Chapter 5: The General Structure of Wave Mechanics

Dirac Notation
Eigenstates are described by vectors in a linear space. The vectors are called a ket and
denoted by a . The label “a” gives the value of the “observables” for the system. If the
vector represents some other state that may be a superposition of eigenstates, we then
denote this by  . These kets have the properties of vectors. Associate with the space
the space of kets in a space of conjugate states, described by bras and denoted by a .
For each ket we have a bra.

Consider the scalar product of two eigenfunctions of some operator


=  dx u ( x)un ( x) =  mn
*
mn m
bra − ket symbol −

where un  n is called the ket vector and um  n is called the bra vector. This
representation is also called bra-ket or Dirac notation. They satisfies the following
properties



 =  dx  ( x) ( x) 
*

−
   =  
*

  =  dx  ( x) ( x) 
*

− 
and

 1 1 + 1 2 = 1   1 + 1   2
 21 +  22  =  2* 1  +  2* 2 

In Dirac language, the eigenvalue equation is written as

Aˆ n = an n 
  Aˆ  n = an  n
 n →  n 

where Aˆ  = Aˆ . Thus Aˆ  =  Aˆ †  .

1
Postulates of Quantum Mechanics

The basic concepts of quantum theory are formulated by five postulates. We now
summarize the postulates of quantum mechanics that have been introduced and applied in
previous chapters.

Postulate 1:

The dynamical state of a particle can be described by a wave function (a ket vector)
which contains all the information that can be known about the particle.

1
 ( x, t ) =
2
 dp
−
 ( p) e − i ( px − Et )/
wave function wave function
in coordinate space in coordinate space

A quantum mechanical state is defined by a specific set such as  ( x1 , t1 ) or  (p1 ) .

Although neither  ( x1 , t1 ) nor  (p1 ) is directly observable, the square of its amplitude is
proportional to the probability that the particle will be observed in that state.
 ( x1 , t1 )
2

p( x1 , t1 ) = 

 dx  ( x , t )
2
probabilty
density 1 1
−

This is the probabilistic interpretation of the wave function. The probability for finding
the particle somewhere in space that is the total probability is unity
 
P=  dx p( x, t ) =  dx  ( x, t ) ( x, t ) = 1
*

− −

As a result, the wave function must satisfy the condition that finding the particle
somewhere in space is unity and this gives us the normalization. The other conditions on
the wave function that arise from the probabilistic interpretation are that it must be single-
valued, continuous and finite. We normally write wave functions with a normalization
constant included.

Not all functions make a suitable wave functions. In order to be physically acceptable,
both the function and its derivative must be finite, continuous, and single-valued
everywhere. The square integrable proper is given by
2
  ( x, t ) dx = N
where N is a finite number. A function which is normalized to unity may be constructed
by writing as
1
 ( x, t ) =  ( x, t )
N
Definition: Two nonzero wave functions are said to be orthogonal if they satisfy

1 if m = n
− dx  m ( x, t ) n ( x, t ) =  mn = 0 if m  n
*

2
Postulate 2:

If a system is in an eigenstate un of an operator  , any measurement of the operator

 will always yield the eigenvalue an that they satisfy the eigenvalue equation
Aˆ un = an un
Although measurement will always yield a value, the initial state does not have to be an
eigenstate of  . An arbitrary function  ( x ) can be expressed as a linear superposition of
a set of orthonormal (normalized and orthogonal) functions of  .

 ( x) =  cnun or  =  cn un
n n

where cn are the expansion coefficients that are related to the probability for finding the
particle in the un state. Here, n may go to infinity. The probability of observing the
2
eigenvalue an is given by the absolute value of the square of the coefficient cn .

Example:
3 3 i
 ( x) = u1 ( x) + u3 ( x) + u5 ( x)
5 10 10

where u1 ( x) , u3 ( x) and are orthogonal functions.

