1
1
1
k|T1 | 2 V U e n k2 =
|T1 | 2 V U e n | |T1 | 2 V U e n
n
n
X
=
e n |U V |T1 |V U e n
n
X
e n |V |T1 |V e n
n
X
e n | |T1 |e n
n
= tr(|T1 |),
because U ,V are partial isometries. The second term is majorised similarly so that tr(|T1 + T2 |) tr(|T1 |) + tr(|T2 |) < , showing that T1 + T2
T1 (H).
3. Using the decomposition of every B B(H) into the combination of four
P
P
unitary operators B = 4i =1 c i Ui , we get tr(T B ) = 4i =1 c i tr(T Ui ) so that
it becomes sufficient to prove that T T1 (p
H) and U U(H) implies that
p
p
1
U T = T T = |T | and |T U | = (T U ) T U =
T
U
,U
T
T
(
H
).
But
|U
T
|
=
(U
T
)
p
U |T |2U = U |T |U ; furthemore U |T |U 0. Hence tr(|T U |) = tr |T | =
tr(|U T |).
90

Exercise 7.2.4. Show the T T1 (H) implies that T T1 (H) (hence T 1 (H) is a
bilateral -ideal of T1 (H).
Exercise 7.2.5. Show that T T1 (H) is not necessarily closed with respect to
the operator norm stemming from the Hilbert norm. Nevertheless, T1 (H) is a
Banach space for the k k1 norm defined by kT k1 = tr |T |.
Definition 7.2.6. If D is a bounded, self-adjoint, non-negative, trace-class operator on H, then D is called a von Neumann operator. If further tr(D) = 1, then
D is said to be a density matrix (operator). The set of density matrices on H is
denoted by D(H).
The states p , for a ray of H, can also be described in another way. Let D
be the projection operator on the one-dimensional subspace1 C. Then D is
trace-class and for every X B(H), it follows that D X is also trace-class. Let
(n )nN be an arbitrary orthonormal basis of H; without loss of generality, we
can then assume that 1 = . We have
tr(D X ) = tr(X D )
=
X
nN
n | X D n
= | X
= E (X ).
In particular, if X = P M for M ,
p (M ) = | P M = tr(D P M ).
Lemma 7.2.7. Let (n )nN be an arbitrary sequence of rays in H and (c n )nN an
P
arbitrary sequence of non-negative reals such that nN c n = 1. Denote by D n the
projection operator on the one-dimensional subspace Cn , for n N. Then
D=
X
nN
cn D n
is a well defined density matrix.

Proof: Exercise.
Exercise 7.2.8. Show that D(H) is convex.

1
We recall that the term subspace always means closed subspace.
91
7.3. Symmetries
Lemma 7.2.9. Let D be a density matrix defined as in lemma 7.2.7 and p : R
the mapping defined by 3 M 7 p(M ) = tr(P M D). Then p S () and moreP
over it can be decomposed into p = nN c n p n .
Proof: First the superposition property follows from the linearity of the trace:
P
P
for all M , we have p(M ) = tr(P M D) = nN c n tr(P M D n ) = nN c n p n (M ). It
is now obvious that p is a state: in fact, p(0) = p({0}) = 0 and p(1) = p(H) = 1.
Conversely, if D is any density matrix, then the map 3 M 7 p(M ) = tr(DP M )
is a state in S (). States of this type are called tracial states. The natural question is whether every state in S () arises as a tracial state. The answer to this
question is one of the most profound results in the mathematical foundations
of quantum mechanics, the celebrated Gleasons theorem:
Theorem 7.2.10 (Gleason). Let H be a complex separable Hilbert space with
3 dim H 0 , D(H) the convex set of density matrices on H, and the logic of
subspaces of H. Then
1. the map D(H) 3 D 7 D S (), defined by D (M ) = tr(DP M ) for all M
, is a convex isomorphism of D(H) on S (),
2. a state p S () is pure if and only if p = p for some ray inH,
3. two pure states p and p are equal if and only if there exists a complex
number c with |c| = 1 such that the rays and verify = c.
The proof, lengthy and tricky, is omitted. It can be found, extending over 13
pages (!), in [31], pages 147160.
7.3 Symmetries
Definition 7.3.1. A linear map S : H H is a symmetry if
1. S is bijective, and
2. for all f , g H, the scalar product is preserved: S f | Sg = f | g .
Exercise 7.3.2. Let Aut() where is the standard quantum logic associated with a given Hilbert space H. Show that
92
1. there exists a symmetry S B(H) such that for all M , (M ) = SM ,

2. if S 0 is another symmetry corresponding to the same automorphism ,
then there exists a complex number c, with |c| = 1 such that S 0 = cS,
3. if S is any symmetry of H, the map 3 M 7 SM is an automorphism
of .
Notice that unitaries are obviously symmetries. It turns out that they are the
only symmetries encountered in elementary quantum systems2 .
In general, anti-unitaries may also occur as symmetries. They are not considered in this
course.
93
7.3. Symmetries
94
8
States, effects, and the corresponding
quantum formalism
8.1 States and effects

8.2 Operations
8.3 General quantum transformations, complete positivity, Kraus theorem
95
8.3. General quantum transformations, complete positivity, Kraus theorem
96
lud on 12 January 2013
9
Quantum formalism based on the
informational approach
97
Chapter 9
/Users/dp/a/ens/mq/iq-infor.tex
98
10
Two illustrating examples
10.1 The harmonic oscillator

In chapter 6, a general formalism, covering both classical and quantum logics,
has been introduced. Here we present a simple physical example, the harmonic
oscillator, in its classical and quantum descriptions. Beyond providing a concrete illustration of the formalism developed so far, this example has the advantage of being completely solvable and illustrating the main similarities and
differences between classical and quantum physics.
10.1.1 The classical harmonic oscillator

The system is described by a mass m attached to a spring of elastic constant
k. The motion is assumed frictionless on the horizontal direction and the mass
originally equilibrates at point 0. The spring is originally elongated to position
q 0 and the system evolves then freely under the equations of motion. The setting is described in figure 10.1. The system was already studied in chapter 2.
99
10.1. The harmonic oscillator
Figure 10.1: The experimental setting of the one-dimensional harmonic oscillator.

The equation of motion, giving the elongation q(t ) as a function of time t , is
) =
m q(t
f (q(t )) = kq(t )
q(0) = q 0
q(0)
= v 0 = 0.
Introducing the new variable p = m q and transforming the second order differential equation into a system of first order equations, we get the vector equation
d
(t ) = A(t ),
dt
where
()

q(t )
q0
(t ) =
, with initial condition (0) =
p0
p(t )
and
0
A=
k
1
m .
The solution to equation () is given by a flow on the phase space = R2 given

by
(t ) = T t (0),
where
T = exp(t A) =
cos(t )
sin(t )
m
k sin(t ) cos(t )
p
k/m. Since det T t = 1, it follows that the evolution
is invertible and

q0
(T t )1 = T t . The orbit of the initial condition (0) =
under the flow reads
0
!
q 0 cos(t )
(T t )t R , where (t ) = T t =
.
q 0 k sin(t )
and =
/Users/dp/a/ens/mq/iq-ilexa.tex
100
The system is classical, hence its logic is a Boolean -algebra; the natural
choice is = B(R2 ). Now observables in O () are mappings x : B(R)
B(R2 ). Identify henceforth indicator functions with Borel sets in B(R2 ) (i.e.
for any Borel set B B(R), instead of considering x(B ) = F B(R2 ) we shall
identify x(B ) = 1 F .)
Let now X : R be any measurable bounded mapping
and chose as
R
x(B ) = 1 X 1 (B ) for all B B(R). Then, on defining X = x(d ), a bijection
is established between x and X . Now since (T t )t R = (exp(t A))t R is the orbit of the initial condition 0 in , the value X (T t ) is well defined for all t R;
we denote by X t () X (T t ). Then
d Xt
d (T t )1
d (T t )2
() = 1 X (T t )
+ 2 X (T t )
dt
dt
dt
d
p
d
q
(t ) + 2 X (T t )
(t ),
= 1 X (T t )
dt
dt
provides the evolution of X under the flow (T t )t .
The Hamiltonian is a very particular measurable bounded map on the phase
space (hence an observable) H : R, having the formula H () = k21 /2 +
22 /2m. It evolves also under the flow (T t )t : Then
p(t )
d Ht
)+
)
() = kq(t )q(t
p(t
dt
m
) + q(t
)(k q(t
))
= kq(t )q(t
= 0.
Thus, the Hamiltonian is a constant of motion. Physically it represents the enkq 2
ergy of the system. Initially, H (q 0 , p 0 ) = 2 0 = E and during the flow, the energy always remains E , so that the energy takes arbitrary (but constant with
) and
respect to the flow) values E R+ . Moreover, 1 H (T t ) = kq(t ) = p(t
p(t )
t
). Hence we recover the Hamilton equations
2 H (T ) = m = q(t
dq
H
(t ) =
= 2 H
dt
p
dp
H
(t ) =
= 1 H .
dt
q
Therefore, ddXt t = 1 X 2 H + 2 X (1 H ) = L H X with L H = (1 H 2 2 H 1 ).
Hence, denoting for every two function f , g C 1 () by { f , g } = 1 f 2 g 2 f 1 g
the Poissons bracket, we have for the flow of an observable, assuming integrability of the evolution equation, X t = exp(t L H )X . This means that the flow
101

(T t )t ) on induces a flow (exp(t L H )X )t on observables. Notice also that
X t = exp(t L H )X is a shorthand notation for
Xt =
(t )n
X
{H , {H , . . . {H , X } . . .}}.
n=0 n!
Theorem 10.1.1 (Liouvilles theorem). Let be the Lebesgue measure on , i.e.

