You are on page 1of 15

LECTURE NOTES 3

Formalism

I.

HILBERT SPACE

A. LINEAR VECTOR SPACE


A linear vector space over the field C of complex number , , ... is an abstract set of elements (vectors), denoted by a ket |a >, with the following properties (in what follows means for all, - exists, - in, or belongs):
1. |a > and |b > H

exists a rule for forming the sum (|a > +|b >) H

2. |a > and |b > H

(|a > +|b >) = (|b > +|a >) (commutative law).

3. |a > , |b > and |c > H


4.

a null vector |0 > H

(|a > +|b >) + |c >= |a > + (|b > +|c >) (associative law).
with the property |a > +|0 >= |a >.

5. |a > H
a negative vector (also called the inverse vector) | a > H
|a > +| a >= |0 >.
6. and C

with the property

the following takes place:


(|a > +|b >) = |a > +|b >
( + )|a >= |a > +|a >
(.)|a >= (|a >)
0|a >= |0 >
1|a >= |a >
(1)|a >= | a >

Example of a finite vector space


All polynomials of degree < N on the interval 1 x 1, i.e. all functions of the following type:
fN (x) = a0 + a1 x + a2 x2 + ... + aN 1 xN 1 |fN (x) >
These polynomials form a linear vector space

2
B. HILBERT SPACE
A Hilbert space H is a linear vector space over the field C of complex number , , ... with an inner (scalar)
product for every ordered pair of vectors. In other words, it is an abstract set of elements (vectors), denoted by a
ket |a >, with the properties (1)-(6) plus a definition of an inner (scalar) product for every ordered pair of vectors.
7. A scalar (inner) product {|a >, |b >} C is defined in H with the following properties:

{|a >, |b >} = ({|b >, |a >})


{|a >, |b > +|c >} = {|a >, |b >} + {|a >, |c >}

or

{|a > +|b >, |c >} = {|a >, |c >} + {|b >, |c >}

{|a >, |b >} = {|a >, |b >}

or

{|a >, |b >} = {|a >, |b >}

Example of a finite vector space with an inner product


All polynomials of degree < N on the interval 1 x 1, i.e. all functions of the following type:
fN (x) = a0 + a1 x + a2 x2 + ... + aN 1 xN 1 |fN (x) >
These polynomials form a linear vector space with inner product defined as follows:
Z 1

{|fN (x) >, |gN (x) >} =


fN
(x)gN (x)dx
1

If, for example,fN (x) = 2 + x2 , and gN (x) = 5xN , then


Z 1
Z
2
N
{|fN (x) >, |gN (x) >} =
(2 + x )(5x )dx = 10
1

1
N

x dx + 5

xN +2 dx

If N is even, we have
{|fN (x) >, |gN (x) >} =

10
20
+
.
N +1 N +2

If N is odd, we have
{|fN (x) >, |gN (x) >} = 0.
8. Two vectors, |a > and |b >, are orthogonal if {|a >, |b >} = 0.
Example of orthogonal vectors: If N is odd the above two vectors, fN (x) = 2 + x2 , and gN (x) = 5xN ,
are orthogonal.
9a. A norm ||a|| < (< is a field of real numbers) of a vector |a > is defined by
p
||a|| = {|a >, |a >}
Example: norm of vector fN (x) = 2 + x2
sZ

||fN (x)|| =

(2 + x2 )2 dx =

9b. The norm satisfies:


- the Schwatrz inequality
||a||

||b||

| {|a >, |b >} |

166
15

3
Example: Let fN (x) = 2 + x2 and g(x) := 5x8 . Norm of fN (x) = 2 + x2 is
sZ
r
1
166
2
2
||fN (x)|| =
(2 + x ) dx =
15
1
Norm of gN (x) is
sZ

||gN (x)|| =

(5x8 )2 dx

=5

Absolute value of |fN (x) > |gN (x) > is


R1
(f (x).gN (x)dx 3.1.
1 N

310
99 .