Postulate 3:

The Schrodinger equation describes the behavior of the wave function in space and time.

i  ( x, t ) = H ( x, t )
t
pˆ 2 2
2
with H = + V ( x) = − + V ( x ) . For time-independent potential, the above
2m 2m x 2
equation is separable in space and time, and  ( x, t ) may be written as the product of a
spatial function and a time function

 n ( x, t ) = un ( x ) e − iEnt /
spatial time − dependent
part part

Spatial part satisfies the energy eigenvalue equation

Hun ( x) = Enun ( x)

3
Although  n ( x, t ) is time-dependent, the probability density p =  n* ( x, t ) n ( x, t ) is
independent of time.

Utilizing the superposition principle, the general solution becomes

 ( x) =  cnun ( x) e−iEnt /
n

where un ( x) is a complete and orthonormal set.

Postulate 4:

Each observable quantity q can be directly associated with a linear, Hermitian operator
Q̂ . The value q is an eigenvalue of the operator.

Qˆ n ( x) = qn n ( x)

This postulate comes from the observation that the expectation value of an operator that
corresponds to an observable must be real and therefore the operator must be Hermitian.

Definition: The operator Q̂ is linear if it satisfies the following property

Qˆ ( +  ) =  Qˆ +  Qˆ
1 2 1 2

Definition: An operator is Hermitian if it satisfies the following property

(Qˆ )* dx =  * (Qˆ )dx
1 2  1 2

In Dirac notation,
 n  m =  ( n )* m dx
and
Qˆ 1  2 =  1 Qˆ †  2
The Hermitian property of an operator thus becomes

Qˆ 1  2 =  1 Qˆ  2 or Qˆ = Qˆ †

4
Example:
d
Prove that pˆ x = is a Hermitian operator.
i dx

 * 
 d  d 1* ( x)
pˆ x 1  2 = −  i dx  1 ( x)   2 ( x)dx = − i 
−
dx
dx
 2 ( x)
u
dv


uv − −  udv
−

 
 
 
  
 
 *  d   d
= −  1 ( x) 2 ( x) −   1 ( x)   2 ( x)  dx  = −  −   1* ( x)   2 ( x)  dx 
*

i −
−  dx   i  −  dx  
 = 0 since xlim
→
 1* ( x ) → 0

 and xlim
→
 2 ( x )→0

 wave function property 
 
 d 
=   1* ( x)   2 ( x)  dx =   1* ( x)  pˆ x 2 ( x)  dx   1 pˆ x  2  pˆ x is Hermitian
−  i dx  −

Example:
ˆ  = Qψ
An oparator Q̂ is Hermitian if it satisfies the condition ψ1|Q ˆ | . If the
2 1 2

momentum operator pˆ x = −i d / dx is Hermitian, show that whether the operator

ˆ  m  x + i pˆ  is Hermitian or not.
A  x
2  m 

 
ˆ m  i  m  i 
Aψ1| 2 =  xψ1| 2 + pˆ x ψ1| 2  =  xψ1| 2 − pˆ x ψ1| 2 
2  m  2  m 
 ψ1|x 2 ψ1|pˆ x 2 
m  i  m  i 
=  ψ1|x 2 − ψ1|pˆ x 2  =  ψ1|x 2 + ψ1| − pˆ x 2 
2  m  2  m 
m   i  
=  ψ1|  x − pˆ x  2 
2   m  
ˆ |  ψ |A
ˆ ˆ
Aψ1 2 1  2  A is not hermitian operator

5
Theorem 1: The eigenvalues of a Hermitian operator are real.

Qˆ n ( x) = qn n ( x)
Qˆ n  n =  n Qˆ  n Hermitian property of Q
qn n qn n

qn n  n =  n qn n   (qn n )* n dx =  n* (qn n )dx

qn*  n  n = qn  n  n  (qn* − qn )  n  n = 0  qn* = qn  qn is real
1

Theorem 2: The eigenfunctions of a Hermitian operator are orthogonal if they

correspond to distinct eigenvalues.

Qˆ  ( x) = q  ( x) 
 where q  q
Qˆ  ( x) = q  ( x) 

Qˆ    =   Qˆ   Hermitian property of Q

q  q 

q    =   q 
q*     = q      (q − q )     = 0      = 0    ⊥  
q 0

  and   are orthogonal each other.

Example:
Let  1 ( x) and  2 ( x) be two eigenfunctions of a Hermitian operator N̂ corresponding
two distinct eigenvalues a1 and a2 . Show that  1 ( x) and  2 ( x) are orthogonal to each
other.