(d 1 d 2 ) = d 1 d 2 . Then
1. the measure is invariant under T t , i.e. (T t B ) = (B ) for all B B(R2 )
and all t R,
2. the operator L H is formally skew-adjoint on L 2 (, F , ).
Proof:
R
1. (T t B ) = T t B d 1 d 2 . Now, if T t B T t B . Hence, denoting
(x 1 , x 2 ) = T t (1 , 2 ), we have
Z
TtB
d 1 d 2 =
=
Z
ZB
B
(1 , 2 )
d x1 d x2
(x 1 , x 2 )
d x 1 d x 2 = (B ),
because the Jacobian verifies

(1 , 2 )
= det exp(t A) = 1.
(x 1 , x 2 )
2. L H is not bounded on L 2 (, F , ). It can be defined on dense subset of
L 2 (, F , ), for instance the Schwartz space S(R2 ). For f , g S(R2 ), we
have
Z
f | LH g =
f ()L H g ()(d )
Z
= L H f ()g ()(d ) + bdry terms.
Now the boundary terms vanish because f and g vanish at infinity. Hence,
on S(R2 ), the operator is skew-adjoint L H = L H and hence formally skewadjoint on L 2 (, F , ).
102

Notice that, as a consequence of the previous theorem, exp(t L H ) is formally
unitary on L 2 (, F , ).
Any probability measure p on is a state. We have for all B B(R), x (B ) =
p(x(B )) = p(X 1 (B )) while x t = p(x t (B )) = p(X t1 (T t B )) = p(x(T t B )). Hence
the flow T t on induces a convex automorphism (p)(x(B

)) = p(x((T t B )) on
states.
10.1.2 Quantum harmonic oscillator

Standard quantum logic coincides with the family of subspaces of an infinitedimensional Hilbert space H. Since all separable Hilbert spaces are isomorphic,
we can chose any of them. The Schrdingers choice for the one-dimensional
harmonic oscillator is H = L 2 (R). States are probability measures p : [0, 1]
and thanks to Gleasons theorem, we can limit ourselves to tracial states, i.e.
3 M 7 p(M ) = tr(P M D) = p D (M ),
for some D D(H). Symmetries are implemented by unitary operators on H
(automorphisms on .) Let U U(H). Then : M 7 (M ) = U M induces a
projection PU M = U P M U . Subsequently, the automorphism induces a convex automorphism on S (), given by
(p)(M
) = p D ((M ))
= tr(PU M D)
= tr(U P M U D)
= tr(P M D (U ) ),
with D (U ) = U DU . Physics remains invariant under time translations. Hence
time translation (evolution) must be a symmetry implemented by a unitary
operator U (t ) acting on H. Define U (t ) = exp(i t H /) (this a definition of
H .) Then H is formally self-adjoint, hence an observable (a very particular
one!) generating the Lie group of time translations. It will be shown below
that H is time invariant. Now U (t ) acts on rays of H to give a flow. Denoting
(t ) = U (t ), we have the Schrdingers evolution equation in the Schrdingers
picture:
d
(t ) = H (t ).
i
dt
103

Thanks to the spectral theorem (and, identifying for x O () and B B(R),
x(B ) with the projection-valued measure corresponding to the subspace x(B )),
thereR is a bijection between x O () and self-adjoint operators
on H through
R
X = x(d ). For every tracial states p D , we have Ep D (X ) = tr(x(d )D) and
Z
E(p
tr(x(d )D U (t ) )
D ) (X ) =
Z
=
tr(U (t )x(d )U (t )D)
= Ep D (X t ),
where we defined X t = U (t )X U (t ). Hence the flow U (t ) on H induces a flow
on observables satisfying
i
d Xt
= [H , X ] = L H X
dt
with L H () = i [H , ]. Notice incidentally that d H t /d t = 0 proving the claim that

H is a constant of motion. Moreover, H has dimensions M L 2 /T 2 (energy),
therefore H is interpreted as the quantum Hamiltonian. If the flow is integrable,
we have
Xt
= exp(t L H )X
(i t )n
X
[H , [H , . . . , [H , X ] . . .]].
=
n
n=0 n!
Physics remains invariant also by space translations. Hence they must correspond to a symmetry implemented by a unitary transformation.
Lemma 10.1.2. The operator x is formally skew-adjoint on L 2 (R).
Proof: For all f , g S(R) (dense in L 2 (R)), we have, f | x g =
R
ddx f (x)g (x)d x + f g |
.
f (x) ddx g (x)d x =
Consequently, the operator exp(x x ) is formally unitary and since exp(x

x )(y) = (y +x), x is the generator of space translations. If we write p = i x
then p is formally self-adjoint, has dimensions L M (L/T 2 ) (1/L) = M L/T
(momentum), and exp(i x p/) is unitary and implements space translations.
Define Hosc = p 2 /2m + kq 2 as the formally self-adjoint operator on L 2 (R),
with p = i x and q(x) = x(x), the multiplication operator. Introduce =
p
p
p
k/m, Q = m/q, P = (1/ m)p, and H = (1/)Hosc . Then H = (1/2)(P 2 +
Q 2 ) where P = i and Q is the multiplication operator; these two latter operators are formally self-adjoint and verify the commutation relation [P,Q] = i 1 .
104
Definition 10.1.3. (Creation and annihilation operators) Define the creation

operator A = p1 (P + iQ) and the annihilation operator A = p1 (P iQ).
2
Exercise 10.1.4. For the creation and annihilation operators, show

1. [A, A ] = 1 ,
2. H = A A + 1 /2,
3. [H , A] = A,
4. [H , A ] = A ,
5. for n N, [H , (A )n ] = n(A )n .
Lemma 10.1.5. If 0 S(R) is a ray (in the L 2 sense) satisfying A0 = 0 then
1. 0 (x) = 1/4 exp(x 2 /2),
2. H 0 = 0 /2, and
3. H (A )n 0 = (1/2 + n)A n 0 , for all n N.
Proof:
1
p (P iQ)0
2
d
i
0 (x) i x0 (x) = 0
dx
0 (x) = c exp(x 2 /2),
A0 = 0
and by normalisation, c = 1/4 .
Lemma 10.1.6. Denote, for n N, n = p1 A n 0 . Then

n!
1. (n )nN is an orthonormal sequence,

p
2. A n = n + 1n+1 , for n 0,
3. An =
p
nn1 , for n 1, and
4. A An = nn , for n 0.
105

Proof: All the assertions can be shown by similar arguments. It is enough to
show the arguments leading to orthonormality:
0 | A n A n 0 = 0 | A n1 A A A n1 0
= 0 | A n1 (1 + A A)A n1 0
..
.
= n 0 | A n1 A n1 0
..
.
= n! 0 | 0 .
Theorem 10.1.7. The sequence (n )nN is a complete orthonormal sequence in

H.
The proof is based on an analogous result for Hermite polynomials that can
be shown using the two following lemmata.
Lemma 10.1.8. Let c n, j =
Then
c n, j = (1
n!
,
(n2 j )!2 j j !
for n N, and j N such that 0 j n/2.
2j
2( j + 1)
)c n+1, j =
)c n+1, j +1
n +1
(n + 1)(n 2 j )
and if
n (x) =
[n/2]
X
(1) j c n, j x n2 j ,
j =0
then
(x
while x n =
P[n/2]
j =0
d
) n (x) = n+1 (x)
dx
c n, j n2 j (x).
Proof: Substitute and make induction.

p
Lemma 10.1.9. (A n 0 )(x) = n ( 2x)0 (x).
Proof: True for n = 0. Conclude by induction.
Corollary 10.1.10. spec(H ) = 1/2 + N.

Therefore the energy is quantised in quantum mechanics i.e. it can take only
discrete values. It is this surprising phenomenon that gave its adjective quantum to the term quantum mechanics.
106
Exercise 10.1.11. Using Diracs notation | n n , for n N,
1. H | n = (1/2 + n)| n ,
2. A | n =
3. A| n =
p
n + 1| n + 1 ,
p
n| n 1 , and
4. A A| n = n| n .
10.1.3 Comparison of classical and quantum harmonic oscillators
Figure 10.2: Comparison of probability densities. In blue is depicted the probability density of the classical oscillator. In red the corresponding density for
the quantum oscillator for n = 10 (left) and n = 60 (right).
107
10.2. Schrdingers equation in the general case, rigged Hilbert spaces
Figure 10.3: Comparison of distribution functions. In blue is depicted the distribution of the classical oscillator. In red the corresponding distribution for the
quantum oscillator for n = 1 (left), n = 10 (middle), and n = 30 (right). Already
for n = 30, the classical and quantum distributions are almost indistinguishable.
10.2 Schrdingers equation in the general case, rigged

Hilbert spaces
10.3 Potential barriers, tunnel effect
10.4 The hydrogen atom
108
11
Quantifying information: classical
and quantum
11.1 Classical information, entropy, and irreversibility