2
17

Obviously, ||fN (x)||.||gN (x)||

5.7

>

|fN (x).gN (x)|

- the triangle inequality:


|||a > +|b > ||

||a|| + ||b||

with the equality only being valid when |a >= |b >, and > 0.
For our above two vectors we calculate:
||fN (x) + gN (x)|| = 4.5

||fN (x)|| + ||gN (x)|| 5.0

10. A vector |a > is called a normalized vector if ||a|| = 1.

C. DISCRETE SPACE OF DIMENSION N


The term Hilbert space H usually is reserved for the infinite-dimensional space. In this Section, we discuss
a N -dimensional (finite-dimensional) linear space VN with a scalar product (which also can be called a Hilbert N dimensional space).
Any set of linearly independent vectors |e1 >, |e2 >, ..., |eN > is called a basis in VN . Recall that |e1 >, |e2 >
, ..., |eN > are linearly independent vectors if
N
X

i |ei >= |0 >

(1)

i=1

is possible only when all i = 0. The basis is called orthonormal basis if


{|ei >, |ej >} = ij

(2)

Any basis can be transform into orthonormal by means of the Gram-Schmidt procedure. Let |E1 >, |E2 >
, ..., |EN > is a basis, but not orthonormal. The new orthonormal basis will be denoted by |e1 >, |e2 >, ..., |eN >. The
first vector from the new orthonormal basis is
|e1 >=

1
|E1 >,
||E1 ||

{|e1 >, |e1 >} = 1

To obtain |e2 >, we first construct the vector


|e02 >= |E2 > {|e1 >, |E2 >}|e1 >
This vector satisfies the condition:
{|e01 >, |e02 >} = 0

4
Thus, the second orhonormal vector is:
|e2 >=

1
|e0 >
||e02 || 2

Similarly,
|e03 >= |E3 > {|e1 >, |E3 >}|e1 > {|e2 >, |E3 >}|e2 >,

|e3 >=

1
|e0 >
||e03 || 3

The same procedure can be used to find |e4 >,|e5 >,..., and |eN >.
Supposed that we have chosen a given orthonormal basis |e1 >, |e2 >, ..., |eN >. Then, any vector |a > from VN
is uniquely defined by the equation:
|a >=

N
X

ai |ei >,

(3)

i=1

where the complex numbers a1 , a2 , ..., an , called the components of the vector |a > in the given basis |e1 >
, |e2 >, ..., |eN >, are defined by the equation:
ai = {|ei >, |a >}

(4)

We now postulate that to each ket-vector |a > in the ket space, there exists a corresponding quantity
called a bra-vector, denoted by < a| in a dual space V 0 . There is one-to-one correspondence between
bra- and ket-vectors
|a >< a|

(5)

It is important that the bra-vector corresponding to |a > is postulated to be


|a > < a|

(6)

Assuming a one-to-one correspondence between bra- and ket-basis


|ei >< ei |,

i = 1, ..., N

the components of < a| in the bra-basis < e1 |, < e2 |, ..., < eN | are a1 , a2 , ..., aN , i.e.
< a| =

N
X

ai < ei |

(7)

i=1

Let us calculate the inner product between two ket-vectors |a >=


{|a >, |b >} =

N
X

PN

ai bj {|ei >, |ej >} =

i,j=1

i=1

N
X

ai |ei > and |b >=


ai bj ij =

i,j=1

N
X

PN

i=1 bi |ei

>:

ai bi

By means of the bra-vectors we redefine the inner product {|a >, |b >} between two ket-vectors |a >=
PN
and |b >= i=1 bi |ei > in terms of product between a bra-vector < a| and a ket-vector |b >:

{|a >, |b >} =< a|b >=

N
X
i,j=1

ai bj < ei |ej >=

N
X
i,j=1

(8)

i=1

ai bj ij =

N
X

ai bi

PN

i=1

ai |ei >

(9)

i=1

We can also introduce one-to-one correspondence between matrices and ket- and bra-vectors in a given orthonormal
basis. The correspondence is as follows: any ket-vector |a > from the discrete space VN can be represented by a
N 1 (one-column) matrix

a1
a
|a > 2
.
aN

5
any bra-vector from the discrete space V0N can be represented by 1 N matrix (one-row matrix):
< a| (a1

a2 ...aN )

