Vous êtes sur la page 1sur 22

Section 1: Vector Space,

Matrix Mechanics, and


Representation Theory
X.Bai
1. Probability Amplitudes
and Their Compositions

Section 1: Vector Space, Matrix


Mechanics, and Representation Theory

2. Hilbert Space and


Dirac Notations

(Textbook: Chapter 9, Chapter 10, Chapter 15)

Xinhua Bai
SDSM&T

1 / 22

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory

1.1 Probability Amplitudes and Their Compositions


The expansion postulate of Quantum Mechanics: Every physical
with
quantity can be represented by a Hermitian operator (A)
eigenfunctions
1 , 2 , ..., n , ... and every physical state by a
P
sum R i ci i , where the coefficients are defined as
ci = i (x)(x)dx.
More specifically, the eigenvalue equation for the Hermitian
reads:
operator A
n (x) = An n (x),
A

X.Bai
1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

(1)

th

where An denotes the n eigenvalue. The eigenfunctions satisfy


the orthonormality (orthogonality & normalization!) condition:
Z
n (x)k (x)dx = nk
(2)
And arbitrary state function (x) can be expressed as the series:
X
(x) =
cn n (x)
(3)
n
2 / 22

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory

with the expansion coefficients given by:


Z
cn = n (x)(x)dx

X.Bai
1. Probability Amplitudes
and Their Compositions

(4)

2. Hilbert Space and


Dirac Notations

The coefficients cn are called the probability amplitudes. We


have to distinguish two cases:
will give
1. An is a simple eigenvalue. Then measurements of A
2
the result An with probability |cn | . If (x) = n (x) then
will yield An with
|cn |2 = 1, i.e. a measurement of A
certainty.
2. An is not a simple eigenvalue, i.e. it corresponds to more
than one linearly independent eigenfunctions. In this case
alone
additional labeling is required, i.e. measurements of A
will not unambiguously identify all the terms in Eq. (3)
expansion.

3 / 22

In Case 2, how do we do the additional labeling of the


such
eigenstates? We have to find another Hermitian operator B
that these two operators have a common set of eigenstates.
These operators are called compatible or simultaneously
and B
to be
measurable. We will prove that in order for A
compatible it is necessary and sufficient that they commute, i.e.
B]=0.

[A,
Then we can label the eigenstates of A using different
If the ambiguity still remains we will have to
eigenvalues of B.
until a complete set of
find another Hermitian operator C
compatible Hermitian operators is found.

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory
X.Bai
1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

The maximum information about the outcome of physical


measurements on a physical system is contained in the
probability amplitudes that belong to a complete set of
observables for the system.
Example: Particle in a spherical potential. Complete set of
L
2 , L
2z }. Corresponding quantum numbers are
operators are {H,
{n, l, m}.

4 / 22

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory
X.Bai

1.2 Properties of the Probability Amplitudes

1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

DiracRnotations. It is convenient to replace an integral


cn = R n (x)(x)dx with a short notation
cn = n (x)(x)dx hAn |i.
The corresponding probability amplitude is hAn |i.
on
|hAn |i|2 gives the probability that a measurement of A
will yield An .
P
2
In this notation, the completeness:
n |hAn |i| = 1.

In this notation, the orthonormality: hAn |Am i = nm .

We will introduce more about the Dirac notation of quantum


quantities/states later on.

5 / 22

1.3 Composition Rule for the Probability Amplitudes


To see the relation between amplitudes for two sets of
observables, we start from the fact that an arbitrary state
function can be equally well specified with eigenstates of any
complete set of observables, for example,
n = An n
by {n } that belong to A
Or
k = Bk k .
by {k } that belong to B
It reads like this:
X
X
(x) =
hBk |ik (x).
hAn |in (x) =
n

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory
X.Bai
1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

On the other hand: k (x) = n hAn |Bk in (x). Insert this to


previous expression, we find:
X
X
X
hAn |Bk in (x)
hAn |in (x) =
hBk |i
n

X
n

XX
hAn |Bk ihBk |in (x)
hAn |in (x) =
n

k
6 / 22

This leads to the composition rule:


X
hAn |i =
hAn |Bk ihBk |i

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory
X.Bai

(5)

Equation (5) embodies the strangeness of quantum mechanics.


on
Indeed, one can calculate the probability of measuring A
using Eq. (5):

1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations


2
X



|hAn |i| = hAn |Bk ihBk |i


2

This differs drastically from a conventional rule for the


calculation of probabilities:
X
2
2
|hAn |Bk i| |hBk |i|
k

This is because there are interference terms in quantum


superposition.
7 / 22

One can replace with j in Eq. (5):