The information content of a message is a probabilistic notion. The less probable a message is, the more information it carries. Let X be a random variable
defined on (, F , P) taking values in the finite set X = {x 1 , . . . , x n } Let PVn = {p
P
Rn+ : ni=1 p i = 1}. To each element p PVn corresponds a probability measure
p
p
P X defined by P X (x i ) = p i , for i = 1, . . . , n. Ask about the information content
carried by the random variable X is the same thing as trying to quantify the predictive power of the law P X . The main idea is that the information content of
X is equal to the average information missing in order to decide the outcome
value of X when the only thing we know is its law P X . Some reasonable requirements on the information content of X are given below:
Suppose that all p i , i = 1, . . . , n but one are 0 and p j = 1, for some j . Then
p
P X (x j ) = 1 and no information is missing, there is no uncertainty about
the possible outcome of X ;
109
11.1. Classical information, entropy, and irreversibility

Suppose on the contrary that p i = 1/n, i = 1, . . . , n. Our perplexity is maximal and this perplexity increases with n.
If S is to be interpreted as a missing information associated with a probability vector p PVn , on denoting PV = nN PVn , the function S : PV
R+ and the first statement implies that S(1, 0, 0, . . . , 0) = 0 while S(1/n, . . . , 1/n)
is an increasing function of n.
The function S must be invariant under permutations of its arguments
i.e. S(p (1) , . . . , p (n) ) = S(p 1 , . . . , p n ) for all the permutations S n .
If we split the possible outcome values into two sets, the function S must
verify the grouping property, i.e.
S(p 1 , . . . , p n ; p n+1 , . . . , p N ) = S(q A , q B )
pn
p1
)
+q A S( , . . . ,
qA
qA
p n+1
pN
+q B S(
,...,
),
qB
qB
where q A = p 1 + . . . + p n and q B = p n+1 + . . . + p N .
Finally, we require S(p 1 , . . . , p n ; 0, . . . , 0) = S(p 1 , . . . , p n ).
Theorem 11.1.1. The only function S : PV R+ satisfying the above requirements is the function defined by
n
X
PV 3 (p 1 , . . . , p n ) 7 S(p 1 , . . . , p n ) = k p i log p i ,
i =1
where k is an arbitrary non-negative constant and the convention 0 log 0 = 0 is

used. The function S is called the (classical) entropy of the probability vector.
Proof: (To be filled in a later version.)
Entropy is closely related to irreversibility since the second principle of thermodynamics states: Entropy of an isolated system is a non decreasing function
of time. It can remain constant only for reversible evolutions. For a system A
undergoing an irreversible transformation the entropy increases; however the
system can be considered as part of a larger isolated composite system (A and
environment), undergoing globally a reversible transformation. In that case
the total entropy (of the system A and of the environment) remains constant
but since the entropy of A must increase, the entropy of the environment must
decrease1 hence the missing information decreases. In other words, when the
1
Notice that this assertion is not in contradiction with the second principle of thermodynamics because the environment is not isolated.
/Users/dp/a/ens/mq/iq-qinfo.tex
110
system A undergoes an irreversible transformation, the environment gains information.

This leads to the Landauers principle: When a computer erases a single bit
of information, the environment gains at least k ln 2 units of information, where
k > 0 is a constant.
111
11.1. Classical information, entropy, and irreversibility
112
12
Turing machines, algorithms,
computing, and complexity classes
All computers, from Babbages never constructed project of analytical machine

(1833) to the latest model of supercomputer, are based on the same principles.
A universal computer uses some input (a sequence of bits) and a programme
(a sequence of instructions) to produce an output (another sequence of bits.)
Universal computers are modelled by Turing machines. Never forget however
that an abstract Turing machine never computed anything. We had to wait until
the first ENIAC was physically constructed to obtain the first output of numbers.
12.1 Deterministic Turing machines

There are several variants of deterministic Turing machines; all of them are
equivalent in the sense that a problem solvable by one variant is also solvable
by any other variant within essentially the same amount of time (see below,
definition ?? and section 12.3.) A Turing machine is a model of computation; it
is to be thought as a finite state machine disposing of an infinite scratch space
113
12.1. Deterministic Turing machines

(an external tape1 .) The tape consists of a semi-infinite or infinite sequence of
squares, each of which can hold a single symbol. A tape-head can read a symbol
from the tape, write a symbol on the tape, and move one square in either direction (for semi-infinite tape, the head cannot cross the origin.) More precisely, a
Turing machine is defined as follows.
Definition 12.1.1. A deterministic Turing machine is a quadruple (A, S, u, s 0 )
where
1. A is a finite set, the alphabet, containing a particular symbol called the

blank symbol and denoted by ]; the alphabet deprived from its blank symbol, denoted A b = A \ {]}, is assumed non-empty,
2. S is a finite non-empty set, the states of the machine, partitioned into the
set S i of intermediate states and the set S f of final states,
3. D = {L, R} {1, 1} is the displacement set,
4. u : A S A S D is the transition function, and
5. s 0 S i the initial state of the machine.
The set of deterministic Turing machines is denoted by DTM.
The machine is presented an input, i.e. a finite sequence of contiguous nonblank symbols, and either it stops by producing an output, i.e. another finite
sequence of symbols, else the programme does never halt.
Example 12.1.2. (A very simple Turing machine) Let M DTM with A = {0, 1, ]},
S = S i S f where S i = {go}, S f = {halt}, and transition function u(a, s) = (a 0 , s 0 , d )
defined by the following table:
a
0
1
]
go
go
go
a0
0
1
]
s0
go
go
halt
d
L
L
R
Mind that during Turings times no computer was physically available. The external tape
was invented by Alan Turing who was fascinated by typewritters as an external storage
device.
/Users/dp/a/ens/mq/iq-turin.tex
114
If the programme, described by this Turing machine, starts with the head over
any non-blank symbol of the input string, it ends with the head over the leftmost non-blank symbol while the string of symbols remains unchanged.
Other equivalent variants of the deterministic Turing machine may have
displacement sets with a 0 (do not move) displacement, have their alphabet A
partitioned into external and internal alphabet, etc. The distinction into internal and external alphabet is particularly useful in the case of semi-infinite tape,
an internal character , identified as first symbol, can be used to prevent the
head from going outside the tape. It is enough to define U (, go) = (, go, R).
Notation 12.1.3. If W is a finite set, we denote by W = nZ+ W n and W =
W = W Z+ . Notice that Z+ = {0, 1, 2, . . .} 6= N = {1, 2, . . .} and that W 0 = {;}. Elements of W are called words of finite length over the alphabet W . For every
w W , there exists n Z+ such that w W n ; we denote then by |W | = n the
length of the word w.
For every A b , we denote by A the completion of the word by
blanks, namely = (1 , . . . , || , ], ], ], . . .).
Considering the example 12.1.2, we can, without loss of generality, always
assume that the machine starts at the first symbol of the input string = A b .
Starting from (, s 0 , h 0 = 1), successive applications of the transition function
U induce a dynamical system on X = A S Z. A configuration is an instantaneous description of the word written on the tape, the internal state of the
machine, and the position of the head, i.e. an element of X.
Let = inf{n 1 : s n S f }. The programme starting from initial configuration (, s 0 , h 0 = 1) stops running if < , it never halts when = . While
1 n < , the sequence ((n) , s n , h n )n is defined by updates of single characters; if, for 0 n < , we have u((n)
, s ) = (a 0 , s 0 , d ), then ((n+1) , s n+1 , h n+1 ),
hn n
is defined by
s n+1 = s 0
h n+1 = h n + d
(n)
(n)
0 (n)
(n+1) = ((n)
1 , . . . , h 1 , a , h +1 , . . . , |(n) | ).
n
If the machine halts at some finite instant, the output is obtained by reading
the tape from left to right until the first blank character. The sequence of words
((n) )n is called a computational path or computational history starting from .
115
12.2. Computable functions and decidable predicates
12.2 Computable functions and decidable predicates

Every M DTM computes a particular partial function M : A b A b . Since the
value of M () remains undetermined when the programme M does not halt,
the function M is termed partial because in general Dom(M ) A b .
Definition 12.2.1. A partial function f : A b A b is called computable if there
exists a M DTM such that M = f . In that case, f is said to be computed by
the programme M .
Exercise 12.2.2. Show that there exist non-computable functions.
Definition 12.2.3. A predicate, P , is a function taking Boolean values 0 or 1. A
language, L, over an alphabet A is a subset of A b .
Thus, for predicates P with Dom(P ) = A b , the set { A b : P ()} is a language. Hence predicates are in bijection with languages.
Definition 12.2.4. A predicate P : A b {0, 1} is decidable, if the function P is
computable.
Let P be a predicate and L the corresponding language. The predicate is
decidable if there exists a M DTM such that for every word , the programme
halts after a finite number of steps and
if L, then the machine halts returning 1, and
if 6 L, then the machine halts returning 0.
Definition 12.2.5. Let M DTM and s M , t M : Z+ R+ be given functions. If for
every A b , the machine stops after having visited at most s M (||) cells, we say
that it works in computational space s M . We say that it works in computational
time t M if t M (||).
12.3 Complexity classes

Computability of a function does not mean effective computability since the
computing algorithm can require too much time or space. We say that r : N
R+ is of polynomial growth if there exist constants c,C > 0 such that r (n) C n c ,
for large n. We write symbolically r (n) = poly(n).
116
Henceforth, we shall assume A b A = {0, 1}.