There is one-to-one correspondence between the inner product {|a >, |b >} and the product between two matrices:

b1
N
X

b2
{|a >, |b >} = (a1 a2 ...aN )
=
ai bi
.
i=1
bN
D. EXAMPLE FOR A FINITE-DIMENSIONAL VECTOR SPACE
The vectors in this example are defined as all polynomials of degree < N on the interval 1 x 1, i.e. all functions
of the following type:
fN (x) = a0 + a1 x + a2 x2 + ... + aN 1 xN 1 |fN (x) >
These polynomials form a linear vector space (one can check that the requirements (1) to (6) are fulfilled). The inner
product of two vectors, |fN (x) > and |gN (x) >, is defined as follows:
Z 1

{|fN (x) >, |gN (x) >} =


fN
(x)gN (x)dx
1

The vectors (polynomials)


Y0 (x) = x0 = 1,

Y1 (x) = x,

Y2 (x) = x2 , ...,

YN 1 (x) = xN 1

from this linear


R 1 space form a linear independent set of N vectors, and therefore, they can be used as a basis in this vector
space. Since 1 Yi (x)Yj (x)dx 6= ij , the linearly independent basis functions Yi (x) do not form an orthonormal basis.
Applying the Gram-Schmidt procedure, we can obtain the following orthonormal basis yi (x) (Legendre polynomials):
r
r
r
r
1
3
5
7
2
, y1 (x) =
x, y2 (x) =
(3x 1), y3 (x) =
(5x3 3x),
y0 (x) =
2
2
8
8
r
y4 (x) =

9
(35x4 30x2 + 3), ...,
128

yi (x)yj (x)dx = ij
1

In this orthonormal basis, any ket vector of the space is defined by providing N coefficients f0 , f1 , ..., fN 1 , i.e.
PN 1
fN (x) = i=0 fi yi (x). We establish one-to-one correspondence between the ket vectors of the linear space and the
N 1 matrices:

f0
f
|fN (x) > 1
.
fN 1
If gN (x) =

PN 1
i=0

gi yi (x) is another ket vector from the same space

g0
g
|gN (x) > 1
.
gN 1

then:
(i) |fN (x) > +|gN (x) >= |hN (x) > is a vector

f0 + g0
f1 + g1

|hN (x) >

.
fN 1 + gN 1

6
from the same space;
(ii) the scalar product of two kets fN (x) and gN (x) is defined by
{|fN (x) >, |gN (x) >} =

N
1
X

fi gi

i=0

As we postulated above, there must be a one-to-one correspondence between bra- and ket-vectors. This requirement
can be fulfilled using the following correspondence: (i) since the Legendre polynomials are real functions, the bra
basis < yi (x)| is formed by the same Legendre polynomials; (ii) to each ket-vector |fN (x) > in the ket space, there
exists a corresponding bra-vector, denoted by < fN (x)| in a dual space V 0 :

< fN (x)| (f0 , f1 , ..., fN


1 )

We establish a one-to-one correspondence between the bra vectors and the 1 N matrices. Thus, we can rewrite the
inner product between two kets as inner product between bet and bra vectors:

g0
N 1
g1 X

=
{|fN (x) >, |gN (x) >} =< fN (x)|gN (x) >= (f0 , f1 , ..., fN
fi gi
)

1
.
i=0
gN 1

E. INFINITE-DIMENSIONAL SPACE
The above equations can be generalized for the case when the space is not a finite-dimensional space, but an infinitedimensional space. As an example, we consider the infinite-dimensional space of all functions f (x), g(x), h(x), ... on the
interval a x b. These functions are vectors from a linear vector space (statements (1) -(6) take place). Instead of a
basis of N linearly independent vectors, one has to use a complete set of vectors |Y1 (x) >, |Y2 (x) >, ..., |Yi (x) >, ...,
and by definition, if the set is complete, then an arbitrary vector |f (x) > can be expanded in terms of it:
|f (x) >=

ai |Yi (x) >

or f (x) =

i=0

ai Yi (x)

(10)

i=0

Applying the Gram-Schmidt orthogonalisation method, we can obtain a complete set of orthonormal vectors
|y1 (x) >, |y2 (x) >, ..., |yi (x) >, ... such that
Z b
yi (x)yj (x)dx = ij
a