X
hAn |Bk ihBk |Aj i = hAn |Aj i = nj

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory
X.Bai

(6)

2. Hilbert Space and


Dirac Notations

On the other hand,


X
X
2
|hAn |Bk i| =
hAn |Bk i hAn |Bk i = 1
k

1. Probability Amplitudes
and Their Compositions

Comparing these equations we obtain


hAn |Bk i = hBk |An i

(7)

and
2

|hAn |Bk i| = |hBk |An i|

(8)

this is a so-called reciprocal relationship: The conditional


in a state that is known to have
probability of finding An for A

the sharp value of Bk for B equals to the probability of finding Bk


in a state that is known to have a sharp value of An for A.

for B
8 / 22

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory

The probability amplitudes hAn |Bk i define a matrix


Snk = hAn |Bk i. This is a unitary matrix because
X
Snk Skj = nj

X.Bai
1. Probability Amplitudes
and Their Compositions

(9)

2. Hilbert Space and


Dirac Notations

or
SS = 1

(10)

where S is a Hermitian conjugate (i.e. transpose and complex


conjugate) of S.
I

The unitary matrix is inverse to its own Hermitian conjugate.

Real unitary matrices are called orthogonal, they are inverse


to their transpose matrices.

Orthogonal matrices describe rotations in Euclidean space.

Similarly, unitary matrices describe rotations in Hilbert


space.

9 / 22

2.1 Bra and Ket vectors


Consider a probability amplitude:

X.Bai
1. Probability Amplitudes
and Their Compositions

Z
hA|Bi h|i =

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory

(x)(x)dx

2. Hilbert Space and


Dirac Notations

We call this quantity inner product. Using the integral


definition of the inner product we can establish the following
properties, for a complex number a:
Z
h|ai = adx = ah|i
Z
ha|i = (a) dx = a h|i
h|i = h|i ,
By analogy with dot product of two conventional 3D vectors
a b, we will follow Dirac and call notations h| - bra-vector, and
|i - ket-vector (bra, ket both stem from bracket).
10 / 22

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory

From linearity of the integral we derive the following properties of


ket- and bra-vectors:
|a1 i + |b2 i = a|1 i + b|2 i

X.Bai
1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

ha1 | + hb2 | = a h1 | + b h2 |
Question: What kind of mathematical objects are these bra- and
ket-vectors?
Before we proceed let us discuss properties of vectors in 3D
Euclidean space R 3 . A vector ~a is defined by three scalars:
~a = (ax , ay , az ) = ax i + ay j + az k
We say that three unit vectors i, j, k span a linear vector space
R 3 . The numbers ax , ay , az define a vector ~a in a particular
coordinate system (basis).

11 / 22

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory

However we can consider an abstract vector ~a and establish


vector calculus without any reference to a particular basis, simply
~a R 3 . What are the essential attributes of R 3 (or any other
linear vector space)?

X.Bai
1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

Mathematically, the Definition of a linear vector space is A set of


objects with the following properties is called a linear vector
space:
I If ~
a R 3 and ~b R 3 then ~a + ~b R 3
I

I
I

For any two vectors we can introduce a dot product


~a ~b = ax bx + ay by + az bz = ab cos

every vector has a length (norm): |~a| = ~a ~a


R 3 is complete, i.e. contains all its limiting points (there are
no holes).

We can straightforwardly generalize this definition for any


number of dimensions.

12 / 22

In general, we can consider a space of infinite dimension, which


will lead to the notion of the Hilbert space. The elements of the
Hilbert space are functions.
The idea is based on the fact that any function can be expanded
in an infinite series with respect to a complete, orthonormal set
of (base) functions. Consider a familiar example: a particle in a
box have eigenstates described by:
p
n (x) = 2/L sin(nx/L).
Every function (x) with
represented as:

RL
0

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory
X.Bai
1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

||2 dx < can be uniquely

an n ,

n=1

p
RL
where an = 2/L 0 sin(nx/L)(x)dx. So, we P
can define an

infinite vector |i (a1 , a2 , ..., an , ...) {an } n=1 an |n i,


where an = hn |i.