Definition 12.3.1. The complexity class P consists of all languages L whose
predicates P are decidable in polynomial time, i.e. for every L in the class, there
exists a machine M DTM such that M = P and t M (||) = poly(||) for all
A .
Similarly, we can define the class PSPACE of languages whose predicates are
decidable in polynomial space. functions computable in polynomial space.
Other complexity classes will be determined in the subsequent sections.
Obviously P PSPACE.
Conjecture 12.3.2.
P 6= PSPACE .
12.4 Non-deterministic Turing machines and the NP

class
Definition 12.4.1. A non-deterministic Turing machine is a quadruplet (A, S, u, s 0 )
where A, S and s 0 are as in definition 12.1.1; u is now a multivalued function,
i.e. there are r different branches u i , i = 1, . . . , r and u i : A S A S D. For
every pair (a, s) A S there are different possible outputs (a i0 , 0i , d i )i =1,...,r , the
choice of a particular branch can be done in a non-deterministic way at each
moment. All such choices are legal actions. The set of non-deterministic Turing
machines is denoted by NTM.
A computational path for a M NTM is determined by a choice of one legal
transition at every step. Different steps are possible for the same input. Notice
that NTM do not serve as models of practical devices but rather as logical tools
for the formulation of problems rather than their solution.
Definition 12.4.2. A language L (or its predicate P ) belongs to the
there exists a M NTM such that
NP
class if
if L (i.e. P () = 1) for some A , then there exists a computational

path with poly(||) returning 1,
if 6 L (i.e. P () = 0) for some A , then there exists no computational
path with this property.
117
12.5. Probabilistic Turing machine and the bpp class

It is elementary to show that P NP. Clay Institute offers you2 USD 1 000 000
if you solve the following
Exercise 12.4.3. Is it true that P = NP?
12.5 Probabilistic Turing machine and the BPP class

be the set of real numbers computable by a determinDefinition 12.5.1. Let R
istic Turing machine within accuracy 2n in poly(n) time. A probabilistic Turing
machine is a quintuple (A, S, u, p, s 0 ) where A, S, u, and s 0 are as in definition
P
+ , with r p i = 1 is a probability vector on the
12.4.1 while p = (p 1 , . . . , p r ) R
i =1
set of branches of u. All branches correspond to legal actions; at each step, the
branch i is chosen with probability p i , independently of previous choices. The
set of probabilistic Turing machine is denoted by PTM.
Each A generates a family of computational paths. The local probability structure on the transition functions induces a natural probability structure
on the computational path space. The evolution of the machine is a Markov
process with the state space A b S Z and stochastic evolution kernel determined by the local probability vector p. Hence, any input gives a set of possible
outputs each of them being assigned a probability of occurrence. A machine in
PTM is also called a Monte Carlo algorithm.
Definition 12.5.2. Let ]0, 1/2[. A predicate P (hence a language L) belongs
to the BPP class if there exists a M BPP such that for any A , poly(||)
and
if L, then P(P () = 1) 1 , and
if 6 L, then P(P () = 1) .
Exercise 12.5.3. Show that the definition of the class does not depend on the
choice of provided it lies in ]0, 1/2[.
Church-Turing thesis . . .
2
http://www.claymath.org/millennium/
118
12.6 Boolean circuits

Notation 12.6.1. For b N and Zb = {0, . . . , b 1}, we denote by x = x n1 x 0 b
the mapping defined by
Znb 3 (x 0 , . . . , x n ) 7 x = x n x 0 b =
n1
X
x k b k Zb n .
k=0
Since conversely for every x Zb n the sequence (x 0 , . . . , x n ) Znb is uniquely

determined, we identify x with the sequence of its digits. For b = 2 we omit the
basis subscript and we write simply .
Definition 12.6.2. Let f : An Am be a Boolean function of n entries and m
outputs. Let B be a fixed set of Boolean functions of different arities. We call
Boolean circuit of f in terms of the basis B a representation of f in terms of
functions from B.
Example 12.6.3. (Addition with carry of 2 binary 2-digit numbers) Let x =
x 1 x 0 and y = y 1 y 0 . We wish to express z = x + y = z 2 z 1 z 0 in terms of
Boolean functions in B = {XOR, AND} = {, }. The truth table is given in table
12.1. We verify immediately that:
z0 = x0 y 0
z 1 = (x 0 y 0 ) (x 1 y 1 )
z 2 = (x 1 y 1 ) [(x 1 y 1 ) (x 0 y 0 )]
Consequently, the Boolean circuit is depicted in figure ??.
A basis B is complete if any Boolean function f can be constructed as a circuit with gates from B.
Example 12.6.4. {NOT, OR, AND} is a complete but redundant basis; {NOT, OR},
{NOT, AND}, and {AND, XOR} are complete minimal bases.
Definition 12.6.5. The minimal number of gates from B needed to compute f ,
denoted by c B ( f ), is circuit complexity of f in B.
The function implementing the addition with carry of table 12.1 over the
basis B = {AND, XOR}, has circuit complexity 7.
Any DTM can be implemented by circuits.
119
12.6. Boolean circuits

x1
x0
y1
y0
z2
z1
z0
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
0
0
0
0
1
1
1
1
0
0
0
0
1
1
1
1
0
0
0
0
0
0
0
1
0
0
1
1
0
1
1
1
0
0
1
1
0
1
1
0
1
1
0
0
1
0
0
1
0
1
0
1
1
0
1
0
0
1
0
1
1
0
1
0
Table 12.1: The truth table of the Boolean function A4 A3 implementing the
addition with carry of two binary 2-digit numbers.
Classical computers are based on gates {XOR, AND} for example. It is easily
shown that these gates are irreversible. Therefore it is intuitively clear why classical computers can produce information. What is much less intuitively clear
is how quantum processes can produce information since they are reversible
(unitary).
In 1973, B ENNETT predicted that it is possible to construct reversible universal gates. In 1982, F REDKIN exemplifies such a reversible gate. Fredkins gate
is a 3 inputs - 3 outputs gate, whose truth tableau is given in table ??. This gate
produces both AND (since inputs 0, x, y return outputs x y, x y, x) and NOT
gates (since inputs 1, 0, x return outputs x, x, x.) The gates AND and NOT forming a complete basis for Boolean circuits, the universality of Freidkins gate is
established.
In 1980, B ENIOFF describes how to use quantum mechanics to implement
a Turing machine, in 1982, F EYNMAN proves that there does not exist a Turing
machine (either deterministic or probabilistic) on which quantum phenomena can be efficiently simulated; only a quantum Turing machine could do so.
Finally, in 1985, D EUTSCH constructs (on paper) a universal quantum Turing
120
Input
a b c
0 0 0
0 0 1
0 1 0
0 1 1
1 0 0
1 0 1
1 1 0
1 1 1
Output
a b0 c 0
0 0 0
0 0 1
0 1 0
1 0 1
1 0 0
0 1 1
1 1 0
1 1 1
0
Table 12.2: The truth table of Fredkins gate. We remark that c 0 = c and if c = 0
then (a 0 = a and b 0 = b) else (a 0 = b and b 0 = a.)
machine.
12.7 Composite quantum systems, tensor products,

and entanglement
Example 12.7.1. Let n = 2 and H1 = H2 = C2 . A basis of H2 is given by (| 00 , | 01 , | 10 , | 11 ).
An arbitrary vector H2 is decomposed as
= 0 | 00 + 1 | 01 + 2 | 10 + 3 | 11 .
If 2 = 3 = 0, then = 0 | 00 + 1 | 01 = | 0 (0 | 0 + 1 | 1 ) and the state is
not entangled. If 1 = 2 = 0 while 0 3 6= 0 then the state is entangled since it
cannot be written as a tensor product.
12.8 Quantum Turing machines

be the set of complex numbers whose real and imagDefinition 12.8.1. Let C
inary part can be computed by a deterministic algorithm with precision 2n
within poly(n) time. A pre-quantum Turing machine is a quadruple (A, S, c, s 0 ),
where A, S, s 0 are as for a deterministic machine and c : (A S)2 D C , where
D is the displacement set.
121
12.8. Quantum Turing machines

Any configuration x of the machine is represented by a triple x = (, s, h)
A S Z = X. The quantum configuration space H is decomposed into HT
HS HH , where the indices T, S, H stand respectively for tape, internal states,
and head. The space H is spanned by the orthonormal system (| )X = (| sh )A ,sS,hZ .
Define now onservables having (| )A , (| s )sS , and (| h )hZ as respective eigenvectors. To do so, identify the sets A with {0, . . . , |A| 1} and S with
and H the self-adjoint operators describing these
{0, . . . , |S|1}. Denoty by T , S,
observables, i.e.
S =
|S|1
X
s| s s |
s=0
h| h h |
hZ
= i Z Ti where Ti =
|A|1
X
a| a a |.
a=0
Due to the linearity of quantum flows, it is enough to describe the flow on

the basis vectors = | , s, h ; a A Z , s S, h Z. The machine is prepared at
some initial pure state = | , s, h , with a string of contiguous non blank
symbols and we assume that the time is discretised:
| n = U n | .
Suppose that the displacement set D reads {1, 0, 1}. Then for = | , s, h and
0 = | 0 , s 0 , h 0
U,0 = 0 , s 0 , h 0 |U , s, h
= [h 0 ,h+1 c(h , s, 0h , s 0 , 1)
+h 0 ,h c(h , s, 0h , s 0 , 0)
+h 0 ,h1 c(h , s, 0h , s 0 , 1)]
Y
j Z\{h}
j ,0 .
j
Definition 12.8.2. A pre-quantum Turing machine is called a quantum Turing