Using this complete set, we have for any kets:


|f (x) >=

fi |yi (x) >

or f (x) =

i=0

Z
fi yi (x),

yi (x)f (x)dx

(11)

i=0

We define the inner product of two kets, |f (x) >=

fi =

fi |yi (x) > and |g(x) >=

{|f (x) >, |g(x) >} =

i gi |yi (x)

>, by the equation:

fi gi

i=0

The dual infinite-dimensional space is a space of all functions f (x), g (x), h (x), ...:
Z b

X
X
< f (x)| =
fi < yi (x)| or f (x) =
fi yi (x), fi =
yi (x)f (x)dx
i=0

(12)

i=0

Note, that we introduced a complete vector space and a complete set of vectors. A complete vector space
means that linear combinations like (10) which involves an infinite number of terms belong to the vector space. The
last requires that < f (x)|f (x) > is a finite:
{|f (x) >, |f (x) >} =< f (x)|f (x) >=

X
i=0

|fi |2 <

(13)

7
II.

LINEAR OPERATORS ON HILBERT SPACE

b on Hilbert space H is an instruction for transforming |a > H into another vector |b > H:
An operator A
b >= |b >
A|a
In other words, a linear operator induces a mapping of H onto itself (or onto a subspace of H).
For linear operators:
b
b > + A|b
b >
A(|a
> +|b >) = A|a
The following definitions are valid |a > H:
b and B,
b are equal, A
b=B
b if
1. Two operators, A
b >= B|a
b >
A|a
2. Sum of two operators:
b + B)|a
b
b > +B|a
b >
(A
>= A|a
3. Product of two operators:
b B)|a
b
b B|a
b >)
(A.
>= A(
4. Zero operator, and unit operator:
b >= |0 >
O|a
5. If f (x) =

b
1|a >= |a >

an xn then:
b =
f (A)

bn
an A

6. If
b >= |a >
A|a
b with eigenvalue . The set of all eigenvalues of A
b is called the spectrum,
then |a > is an eigenvector of the operator A
which can be discrete n or continuous values (or both).
b acts on ket-vectors from the left. Now, we introduce an operator A
b
7. We wrote earlier that an operator A
b
named the adjoint of A, which acts in the dual space of the bra-vectors, and acts on any bra-vector from the right:
b > (< a|)A
b
A|a
b
8.
 The last equation states that if |a > and | < a| are the corresponding ket- and bra-vector, then A|a > and
b also should be corresponding ket- and bra-vector.
< a|A
It can be proved that:
 
b = A
b ;
9a. A


b+B
b =A
b + B
b;
9b. A


bB
b =B
bA
b ;
9c. A
 
b = A
b
9d. A
b = A
b then the operator A
b is called Hermitian operator.
10. If A
b = A
b we have:
For any Hermitian operator A
b >=< g|A|f
b >
< f |A|g

(14)

b >,
|G >= A|g

b >=
< f |A|g

fi Gi



X
b > = < g|A
b |f > =< G|f > ==
< g|A|f
fi Gi
i

b=A
b , then all eigenvalues n are real.
11a. If A
b
b , the eigenvectors, |a > and |b >, corresponding to the different eigenvalues and , are orthogonal,
11b. If A = A
b
b >= |b >, and 6= , then < a|b >= 0.
i.e. if A|a >= |a >, A|b
b is called unitary operator if
12. The operator U
bU
b = U
b U
b =b
U
1
13. Commutator
b B]
b =A
bB
bB
bA
b
[A,
14. Properties of commutators:
b B]
b = [B,
b A]
b
14a. [A,
b
b
b
b B]
b + [A,
b C]
b
14b. [A, B + C] = [A,
b
b
b
b
b
b
b
b
b
14c. [A, B.C] = [A, B].C + B.[A, C]
b [B,
b C]]
b + [B,
b [C,
b A]]
b + [C,
b [A,
b B]]
b =O
b
14d. [A,

b
b
b
b
14e. [A, B] = [B , A ]
15. Outer product:
We define the so-called outer product |f >< g| between ket |f > and bra < g| vectors. The result of the outer
product is not a complex number, but an operator. By definition, the outer product |f >< g| is an operator: when
acting on a ket-vector |h >, it gives another ket-vector |f > times some number < g|h >:
|f >< g|(|h >) =< g|h > |f >