13 / 22

An abstract Hilbert space L is a set of vectors (functions) that


have the following properties:
1. Linearity. If |i L and |i L then
|i + |i = |i + |i L.
2. There is an inner product for any two elements of space (the
operation):
Z
(, ) h|i = (x)(x)dx

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory
X.Bai
1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

3. Every element has a length or norm |||| defined as


follows:
||||2 = h|i
4. L is complete. Every Cauchy sequence of functions in L
converges to an element of L.
Note: A Cauchy sequence is a sequence whose elements
become arbitrarily close to each other as the sequence
progresses.
14 / 22

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory
X.Bai

That is P
to say, the completeness requires that, if a series of

vectors
k=0 uk converges absolutely in the sense that
P
||u
||
< , then the series converges in L, in the sense
k
k=0
that the partial sums converge to an element of L.

1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

In other words, Hilbert space contains all of its limit points. You
will see how this may affect in the actual applications of quantum
mechanics in a strict (as it has to be) mathematical manner.
The elements of the space L are the ket vectors |i. We call L a
ket-space.
Why is the completeness is needed from physics point of view?
(Look at the example of a particle in a box.)

15 / 22

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory
X.Bai

2.2 Two Definitions: Functional & Operator


Linear functional: , defined on the kets |i L, is a linear
mapping which associates a complex number with every ket |i
(a function!):

1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

|i L complex number (|i)


(1 |1 i + 2 |2 i) = 1 (|1 i) + 2 (|2 i)
Note. Linear functionals and linear operators (see below) must
not be confused. They both deal with linear operations, but
I

A linear function associates each ket with a complex number

A linear operator associates each ket with another ket.

A set of linear functionals defined on the kets |i L constitutes


a vector space, which is called the dual space L of L.

16 / 22

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory
X.Bai

Every element of the dual space L is called a bra-vector or


simply bra. It is symbolized as h|. Thus

1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

(|i) = h|i
is the number obtained by causing the linear functional h| L
to act on the ket |i L. We do require:
I

To every ket corresponds a bra.

The correspondence is anti-linear, i.e.


(1 |1 i + 2 |2 i, |i) = (1 h1 | + 2 h2 |)|i
or
1 |1 i + 2 |2 i

corresponding to

1 h1 | + 2 h2 |

17 / 22

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory
X.Bai

Note: If |i is a ket and is a complex number then |i is


another ket. Sometimes we write

1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

|i = |i
The corresponding bra is (please verify yourself):
h| = h|
The Geometrical Properties
I

Cauchy-Schwartz inequality: |h|i| |||| ||||

Triangle inequality: || + || |||| + ||||

18 / 22

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory

Proof of the Cauchy-Schwartz inequality. We start from the


inequality:
2

|| || = h | i 0

X.Bai
1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

which gives
|| ||2 = h|i h|i h|i + ||2 h|i 0
Let us take =

h|i
||||2 .

Then we have

|| ||2 = ||||2

|h|i|2
0
||||2

||||2 ||||2 |h|i|2


|||||||| |h|i|
Q. E. D.

19 / 22

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory

is a map between two sets of ket


Operators: An operator A
vectors:

| 0 i = A|i

X.Bai
1. Probability Amplitudes
and Their Compositions

We will consider only linear operators such that:

2. Hilbert Space and


Dirac Notations

A(|
1 i + |2 i) = A|1 i + A|2 i

We need to answer this question: We can write A|i


= |Ai.

So, what is hA|?


The quantity
Z

h|A|i = (x)A(x)dx
between |i and |i. This is a
is called a matrix element of A
familiar integral which can be used to define an adjoint operator
as following:
A
Z

Z
Z



(x)A (x)dx = (x) A(x) dx =


(x)A(x)dx
(11)
20 / 22

Rewriting Eq. (11) in Dirac notations we obtain:


|i = h|A|i

h|A
This formula defines an adjoint operator in Dirac notations.
Thus a Hermitian (or self-adjoint) operator is such that (in terms
of elements of the matrix)

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory
X.Bai
1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations


h|A|i
= h|A|i

We now can find the bra-vector corresponding to A|i:


Z

Z

|i
hA|i
= (A(x))
(x)dx =
(x)A(x))dx
= h|A
Thus
= h|A

hA|
and we have an association

A|i
h|A

21 / 22

Section 1: Vector Space,


Matrix Mechanics, and
Representation Theory
X.Bai

Consider an operator:
= |ih|
P

1. Probability Amplitudes
and Their Compositions
2. Hilbert Space and
Dirac Notations

This operator is called a projector or projection operator. Indeed

P|i
= h|i|i

= |i, which
gives a projection of |i onto |i. Also P|i
means that
2 = P

P
Such operators are also called idempotent operators.
Lets take a look at the Hermitian Adjoint again and summarize
a lit bit. (Review01 Hermitian.pdf)

22 / 22

Vous aimerez peut-être aussi