machine if the function c is such that the operator U is unitary.
Exercise 12.8.3. Find the necessary and sufficient conditions on the function
c so that U is unitary.
Wavelets, Cuntz-Krieger algebras, Bratteli-Jrgensen . . .
To halt the machine, we can not perform intermediate measurements of
the composite state because quantum mechanical measurement perturbs the
122
system. To proceed, suppose that S f = {halt} {0} and introduce a halting flag
operator F = | 0 0 |. Once the state s is set to 0, the function c is such that U
does not any longer change either the state s or the result of the computation.
A predicate is a projection operator P = | |. Let the machine evolve
for some time n: it is at the state | n = U n | . Perform the measurement
n | P F I n = p [0, 1].
Definition 12.8.4. A language L belongs to the BQP complexity class if there is
a machine M QTM such that
if L, then the machine accepts with probability p > 2/3,
if 6 L, then the machine rejects with probability p > 2/3,
within a running time poly(||).
Theorem 12.8.5.
P BPP BQP PSPACE .
123
12.8. Quantum Turing machines
124
13
Cryptology
Cryptology, grouping cryptography and cryptanalysis, is an old preoccupation

of mankind because information is, as a matter of fact, a valuable resource.
Nowadays classical technology allows secure ciphering of information that cannot be deciphered in real time. However, the cryptologic protocols used nowadays are all based on the unproven conjecture that factoring large integers is a
hard computational task. Should this conjecture be proved false, and an efficient polynomial factorisation algorithm be discovered, the security of our
communication networks could become vulnerable. But even without any technological breakthrough, the ciphered messages we exchange over public channels (internet, commutated telephone network, fax, SMS, etc.) can be deciphered by spending 810 months of computing time; hence our information
exchange is already vulnerable for transporting information that remains important 10 months after its transmission.
Quantum information acquired an unprecedented impetus when Peter Shor
[30] proved that on a quantum computer, factoring is a polynomially hard problem. On the other hand, quantum communication can use the existing technology to securely cipher information. It is therefore economically and strategically important to master the issues of advanced cryptography and to invent
new cryptologic methods.
125
13.1. An old idea: the Vernams code
13.1 An old idea: the Vernams code

In 1917, Gilbert V ERNAM proposed [34] the following ciphering scheme.1 Let
A be a finite alphabet, identified with the set {0, . . . , |A| 1} and m a message
of length N over the alphabet A, i.e. a word m A N . The Vernams ciphering
algorithm uses a ciphering key of same length as m, i.e. a word k A N and
performs character-wise addition as explained in the following
Algorithm 13.1.1. VernamsCiphering
Require: Original message m A N and U NIF R ANDOM G ENERATOR(A N )
Ensure: Ciphered message c A N
Choose randomly ciphering key k A N
i 1
repeat
Add character-wise c i = m i + k i mod |A|
i i +1
until i > N
The recipient of the ciphered message c, knowing the ciphering key k performs the following
Algorithm 13.1.2. VernamsDeciphering
Require: Ciphered message c A N and ciphering key k A N
Ensure: Original message m A N
i 1
repeat
Subtract character-wise m i = c i k i mod |A|
i i +1
until i > N
As far as the ciphering key is used only once and the key word has the same
length as the message, the Vernams algorithm is proved [28] to be perfectly
secure. The main problem of the algorithm is how to securely communicate
the key k?
Appeared as a first patent US Patent 1310719 issued on 22 July 1919, and further improved
in a series of patents: US Patent 1416765, US Patent 1584749, and US Patent 1613686.
/Users/dp/a/ens/mq/iq-crypt.tex
126
13.2 The classical cryptologic scheme RSA

Theorem 13.2.1. (Fermats little theorem) Let p be a prime. Then
1. any integer a satisfies a p = a mod p,
2. any integer a, not divisible by p, satisfies a p1 = 1 mod p.
Definition 13.2.2. The Eulers function : N N is defined by
(n) = card{0 < a < n : gcd(a, n) = 1}, n N.
In particular, if p is prime, then (p) = p 1.
Theorem 13.2.3. (Eulers) If gcd(a, m) = 1, then a (m) = 1 mod m.
Proposition 13.2.4. Let m be an integer, strictly bigger than 1, without square
factors, and r a multiple of (m). Then
a r = 1 mod m, for all integers a relatively prime with respect to m, and
a r +1 = a mod m for all integers.
The proofs of all the previous results are straightforward but outside the
scope of the present course; they can be found in pages 5060 of [?].
The RSA protocols, named after its inventors Rivest, Shamir, and Adleman
[24], involves two legal parties: Alice and Bob, and an eavesdropper, Eve. Bob
produces by the classical key distribution algorithm a private key d and a public
key . Alice uses the public key of Bob to cipher the message and Bob uses his
private key to decipher it. Eve, even if she intercepts the ciphered message,
cannot decipher it in real time.
Algorithm 13.2.5. ClassicalKeyDistribution
Require: Two primes p and q
Ensure: Public, , and private, d , keys of Bob
n pq (hence (n) = (p 1)(q 1))
Choose any e < n, such that gcd(e, (n)) = 1
d e 1 mod (n)
(e, n)
Bob publishes his public key on his internet page. Alice uses to cipher
the message m using the following
127
13.2. The classical cryptologic scheme RSA

Algorithm 13.2.6. Ciphering
Require: Public key = (e, n) and message m N, with m < n
Ensure: Ciphered text c N
c m e mod n
Alice transmits the ciphered text c through a vulnerable public channel to
Bob. He uses his private key to decipher by using the following
Algorithm 13.2.7. Deciphering
Require: Private key d and ciphered message c N
Ensure: Deciphered text N
c d mod n
Theorem 13.2.8. = m
Proof:
cd
= m ed
ed
= 1 + k(n), for some k N
ed
mod n
= m 1+k(n) ,
and since n = pq has no square factors, by using proposition 13.2.4, we get

m 1+k(n) mod n = m mod n.
If Eve intercepts the message, to compute d she must know (n), hence
the factoring of n into primes. Security of the protocol is based on the conjecture that it is algorithmically hard to factor n. If we denote by N = log n, then
it is worth noticing that when the RSA protocol has been introduced, the best
known algorithm of factor n run in exp(N ) time. The best 2 known algorithm
nowadays [18] runs in exp(N 1/3 (log N )2/3 ) time. This algorithmic improvement,
combined with the increasing in the computational capabilities of computers,
allows the factoring of a 1000 digits number in ca. 8 months instead of a time
exceeding the age of the universe at the moment the algorithm has been proposed. Until May 2007, the RSA company ran an international contest offering
several hundreds thousand dollars to whoever could factor multi-digit numbers they provided on line. When the contest stopped the company gave the
official reasons explained in RSA factoring challenge.
2
See also [20] for an updated state of the art.
128
13.3 Quantum key distribution

13.3.1 The BB84 protocol
Theorem 13.3.1. (No cloning theorem) Let | and | be two rays in H such
that | 6= 0 and | 6= exp(i )| . Then there does not exist any quantum
device allowing duplication of and .
Proof: Suppose that such a device exists. Then, for some n 1, there exists a
unitary U : H(n+1) H(n+1) and some ancillary ray | 1 n Hn such that
we get
| 1 n1 = U | 1 n
| 1 n1 = U | 1 n .
Then
| = 1 n |U U | 1 n
n1
Y
i | i .
= | 2
i =1
Q
Since | 6= 0 we get | n1
i =1 i | i =Q1 and since | 6= exp(i )| psi ,
it follows that 0 < | | | < 1. Subsequently, n1
i =1 | i | i | > 1 but this is impossible since for every i , | i | i | 1.
This theorem is at the basis of the BB84 quantum key distribution protocol [8]. Alice and Bob communicate through a quantum and a classical public
channels; they agree publicly to use two different orthonormal bases of H = C2
(describing the photon polarisation):
+
B + = {+
0 = | 0 , 1 = | 1 }
|0|1 |0+|1
, 1 =
}.
B = {
p
p
0 =
2
2
The first element of each basis is associated with the bit 0, the second with the
bit 1. Moreover Alice and Bob agree on some integer n = (4 + )N with some
> 0, where N is the length of the message they wish to exchange securely;
it will be also the length of their key. Alice needs also to know the function
129
13.3. Quantum key distribution

T : {0, 1}2 H defined by
+
0
+
1
T (x, y) =
if (x, y) = (0, 0)
if (x, y) = (0, 1)
if (x, y) = (1, 0)
if (x, y) = (1, 1).
Algorithm 13.3.2. AlicesKeyGeneration

Require: U NIF R ANDOM G ENERATOR({0, 1}), T , n
Ensure: Two strings of n random bits a, b {0, 1}n and a sequence of n qubits
(| i )i =1,...,n
Generate randomly a 1 , . . . , a n
a (a 1 , . . . , a n ) {0, 1}n
Generate randomly b 1 , . . . , b n
b (b 1 , . . . , b n ) {0, 1}n
i 1
repeat
| i T (a i , b i )
Transmit | i to Bob via public quantum channel
i i +1
until i > n
On reception of the i th qubit, Bob performs a measurement of the projec]
]
tion operator P ] = | 1 1 |, where ] {+, }.
130
Algorithm 13.3.3. BobsKeyGeneration