16. The following takes place:

(|f >< g|) = |g >< f |

which follows from the relation |f >< g||h >< h| (|f >< g|) .
17. Projection operator Fb = |f >< f | projects any vector |g > into its component along |f >:
X
X
X
Fb|g >= |f >< f |
gi |ei >=
gi < f |ei > |f >
gi fi |f >
i

18. The unit operators can be written as


b
1=

|ei >< ei |

This very important and useful property is called the completeness relation or closure. The proof is simple:
X
X
X
b
1|f >= (
|ei >< ei |)|f >=
< ei |f > |ei >=
fi |ei >= |f >
i

b is a Hermitian operator, |n > and n are the corresponding eigenvectors and eigenvalues, i.e.
19. Let A
b
A|n >= n |n >. It can be proved that:

9
(i) the eigenvalues are real and form a discrete spectrum, and
(ii) its eigenvectors form a complete orthonormal system. In other words, its eigenvectors can be used as an
orthonormal basis in the Hilbert space.
Thus, any |f > H can be written in terms of the eigenvectors |n > of a Hermitian operator:
X
X
|f >= b
1|f >= (
|n >< n|)|f >=
< n|f > |n >
n

(15)

The set of complex numbers fn =< n|f > are called the A-representation of the vector |f >.
b
21. A-representation of the operator B:
b
Suppose we have some operator B. We can write:
!
!
X
X
X
0
0
b
b
b
b 0 > |n >< n0 |Bnn0 =< n|B|n
b 0>
b
b
B = 1B 1 =
|n >< n| B
|n >< n | =
< n|B|n
n0

(16)

n,n0

b there are N 2 terms in the sum, and hence N 2 numbers < n|B|n
b 0 >. We
If there are N eigenvectors of the operator A,
can place these numbers in a N N matrix, where the matrix elements are arranged by columns and rows according
b 0 >= Bnn0 . Thus, we can write for any operator B:
b
to < n|B|n
X
b=
b 0>
B
Bnn0 |n >< n0 |; Bnn0 =< n|B|n
(17)
n,n0

b in the basis |n >. Note that that


The matrix Bnn0 formed in this way is said to be a representation of the operator B
b has been chosen to determine the basis for that representation.
the matrix representation depends on which operator A

b
Similarly, the operator B can be represented by the matrix (Bnn0 ) :
X
b =
B
Bn 0 n |n >< n0 |
(18)
n,n0

We can prove that




bC
b
B


nn0

Bnn00 Cn00 n0

(19)

n00

b
22. Trace of B:
  X
b =
Tr B
Bnn
n

The trace is independent of the particular orthonormal basis |n > that is chosen for its evaluation.
b and B
b are Hermitian operators, and if [A,
b B]
b = 0, then there exists a complete set of eigenvectors
23. If A
b and B.
b
which are eigenvectors for both A
b If the system is in state | >, then the corresponding
24. Suppose, we two Hermitian operators, Fb and G.
b >, respectively. We introduce the following vectors:
expectations values are f =< |Fb| > and g =< |G|
|f >= (Fb f )| >,

b g)| >
|g >= (G

Let us calculate < f |f >:


 


< f |f >= < |(Fb f ) | (Fb f )| >
The state | > in F -representation is:
| >=

X
n

n |n >

10
where Fb|n >= fn |n >. Thus, we calculate:

 
 X
2
< f |f >= < |(Fb f ) | (Fb f )| > =
|n |2 fn f
n

p
The last result shows that < f |f > is the standard deviation of the observable f :
X
2
f2 =< f |f >=
|n |2 fn f
n

Similarly, the standard deviation of the observable g is:


g2 =< g|g >=

|n |2 (gn g)

b n >= gn |n >.
where G|
According to the Schwarz inequality:
f2 g2 =< f |f >< g|g > | < f |g > |2
Since, for any z = a + b we have |z|2 Imz = (z z )/2, one can write:

2
1
(< f |g > < g|f >)
f2 g2
2
But
b g)| >=< |FbG|
b > f .g < | >
< f |g >=< |(Fb f )(G
b g)(Fb f )| >=< |G
b Fb| > f .g < | >
< g|f >=< |(G
and therefore,
s
f g

2
1
b
b
< |[F , G]| >
2

The last inequality is called the uncertainty principle in its most general form.