N̂ψ1 = a1ψ1 
 a1  a2
N̂ψ 2 = a2 ψ 2  ˆ | = ψ |N
ˆ
Nψ n n n n
ˆ | = ψ |N ˆ
Nψ 1 2 1 2
(a*
n )
− an ψ n | n = 0  an* = an
a1ψ1| 2 = ψ1|a 2 2 0

(a *
1 )
− a2 ψ1| 2 = 0  ψ1| 2 = 0  ψ1 ⊥  2
0

6
Postulate 5:

The expectation value (the average value of the observable corresponding to Q) of a

measurement of a variable q is given mathematically as
ψ |Q ψ
q =
ψψ
If  is normalized to unity,   = 1 , q =  Q . Each observable quantity q can be
directly associated with a linear, Hermitian operator. The value q is an eigenvalue of the
operator. If   ( x ) is a linear superposition of a set of orthonormal
functions ( x) =  cnun ( x) ,
n

  = c1 + c2 + ....+ =  cn = 1
2 2 2

 n n
2
c q
 Q
2 2
c c
q = = n = 2 1 2 q1 + 2 2 2 q2 + ...
  n + + + +
2
c c1 c2 .... c1 c2 ....
n

= c1 q1 + c2 q2 + ...
2 2

Example: Find the energy eigenvalue of a system that is given by

3 3 i
 ( x) = u1 ( x) + u3 ( x) + u5 ( x)
5 10 10
E =  ( x) H  ( x)
2
3 3 3 3 i
= u1 H u1 + u1 H u3 + u1 H u5
5 5 10 5 10
E1 u1 E u 3 E u
3 5 5

E1 u1 u1 = E1 E3 u1 u3 = 0 E5 u1 u5 = 0
2
3 3 3 3 i
+ u3 H u1 + u3 H u3 + u3 H u5
10 5 10 10 10
E1 u1 E3 u3 E u 5 5

E1 u3 u1 = 0 E3 u3 u3 = E3 E5 u3 u5 = 0
* * 2
 i  3  i  3 i
+  u5 H u1 +   u5 H u3 + u5 H u5
 10  5
E u
 10  10
E u
10 E5 u5
1 1 3 3

E1 u5 u1 = 0 E3 u5 u3 = 0 E5 u5 u5 = E5

3 3 1
E = E1 + E3 + E5
5 10 10

Eigenvalues E1 E3 E5
Probabilities 3/ 5 3 /10 1/10
Total probability = 3 / 5 + 3 /10 + 1/10 = 1

7
Measurements of Compatible Operators

For any operator Q̂ , the combinations Q̂ + Q + , i(Q̂ − Q + ) and Q̂Q + are Hermitian.

Example 1:
Q̂ is a non-Hermitian operator, however, Â =
1
2

1

(Q̂ + Q + ) + i(Q̂ − Q + ) is Hermitian.
2i

Example 2:
1
H = w( Aˆ † Aˆ + ) is harmonic oscillator Hamiltonian. † and  are not Hermitian
2
operators but Aˆ † Aˆ is Hermitian where Nˆ = Aˆ † Aˆ is called the number operator).

If  and B̂ are Hermitian ( Aˆ † = Aˆ and Bˆ † = Bˆ ), the product of these operators gives

ˆ ˆ )† = Bˆ † Aˆ † = BA
( AB ˆ ˆ  BA
ˆ ˆ − AB
ˆ ˆ + AB
ˆ ˆ = AB
ˆ ˆ −  Aˆ , Bˆ 
 
−  Aˆ , Bˆ 

where  Aˆ , Bˆ  = AB ˆ ˆ is called the commutator of  and B̂ . If  Aˆ , Bˆ  = 0 , the

ˆ ˆ − BA
 
product of operators  and B̂ is Hermitian.