Require: U NIF R ANDOM G ENERATOR({0, 1}), n, sequence | i for i = 1, . . . , n, P ]
for ] {+, }
Ensure: Two strings of n bits 0 , b0 {0, 1}n
Generate randomly b 10 , . . . , b n0
b0 (b 10 , . . . , b n0 ) {0, 1}n
i 1
repeat
if b i0 = 0 then
ask whether P + takes value 1
else
ask whether P takes value 1
end if
if Counter triggered then
a i0 1
else
a i0 0
end if
i i +1
until i > n
a0 (a 10 , . . . , a n0 ) {0, 1}n
Transmit string b0 {0, 1}n to Alice via public classical channel
When Alice receives the string b, she performs the conciliation algorithm
described below.
Algorithm 13.3.4. Conciliation
Require: Strings b, b0 {0, 1}
Ensure: Sequence (k 1 , . . . , k L ) with some L n of positions of coinciding bits
c b b0
i 1
k 1
repeat
k min{ j : k j n such that c j = 0}
if k n then
ki k
i i +1
end if
until k > n
L i 1
transmit (k 1 , . . . , k L ) to Bob via public classical channel
131

Theorem 13.3.5. If there is no eavesdropping on the quantum channel then
P((a k0 1 , . . . , a k0 L ) = (a k1 , . . . , a kL )|a, b) = 1.
Proof: Compute i | P + i and i | P i for all different possible choices of

i B + B . We observe that for those i s such that b i0 = b i we have P(a i0 =
a i ) = 1. Hence on deciding to consider only the substrings of a and a0 defined
on the locations where b and b0 coincide, we have the certainty of sharing the
same substrings, although a and a0 have never been exchanged.
Lemma 13.3.6. If there is no eavesdropping, for N large enough, L is of the order

2N .
Proof: Elementary use of the law of large numbers.
If Eve is eavesdropping, since she cannot copy quantum states (no-cloning

theorem), she can measure with the same procedure as Bob and in order for
i to Bob.
the leakage not to be apparent, she re-emits a sequence of qubits |
Now again L is of the order 2N but since Eves choice of the bs is independent of
the choices of Alice and Bob, the string a0 computed by Bob will coincide with
Alices string a at only L/2 ' N positions.
Hence to securely communicate, Alice and Bob have to go through the eavesdropping detection procedure and reconciliation.
Bob randomly chooses half of the bits of the substring (a k0 , . . . , a k0 ), i.e. (a r0 1 , . . . , a r0 L/2 )
1
L
with r i {k 1 , . . . , k L } and r i 6= r j for i 6= j , and sends the randomly chosen positions (r 1 , . . . , r L/2 ) and the corresponding bit values (a r0 1 , . . . , a r0 L/2 ) to Alice. If
(a k0 , . . . , a k0 ) = (a k1 , . . . , a kL ) (reconciliation) then Alice announces this fact to
1
L
Bob and they use the complementary substring of (a k0 , . . . , a k0 ) (of length L/2 '
1
L
N ) as their key to cipher with Vernams algorithm. Else, they restart BB84 protocol.
Notice that Alice and Bob never exchanged the ultimate substring of N bits
they use as key.
132
13.3.2 Simple eavesdropping strategies, disturbance and informational gain

13.3.3 Other cryptologic protocols
133
134
14
Elements of quantum computing
In this chapter, B denotes the set {0, 1} and elements b B are called bits; H
will denote C2 and rays | H are called qubits. Similarly, arrays of n bits are
denoted by b = (b 1 , . . . , b n ) B n ; arrays of n qubits by | = | 1 n Hn .
14.1 Classical and quantum gates and circuits

A classical circuit implements a Boolean mapping f : B n B n by using elementary gates of small arities1 , chosen from a family G; A quantum circuit implements a unitary mapping U : Hn Hn by using unitary elementary gates
of small arities2 , chosen from a family G.
Definition 14.1.1. Let U : Hn Hn for some n and G be a fixed family of
unitary operators of different arities. A quantum circuit over G is a product of
operators from G acting on appropriate qubit entries.
It is usually assumed that G is closed under inversion.
1
2
usually acting on O (1) bits.

usually acting on O (1) qubits.
135
14.2. Approximate realisation

Definition 14.1.2. Let V : Hn Hn be a unitary operator. This operator is
said to be realised by a unitary operator W : HN HN , with N n entries,
acting on n qubits and N n ancillary qubits, if for all | Hn ,
W (| | 0N n ) = (V | ) | 0N n .
Ancillary qubits correspond to some memory in a fixed initial state we borrow for intermediate computations that is returned into the same state. Returning ancillary qubits into the same state can be relaxed. What cannot be relaxed
is that ancilla must not be entangled with the n qubits (it must remain in tensor
form); otherwise the anicllary subsystem could not be forgotten.
Quantum circuits are supposed to be more general than classical circuits.
However, arbitrary Boolean circuits cannot be considered as classical counterparts of quantum ones because the classical analogue of a unitary operator on
Hn is an invertible map on B n , i.e. a permutation S 2n . Since to any n-bit array = (1 n ) B n corresponds a basis vector | = | 1 n Hn , to every
defined by
permutation S 2n naturally corresponds a unitary operator ,
= | () ,
|
d
1 . Hence we can define:
with = 1 =
Definition 14.1.3. Let G S 2n . A reversible circuit over G is a sequence of permutations from G.

An arbitrary Boolean function F : B m B n can be extended to a function
F : B m+n B m+n , defined by
F (x, y) = (x, y F (x)),
where the symbol in the right hand side stands for the bit-wise addition modulo 2. It is easily checked that F is a permutation. Moreover F (x, 0) = (x, F (x)).
Notice that 2-bit permutation gates do not suffice the realise all functions
of the form F . On the contrary G = {NOT, } with : B 3 B 3 the Toffoli gate,
defined by (x, y, z) = (x, y, z (x y)), is a basis.
14.2 Approximate realisation

There are uncountably many unitary operators U : Hn Hn . Hence if a
quantum computer is to be constructed, the notion of exact realisation of a
/Users/dp/a/ens/mq/iq-qcomp.tex
136
lud on 16th March 2004
unitary operator must be weakened to an approximate realisation. The same

rationale prevails also in classical computing, instead of all real functions (uncountably many), only Boolean functions are implemented.
Lemma 14.2.1. An arbitrary unitary operator U : Cm Cm can be represented
Q
V (i ) of matrices of the form
as a product V = m(m1)/2
i =1
1
.
..
a b
c d
..
.
a b
U(2).
,
c d
1
Moreover, the sequence of matrices appearing in the product can be explicitly
constructed in a running time O (m 3 )poly(log(1/)) where = kU V k.
Proof: An exercise if one recalls that for all c 1 , c 2 C, there exists a unitary operator W U(2) such that
p
c1
|c 1 |2 + |c 2 |2
.
W
=
c2
0
Basic properties of the operator norm are recalled below:

kX Y k kX kkY k
kX k = kX k
kU k = 1
kX Y k = kX kkY k,
where X and Y are arbitrary operators and U is a unitary.
Definition 14.2.2. A unitary operator U 0 approximates a unitary operator U
within if kU U 0 k .
Lemma 14.2.3. If a unitary U 0 approximates a unitary U within , then U 01
approximates U 1 within .
137
14.2. Approximate realisation

Proof: Since U 01 (U 0 U )U 1 = U 1 U 01 , it follows that kU 1 U 01 k kU 0
U k .
Lemma 14.2.4. If unitary operators (Uk0 )k=1,...,L approximate unitary operators

(U )
within k , then U 0 = UL0 U10 approximates U = UL U1 within
PLk k=1,...,L
.
k=1 k
Proof: kU20 U10 U2U1 k kU20 (U10 U1 ) + (U20 U2 )U1 k 1 + 2 .
Definition 14.2.5. A unitary operator U : Hn Hn is approximated by a unitary operator U : HN HN , with N n, within if for all | Hn
kU 0 (| | 0N n ) U | | 0N n k kk.
Definition 14.2.6. For every unitary operator U : Hn Hn there exists a unitary operator C (U ) : H Hn H Hn , called the controlled-U operator, defined for all | Hn by
C (U )| | =
||
if = 0
| U | if = 1
Similarly, multiply controlled-U C k (U ) : Hk Hn Hn Hn , is defined by

k
C (U )| 1 k | =
| 1 k |
if 1 k = 0
| 1 k U | if 1 k = 1
0 1
Example 14.2.7. Let 1 =
be the unitary operator corresponding to the
1 0
where is the Toffoli gate.
classical NOT gate. Then C 2 (1 ) = ,
Definition 14.2.8. The set

G = {H , K , K 1 ,C (1 ),C 2 (1 )},
1 1
1 0
(Hadamard gate) and K =
(phase gate), is called
with H =
1 1
0 i
the standard computational basis.
p1
2
Theorem 14.2.9. Any unitary operator U : Hn Hn can be approximated

within by a poly(log(1/))-size circuit over the standard basis using ancillary
qubits. There is a poly(n)-time algorithm describing the construction of the approximating circuit.
Proof: An exercise, once you have solved the exercise 14.2.10 below.
138
Exercise 14.2.10. Let 0,...,3 be the 3 Pauli matrices augmented by the identity
1
0
.
matrix, H the Hadamard gate, and () =
0 exp(2i )
1. Show that if A M2 (C) with A 2 = 1 and R, then
exp(i A) = cos 0 + i sin A.
), where
2. Let R j () = exp(i 2 j ), for j = 1, 2, 3 and R n () = exp(i 2 n ~
2
2
2
= (1 , 2 , 3 ). Express R j ()
n = (N1 , n 2 , n 3 ) with n 1 + n 2 + n 3 = 1 and ~
and R n () on the basis 0 , . . . , 3 .
3. Show that H = exp(i )R 1 ()R 3 (), for some , , to be determined.
4. If | C2 is a ray represented by a vector of the Bloch sphere S2 = {x
R3 : kxk2 = 1}, show that
R n ()| = | Tn ()x
where Tn ()x is the rotation of x around n by an angle .
5. Show that every U U(2) can be written as
U = exp(i )R n ()
for some , R.
6. Show that every U U(2) can be written as
U = exp(i )R 3 ()R 2 ()R 3 ()
for some , , , R.
and n are two not parallel vectors of S2 . Show that every
7. Suppose that m
U U(2) can be written as
U = exp(i )R n (1 )R m (1 )R n (2 )R m (2 ) .
8. Establish identities
H 1 H
= 3
H 2 H
= 2
H 3 H
H ( )H
8
= 1
= exp(i )R 1 ( )
4
for some .
139
14.3. Examples of quantum gates
14.3 Examples of quantum gates