III.

EXAMPLES OF LINEAR VECTOR SPACES AND OPERATORS ON IT

Example A. Space of functions, position and momentum operators


1. The differentiable functions F (x, y, z), G(x, y, z),.... , which are square integrable:
Z +
|F (x, y, z)|2 dxdydz <

behave as vectors, i.e. they are vectors from a linear vector space, in which the vectors are functions of x, y and z.
We define the inner product by the equation:
Z + Z + Z +
Z +
< F |G >=
F (x, y, z)G(x, y, z)dxdydz =
d3 rF (r)G(r)
(20)

2. The position operator b


r is one of the operators used in quantum mechanics. By definition, the position operator
acts on a given function F (x, y, z) F (r) simply multiplying that function by corresponding coordinate:
b
rj F (x, y, z) = xj F (x, y, z);

j = 1, 2, 3 or

rj = xj

x1 = x, x2 = y, x3 = z

(21)

11
The position operator b
r possesses eigenvalues < xj < + lying in the continuous range. The ket-vector
corresponding to the eigenvalue xj will be denoted by |x, y, z > (or shortly |r >):
b
rj |x, y, z >= xj |x, y, z >,

xj = x, y, z

(22)

d3 r|r >< r|,

(23)

Those vectors are orthonormal:


0

< r|r >= (r r ),

b
1=

where (r) = (x)(y)(z), and (x) is the Dirac delta function


Z +
d3 rF (r)(r r0 ) = F (r0 )

Here are some useful equations, related to the Dirac function:



(x) =

0 x 6= 0
x=0

(x2 x20 ) =


F (x)(x x0 ) =

,
a

F (x0 ),
x0 [a, b]
0, x0 < a, or x0 > b

1
[(x x0 ) + (x + x0 )] ,
|2x0 |

G(x)(F (x) x0 )dx =

(ax) =

1
(x)
|a|

G(x0 )
| dFdx(x) | x=x0

The position representation of any vector |F > from the vector space of functions is:
Z
Z
|F >= d3 r|r >< r|F >= d3 rF (r)|r >

(24)

By means of the equations (22) and (23), one can prove that the matrix elements of the above operators are:
< r0 |b
rj |r >= xj (r r0 ),

(25)

b acts on a given function F (x, y, z) F (r) according to the equation:


3. By definition, the momentum operator p
b j F (x, y, z) =
p

~
F (x, y, z);
xj

j = 1, 2, 3 or

rj = xj

x1 = x, x2 = y, x3 = z

(26)

In the one-dimensional case the momentum operator is


px =

~
x

defined on the space of differentiable functions F (x) for a x b. If the momentum operator is a Hermitian operator,
then according to eq. (14) we have:
Z
~ b
~
dG(x)
~ d
|G >=
F (x)
dx = F (x)G(x)|ba
< F |px |G >=< F |
x
a
dx

Z b
~
dF (x)
~
~ d
(27)
+
G (x)
dx = F (x)G(x)|ba + < G|
|F >
a
dx

x
~
= F (x)G(x)|ba + < G|px |F >

Thus, the momentum operator is a Hermitian operator if only the boundary conditions are imposed, i.e. the term
F (x)G(x)|ba = 0

(28)

Performing exactly the same considerations as in the case of position operator, we can introduce the ket-eigenvectors
b:
|p > of the momentum operator p
b j |p >= pj |p >,
p

j = x, y, z

(29)

12
Those ket-eigenvectors are orthogonal and normalized:
< p|p0 >= (p p0 ),

Z
b
1=

d3 p|p >< p|

(30)

The momentum representation of any ket-vector |F > is:


Z

Z
3
d p|p >< p| |F >= d3 pF (p)|p >;
|F >=

(31)

where F (p) =< p|F > is the Fourier transform of F (r) =< r|F >, defined as follows:
Z
Z
 p.r 
 p.r 
d3 r
d3 p
F
(p)
exp