14.3.1 The Hadamard gate
1 1 1
H=p
.
2 1 1
1
H | = p ((1) | + | 1 ), B.
2
H
7
1 X
| 000 = p
| x .
8 x=0
14.3.2 The phase gate
1
0
() =
.
0 exp(2i )
()| = exp(2i )|

+ )H ()H | 0 = cos | 0 + exp(i ) sin | 1 .
4 2
14.3.3 Controlled-NOT gate
0
C (1 ) =
0
0
0
1
0
0
0
0
0
1
0
.
1
0
For any x B , C (1 )| x0 = | xx , but for arbitrary | = | 0 + | 1 ,

C (1 )| 0 = | 00 + | 11 6= | .
140
14.3.4 Controlled-phase gate
0
C (()) =
0
0
0
0
0
0
.
1
0
0 exp(2i )
0
1
0
0
For x, y B ,
C (())| x y = exp(2i x y)| x y .
14.3.5 The quantum Toffoli gate

For all x, y, z B ,
C 2 (3 )| x y z = | x, y, (x y) z .
Suppose that f : B m B n is a Boolean function, implemented by the uni1 P

tary operator U f : H(n+m) H(n+m) . If | = 2m/2
1 ,...,m B | 1 , . . . , m then
n
Uf ||0 =
2m
1
X
2m/2
x=0
| x, f (x) .
Hence computing simultaneously all values of f over its domain of definition

requires the same computational effort as computing the value over a singleton
of the domain.
141
14.3. Examples of quantum gates
142
15
The Shors factoring algorithm
143
Chapter 15
/Users/dp/a/ens/mq/iq-shora.tex
144
16
Error correcting codes, classical and
quantum
145
References
146
Bibliography
[1] N. I. Akhiezer and I. M. Glazman. Theory of linear operators in Hilbert

space. Dover Publications Inc., New York, 1993. Translated from the Russian and with a preface by Merlynd Nestell, Reprint of the 1961 and 1963
translations, Two volumes bound as one. 31
[2] William Arveson. A short course on spectral theory, volume 209 of Graduate
Texts in Mathematics. Springer-Verlag, New York, 2002. 48, 52, 59
[3] Alain Aspect, Jean Dalibard, and Grard Roger. Experimental test of
5brells inequalities using time-varying analyzers. Phys. Rev. Lett., 49:1804
1807, Dec 1982. 16, 20
[4] Alain Aspect, Philippe Grangier, and Grard Roger. Experimental tests of
realistic local theories via Bells theorem. Phys. Rev. Lett., 47:460463, Aug
1981. 16
[5] Alain Aspect, Philippe Grangier, and Grard Roger. Experimental realization of Einstein-Podolsky-Rosen-Bohm Gedankenexperiment: A new violation of Bells inequalities. Phys. Rev. Lett., 49:9194, Jul 1982. 16, 19
[6] William Aspray. John von Neumann and the origins of modern computing.
MIT Press, Cambrdige, MA, 1990. 23
[7] John S. Bell. On the problem of hidden variables in quantum mechanics.
Rev. Modern Phys., 38:447452, 1966. 16
[8] C. H. Bennett and G. Brassard. Quantum public key distribution system.
IBM Technical disclosure bulletin, 28:31533163, 1985. 129
[9] David Bohm. A suggested interpretation of the quantum theory in terms
of "hidden" variables. I. Phys. Rev., 85:166179, Jan 1952a. 16
[10] David Bohm. A suggested interpretation of the quantum theory in terms
of "hidden" variables. II. Phys. Rev., 85:180193, Jan 1952b. 16
147
References
[11] Giulio Chiribella, Giacomo Mauro DAriano, and Paolo Perinotti. Theoretical framework for quantum networks. Phys. Rev. A (3), 80(2):022339, 20,
2009. 23
[12] Giulio Chiribella, Giacomo Mauro DAriano, and Paolo Perinotti. Informational derivation of quantum theory. Phys. Rev. A, 84:012311, Jul 2011.
23
[13] A. Einstein, B. Podolsky, and N. Rosen. Can quantum-mechanical description of physical reality be considered complete? Phys. Rev., 47:777780,
May 1935. 15
[14] Nicolas Gisin, Grgoire Ribordy, Wolfgang Tittel, and Hugo Zbinden.
Quantum cryptography. Rev. Mod. Phys., 74:145195, Mar 2002. 28
[15] Anthony J.G. Hey. Feynman and computation. Perseus Books Publishing,
1998. 22
[16] John K. Hunter and Bruno Nachtergaele. Applied analysis. World Scientific
Publishing Co. Inc., River Edge, NJ, 2001. 31, 33
[17] Richard V. Kadison and John R. Ringrose. Fundamentals of the theory of
operator algebras. Vol. I, volume 15 of Graduate Studies in Mathematics.
American Mathematical Society, Providence, RI, 1997a. Elementary theory, Reprint of the 1983 original. 36, 50
[18] A. K. Lenstra and H. W. Lenstra, Jr. Algorithms in number theory. In Handbook of theoretical computer science, Vol. A, pages 673715. Elsevier, Amsterdam, 1990. 4, 128
[19] A. K. Lenstra and H. W. Lenstra, Jr., editors. The development of the number
field sieve, volume 1554 of Lecture Notes in Mathematics. Springer-Verlag,
Berlin, 1993. 6
[20] Arjen K. Lenstra. Integer factoring. Des. Codes Cryptogr., 19(2-3):101128,
2000. Towards a quarter-century of public key cryptography. 128
[21] H. Maassen. Quantum probability and quantum information theory. In
Quantum information, computation and cryptography, volume 808 of Lecture Notes in Phys., pages 65108. Springer, Berlin, 2010. 16, 20
[22] Dimitri Petritis. Markov chains on measurable spaces, 2012. Lecture notes
of the University of Rennes 1. 21
148
BIBLIOGRAPHY
[23] Michael Reed and Barry Simon. Methods of modern mathematical physics.
I. Functional analysis. Academic Press, New York, 1972. 31, 36
[24] R. L. Rivest, A. Shamir, and L. Adleman. A method for obtaining digital signatures and public-key cryptosystems. Comm. ACM, 21(2):120126, 1978.
6, 127
[25] Walter Rudin. Real and complex analysis. McGraw-Hill Book Co., New
York, third edition, 1987. 31
[26] Walter Rudin. Functional analysis. International Series in Pure and Applied Mathematics. McGraw-Hill Inc., New York, second edition, 1991. 48
[27] Raymond A. Ryan. Introduction to tensor products of Banach spaces.
Springer Monographs in Mathematics. Springer-Verlag London Ltd., London, 2002. 36
[28] Claude Shannon. Communication theory of secrecy systems. Bell System
Technical Journal, 28:656715, 1949. 126
[29] Albert Nikolaevich Shiryayev. Probability, volume 95 of Graduate Texts in
Mathematics. Springer-Verlag, New York, 1984. Translated from the Russian by R. P. Boas. 11
[30] Peter W. Shor. Polynomial-time algorithms for prime factorization and
discrete logarithms on a quantum computer. SIAM J. Comput., 26(5):1484
1509, 1997. 4, 6, 125
[31] V. S. Varadarajan. Geometry of quantum theory. Vol. I. D. Van Nostrand Co.,
Inc., Princeton, N.J.-Toronto, Ont.-London, 1968. The University Series in
Higher Mathematics. 92
[32] V. S. Varadarajan. Geometry of quantum theory. Springer-Verlag, New York,
second edition, 1985. 23
[33] V. S. Varadarajan. Geometry of quantum theory. Springer-Verlag, New York,
second edition, 1985. 73
[34] Gilbert S. Vernam. Cipher printing telegraph systems for secret wire and
radio telegraphic communications, volume 55. 1926. 126
[35] Johann von Neumann.
Mathematische Grundlagen der Quantenmechanik. Unvernderter Nachdruck der ersten Auflage von 1932. Die
Grundlehren der mathematischen Wissenschaften, Band 38. SpringerVerlag, Berlin, 1968. 22
149
Index
[36] John von Neumann. Collected works. Vol. I: Logic, theory of sets and quantum mechanics. General editor: A. H. Taub. Pergamon Press, New York,
1961a. 23
[37] John von Neumann. Collected works. Vol. II: Operators, ergodic theory and
almost periodic functions in a group. General editor: A. H. Taub. Pergamon
Press, New York, 1961b. 23
[38] John von Neumann. Collected works. Vol. III: Rings of operators. General
editor: A. H. Taub. Pergamon Press, New York, 1961c. 23
[39] John von Neumann. Collected works. Vol. IV: Continuous geometry and
other topics. General editor: A. H. Taub. Pergamon Press, Oxford, 1962. 23
[40] John von Neumann. Collected works. Vol. V: Design of computers, theory of
automata and numerical analysis. General editor: A. H. Taub. A Pergamon
Press Book. The Macmillan Co., New York, 1963a. 23
[41] John von Neumann. Collected works. Vol. VI: Theory of games, astrophysics, hydrodynamics and meteorology. General editor: A. H. Taub. A
Pergamon Press Book. The Macmillan Co., New York, 1963b. 23
[42] Joachim Weidmann. Linear operators in Hilbert spaces, volume 68 of Graduate Texts in Mathematics. Springer-Verlag, New York, 1980. Translated
from the German by Joseph Szcs. 36
150
Index
algebra, 42
Banach, 43
Boolean, 72
Boolean -, 72
commutative, 42
C , 44
normal element of an, 43
normed, 43
self-adjoint element of an, 43
unital, 42
unitary element of an, 43
algebraic
adjoint, 43
isometry, 43
tensor product, 24
algorithm
Monte Carlo, 118
automorphism
convex, 84
computational
space, 116
time, 116
computational history, 115
computational path, 115
conditional expectation
classical, 34
consistency, 10
density matrix, 91
determinant, 52
direct sum, 34
eigenvalue, 55
entanglement, 25
entropy, 110
equation
Hamilton, 13
Schroedinger, 103
Euler function, 127
expected value, 79
Banach algebra
general linear group of invertible function-valued measure, 61
elements of, 56
gate
invertible element of, 56
Freidkin, 120
Boolean circuit, 119
complement, 72
completion, 32
complexity class, 116
BPP , 118
NP , 117
P , 117
PSPACE , 117
computable function, 116
harmonic oscillator
classical, 99
quantum, 103
Hasse diagram, 71
hatling flag, 123
Hilbert
norm, 32
inequality
151
Index
Cauchy-Schwarz-Buniakovski, 32
involution, 42
isometry, 47
partial, 48
joint probability distribution, 84
language, 116
lattice, 70
atom (in a), 72
atomic, 72
automorphism, 72
complemented, 72
Dilworth, 72
distributive, 71
isomorphism, 72
modular, 72
orthocomplemented, 74
orthomodular, 72
lattice
-complete, 72
lattice of propositions, 70
logic, 74
classical, 75
probability measure on, 78
standard quantum, 75
state on, 78
measurement, 7
classical, 14
quantum, 23
momentum operator, 67
observable, 7
associated with a logic, 76
bounded, 76
classical, 14, 73
constant, 76
discrete, 76
quantum, 23
spectral value of, 76
spectrum of, 76
strict value of an, 76

operator
adjoint, 66
annihilation, 105
bounded, 33
closable, 66
closed, 66
closure, 66
creation, 105
densely defined, 66
density, 91
diagonalisable, 59
extension of an, 66
graph of, 66
Hermitean, 46
invertible, 54, 66
isometric, 47
norm of , 33
normal, 49
orhoprojection, 46
positive, 46
projection, 46
restriction of an, 66
self-adjoint, 46
trace-class, 89
unbounded, 66
unitary, 47
von Neumann, 91
orhogonal
subsets, 34
vectors, 34
orthocomplementation, 74
orthogonal
complement, 34
orthoprojection, 35, 46
othogonal
orthoprojections, 46
partial isometry, 48
partially ordered set, 70
phase space
152
INDEX
classical, 13
quantum, 23
Poisson bracket, 101
polynomial growth, 116
poset, 70
position operator, 66
predicate, 116
predicate:decidable, 116
probability distribution, 79
projection, 35, 46
proposition
simultaneously verifiable, 82
quantum circuit, 135
question, 78
associated with proposition, 78
random variable, 9
law of, 9
ray, 89
representation
faithful, 50
of locally compact group, 85
space, 50
, 49
unitarily equivalent, 50
resolvent
set, 54
scalar product, 32
sequence
Cauchy, 32
fundamental, 32
simultaneous observability, 82
space
Banach, 32
complete, 32
Hilbert, 32
metric, 32
normed, 32
pre-Hilbert, 32
spectral measure, 62
spectral radius, 58
spectrum, 54, 57
continuous, 55
point, 55
residual, 55
-algebra, 43
-homomorphism, 43
state
classical, 13
entagled, 25
entangled, 25
pure, 80
superposition, 80
tracial, 92
state function, 77
stochastic process, 10
symmetry, 92
tensor product, 24
theorem
Gelfand-Namark, 50
Gleason, 92
Kolmogorovs existence, 11
Liouville, 102
of spectral decomposition, 64
time evolution
classical, 14
quantum, 23
trace, 89
Turing machine
deterministic, 114
non-deterministic, 117
pre-quantum, 121
probabilistic, 118
quantum, 122
variance, 79
yes-no question, 14
153

2 Pdfsam Iq PDF

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

2 Pdfsam Iq PDF

Загружено:

Авторское право:

Доступные форматы

Master de mathmatiques

Rennes (draft of 2 October 2013 at 13:45)

2003 2013 D Petritis

1 Physics, mathematics, and mathematical physics

2 Phase space, observables, measurements, and yes-no experiments

2.2 Classical physics as a probability theory with a dynamical law . .

2.2.1 Some reminders from probability theory . . . . . . . . . . .

2.2.2 Postulates for classical systems . . . . . . . . . . . . . . . . .

2.3 Classical probability does not suffice to describe Nature! . . . . . .

2.3.1 Bells inequalities . . . . . . . . . . . . . . . . . . . . . . . . .

2.3.2 The Orsay experiment . . . . . . . . . . . . . . . . . . . . . .

2.4 Quantum systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.4.1 Postulates of quantum mechanics: the Hilbert space approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.4.2 Interpretation of the basic postulates . . . . . . . . . . . . .

2.5 First consequences of quantum formalism . . . . . . . . . . . . . .

2.5.1 Irreducibility of quantum randomness and Heisenbergs uncertainty principle . . . . . . . . . . . . . . . . . . . . . . . . 29

3 Short resum of Hilbert spaces

3.1 Scalar products and Hilbert spaces . . . . . . . . . . . . . . . . . . .

3.2 Orthogonality and projection; direct and orthogonal sums . . . .

3.3 Duality and Frchet-Riesz theorem; adjoint . . . . . . . . . . . . .

3.4 Tensor product of Hilbert spaces . . . . . . . . . . . . . . . . . . . .

3.5 Orhonormal systems, ortonormal bases; criteria of completeness

3.6 Diracs bra and ket notation . . . . . . . . . . . . . . . . . . . . . . .

3.7 Some more involved quantum mechanical phenomena . . . . . .

3.7.1 Composite systems and entanglement . . . . . . . . . . . .

3.7.2 Decoherence and quantum to classical transition . . . . . .

3.8 Rigged Hilbert spaces and generalised kets . . . . . . . . . . . . . .

4.1 Introduction and motivation . . . . . . . . . . . . . . . . . . . . . .

4.2 Algebra of operators . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.3 Convergence of sequences of operators . . . . . . . . . . . . . . . .

4.4 Classes of operators in B(H) . . . . . . . . . . . . . . . . . . . . . .

4.4.1 Self-adjoint and positive operators . . . . . . . . . . . . . . .

4.4.3 Unitary operators . . . . . . . . . . . . . . . . . . . . . . . . .

4.4.4 Isometries and partial isometries . . . . . . . . . . . . . . . .

4.4.5 Normal operators . . . . . . . . . . . . . . . . . . . . . . . . .

4.5 States on algebras, GNS construction, representations . . . . . . .

5 Spectral theory in Banach algebras

5.2 The spectrum of an operator acting on a Banach space . . . . . . .

5.3 The spectrum of an element of a Banach algebra . . . . . . . . . .

5.4 Relation between diagonalisability and the spectrum . . . . . . . .

5.5 Spectral measures and functional calculus . . . . . . . . . . . . . .

5.6 Some basic notions on unbounded operators . . . . . . . . . . . .

6 Propositional calculus and quantum formalism based on quantum logic 69

6.2 Classical, fuzzy, and quantum logics; observables and states on

6.2.2 Observables associated with a logic . . . . . . . . . . . . . .

6.2.3 States on a logic . . . . . . . . . . . . . . . . . . . . . . . . . .

6.3 Pure states, superposition principle, convex decomposition . . . .

6.4 Simultaneous observability . . . . . . . . . . . . . . . . . . . . . . .

6.5 Automorphisms and symmetries . . . . . . . . . . . . . . . . . . . .

7 Standard quantum logics

8 States, effects, and the corresponding quantum formalism

8.1 States and effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

8.3 General quantum transformations, complete positivity, Kraus theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

9 Quantum formalism based on the informational approach

10 Two illustrating examples

10.1 The harmonic oscillator . . . . . . . . . . . . . . . . . . . . . . . . .

10.1.1 The classical harmonic oscillator . . . . . . . . . . . . . . . .

10.1.2 Quantum harmonic oscillator . . . . . . . . . . . . . . . . . 103

11.1 Classical information, entropy, and irreversibility . . . . . . . . . . 109

12.1 Deterministic Turing machines . . . . . . . . . . . . . . . . . . . . . 113

12.7 Composite quantum systems, tensor products, and entanglement 121

13.1 An old idea: the Vernams code . . . . . . . . . . . . . . . . . . . . . 126

14 Elements of quantum computing