,
F
(p)
=
F
(r)
exp

F (r) =
~
~
(2~)3/2
(2~)3/2

(32)

The fundamental assumption in physics is that there exists a particular symmetry between coordinates and the
corresponding momenta - all the coordinates can be transformed into their momenta, and vice versa. In terms of
Dirac notations, this can be expressed by the following relationships:
 p.r 
 p.r 
1
1
< r|p >=
exp

exp

,
<
p|r
>=
(33)
~
~
(2~)3/2
(2~)3/2
By means of the equations (33 ), one can prove that the matrix elements of the position and momentum operators
are:
< r0 |b
rj |r >= xj (r r0 ),

< p0 |b
rj |p >=

~ (p p0 )

p0j

(34)

< r0 |b
pj |r >=

~ (r r0 )

x0j

(35)

< p0 |b
pj |p >= pj (p p0 ),

IV.

THE POSTULATES OF THE NON-RELATIVISTIC QUANTUM MECHANICS

Postulate I: The state of the system is represented by a state vector |(t) > in a Hilbert space. The state vector
contains all needed information about the system. Any superposition of state vectors ia also a state vector.
According to this postulate, the state of the system is described by a ket |(t) > in a Hilbert space. If we
choose a position representation, the state of the system at any instant of time may be represented by a state (or
wave function) (r, t) =< r| >. All information regarding the state of the system is contained in the wave function.
Example Suppose, we have three operators in the Hilbert space, defined by their matrices:

0
1
0
1
Lx = 1 0 1 ,
2 0 1 0

0
1
Ly = 0
2 0 0

1
0
0
1
Lz = 0 0 0
2 0 0 1

(36)

By calculating the corresponding conjugate and transpose matrices we find that all of the above matrices are
Hermitian ones. The corresponding eigenvalues and eigenvectors are as follows:
For Lx :

1 0
det 1 1 = 0
0 1

1
1
1 = 1 |X1 >=
2 ,
2
1

1
1
2 = 1 |X2 >=
2 ,
2
1

3 = 0

1 1
0
|X3 >=
2
1

(37)

13

For Ly :

0
det = 0
0

1
1
1 = 1 |Y1 >=
2 ,
2
1

2 = 1

1
1
|Y2 >=
2 ,
2
1


11
0
3 = 0 |Y2 >=
2 1

(38)

For Lz :

1 0
0

0 =0
det 0
0
0 1


1
0
1 = 1 |Z1 >= 0 , 2 = 1 |Z2 >= 0 ,
0
1

0
|Z3 >= 1
0

3 = 0

(39)

To find the corresponding bra vectors, one has to conjugate and transpose the above ket vectors. For example, the
bra < Y1 | is

1


1
1
< Y1 | =
=
1 2 1
2
2
2
1
As we mentioned, eigenvectors of an Hermitian operator form a basis. This means, that one can use the eigenvectors
b x in the basis
|Zn >, for example, as a basis. Let us find the normalized eigenvectors |Xn > and eigenvalues of L
|Zn >. The eigenvector |Xi > could be expended in the basis |Zn >:
!
3
3
X
X
b
Xin |Zn >,
(40)
|Xi >= 1|Xi >=
|Zn >< Zn | |Xi >=
n=1

n=1

where

1
1 2
1
1
2
1
1

= 1
,
X
=
2
1
1 1 2

2
2
2 2 0
2 2 0

Xin =< Zn |Xi >

i, n = 1, 2, 3

b = Xin
X

(41)

b = X
bX
b = 1) and it transforms the eigenvectors |Zn > into the
b = Xin is a unitary matrix (X
b X
The matrix X
b matrix transforms |Xi > into |Zn >.
eigenvectors |Xi >. The X
Postulate II: To any observable f (such as position, linear momentum, energy, angular momentum, or number of particles) in physics, there exists a corresponding Hermitian operator Fb such that:
(i) the measurement of f yields values (call these measured values of f ) which are the eigenvalues f1 , f2 , ..., fn , ...
of Fb, i.e. Fb|n >= fn |n >, where |n > is the eigenvector which corresponds to the eigenvalue fn ;
(ii) if the system is in a state | >, a measurement of the observable f will yield the value fn with probability
| < n | > |2 , and therefore, the average (or the expectation value) of f is f =< |Fb| >.
Example. Suppose that the system is in the state in which the outcome of the measurement of the observable Lz
is 2 = 1. This means that the state vector is:

1
|(t) >= |Z2 >= 0
(42)
0

14
Let us calculate Lx , L2x and Lx .
bx :
First, we have to expand the state vector |Z2 > in terms of the eigenvectors of L
|Z2 >=

3
X

< Xn |Z2 > |Xn >=

n=1

1
1
1
|X1 > + |X2 > |X3 >
2
2
2

(43)

In this state, if we measure the observable Lx , the outcomes are as follows:


Lx = 1 = 1

with

probability P1Lx = | < X1 |Z2 > |2 =

1
4

Lx = 2 = 1

with

probability P2Lx = | < X2 |Z2 > |2 =

1
4

Lx = 3 = 0

with

probability P3Lx = | < X3 |Z2 > |2 =

1
2

Thus, the average of the observable Lx is:


Lx =

3
X

PnLx n =

n=1

1
1
1
(1) + (1) + (0) = 0
4
4
2

c2 , L
c2 must have common eigenvectors |Xn >, where
cx ] = 0, and therefore, L
cx and L
Let us calculate L2x . Note that [L
x
x
c2 are 2 . Thus, L2 is calculated to be:
n = 1, 2, 3. The eigenvalues of L
n
x
x
L2x

3
X

PnLx 2n =

n=1

1
1
1
1
(1)2 + (1)2 + (0)2 =
4
4
2
2

Finally, we find for the standard deviation:


v
r
u 3
uX
2
1
1
1
1
Lx
t
(1 0)2 + (1 0)2 + (0 0)2 =
Pn n Lx =
Lx =
4
4
2
2
n=1
Example. Suppose the system is in a state with Lz = 1, which means the state vector is |Z1 > instead of |Z2 >
bx :
in the previous example. Now, we expand the state vector |Z1 > in terms of the eigenvectors of L
|Z1 >=

3
X

< Xn |Z1 > |Xn >=

n=1

1
1
1
|X1 > + |X2 > + |X3 >
2
2
2

(44)

In this state, if we measure the observable Lx , the outcomes are as follows:


Lx = 1 = 1

with

probability P1Lx = | < X1 |Z1 > |2 =

1
4

Lx = 2 = 1

with

probability P2Lx = | < X2 |Z1 > |2 =

1
4

Lx = 3 = 0

with

probability P3Lx = | < X3 |Z1 > |2 =

1
2

Postulate III: A measurement of the observable f that yields the value fn leaves the system in the state
n , i.e. the measurement changes the state of the system. In other words, if the state of the system is |(t) > and the

15
result of the measurement on state |(t) > is fn , the state of the system immediately after the measurement is
given by a projection of |(t) > onto the eigenvector |n > corresponding to fn :
| >af ter = Pb|n > |(t) >=< n |(t) > |n >
Here, Pb|n > = |n >< n | is the projection operator that projects any vector onto its component along |n >.
Example Suppose that the state of the system in |Zn > basis is:

1/2
= 1 |Z1 > + 1 |Z2 > + 1 |Z3 >
| >= 1/2

2
2
2
1/ 2

(45)

Then, if L2z is measured and the result is +1, what is the state immediately after the measurement?
b z and L
b 2z have common eigenvectors |Zn >, where n = 1, 2, 3. The eigenvalues an of L
b 2z are an = 2n :
First, L
a1 = 1,

a2 = 1,

a3 = 0

In the case of degeneracy, we have a doublet of orthonormal vectors |Z1 > and |Z2 > with the same eigenvalue
a1 = a2 = 1. The probability to measure a1 = 1 is:
| < Z1 | > |2 =

1
2

| < Z2 | > |2 =

1
4

The probability to measure a2 = 1 is:

After the measurements, the system could be in a state |Z1 > with a probability 2/3 and in a state |Z2 > with a
probability 1/3.
Postulate IV: The state vector |(t) > obeys the Schrodinger equation:
~
b is the Hamiltonian operator.
where H

d
b
|(t) >= H|(t)
>,
dt