ˆ ˆ )† = AB
( AB ˆ ˆ −  Aˆ , Bˆ   ( AB
ˆ ˆ )† = AB
ˆˆ
 
=0

If the commutator vanishes when acting on any wave function, the operators  and B̂
are said to commute,
 Aˆ , Bˆ  = ( AB
ˆ ˆ − BA
ˆ ˆ ) = 0  AB
ˆ ˆ = BA
ˆˆ
 
 
When the operators commute, Â, B̂ = 0 , their observables A and B are said to be
compatible. They are non-compatible if Â, B̂  0 .  
If two observables are compatible, their corresponding operators have the simultaneous
eigenfunctions and A and B are said to be simultaneously measurable. Thus, compatible
observables can be measured simultaneously with arbitrary accuracy, non-compatible
observables cannot.
A ˆ and B ˆ =a ψ and Bψ
ˆ have simultaneous eigenfunctions: Aψ ˆ =b ψ .
 Aˆ , Bˆ  = 0   n n n n n n
  ˆ and B
ˆ have simultaneous eigenfunctions, they commute.
If A

 Aˆ , Bˆ  = 0  Aˆ ( Bˆ n ) = Bˆ ( Aˆ n )
 
an n

Aˆ ( Bˆ n ) = an ( Bˆ n )  Bˆ n is an eigenfunction of Aˆ

8
(B̂ψ )
n is an eigenfunction of  belonging to the eigenvalue a n . Since a n is non
( )
degenerate, B̂ψ n can only differ from ψ n by a multiplicative constant which we can
call b n
B̂ψ n = b n ψ n
Thus we see that ψ n is simultaneously an eigenfunction of the operators  and B̂
belonging to the eigenvalues a n and b n , respectively. If Â, B̂, Ĉ, . . . are a set of
commuting operators, there is a simultaneous eigenfunction ψ n of Â, B̂, Ĉ, . . . with the
eigenvalues a n , b n , c n ,... .

Question: What happens if an operator has degenerate eigenvalues?

Suppose that an operator has the same eigenvalue  corresponding to the eigenkets
u1 and u2 . The wave function becomes a superposition of eigenkets which have the
same eigenvalue  as
 = c1 u1 + c2 u2

Example:

pˆ 2 2
d2
H= =−
2m 2m dx 2
 H , pˆ  = 0  They have simultaneous eigenfunctions
uk ( x) = eikx and u− k ( x) = e −ikx

p̂uk ( x) = kuk ( x) 

pˆ u− k ( x) = − ku− k ( x)   u k ( x) aresimultaneous eigenfunctions of pˆ and H
2 2 
k
Hu k ( x) = u k ( x ) 
2m 

The incompatible quantum observables are correlated with Heisenberg's uncertainty

principle. Heisenberg uncertainty relation says that not all properties of a quantum
particle can be measured with unlimited accuracy. This relations can be written in terms
of commutation relationships between incompatible observables according to two
hermitic operators and their commutator as
1  ˆ ˆ
(A) 2 (B) 2  i A, B 
4 

No actual physical measurements can avoid the limitations given by these relationships.

Example:

xp  comes from the canonical commutation relation  x, p  = i .

2

9
Example:
Using the uncertainty principle, show that the lowest energy of an oscillator is w / 2 .

xp   p  =
2 2x 2a

Total energy as a function of a

(p ) 2 1 ( / 2 a) 2 1
E (a) = + mw2 ( x) 2 = + mw2 a 2
2m 2 2m 2
The minimization of E with respect to a
2
dE (a) 1
=− + mw2 a0 = 0  a0 = / 2mw
da a = a0 2m 2a03
1 2 1
E (a0 ) 2
+ mw2 a02
2m 4a0 2
2
1 1 w w w
= + mw2 / 2mw = +
2m 4 / 2mw 2 4 4 2

 d d 
Example: Evaluate the commutator  x + , x −  .
 dx dx 

 d d  d d  d d 
 x + dx , x − dx  f ( x) =  x, x  f ( x) −  x, dx  f ( x) +  dx , x  f ( x) −  dx , dx  f ( x)
0
 d 0
−  x, 
 dx 

d  d d   d d 
= 2  , x  f ( x) = 2  ( xf ) − x f  = 2 f + x f −x f  = 2 f ( x)
 dx   dx dx   dx dx 
 d d
 x + dx , x − dx  = 2

Prb 8 (Gasiorowicz):
Show that the eigenvalue  of a unitary operator must be the form  = ei .

Let us consider the eigenvalue equation of unitary operator U as

U  ( x) =   ( x)
where  is the eigenvalue of U. An operator is unitary if U †U = 1 . Consider
  
* † 
−        
=     dx  ( x)   ( x)  = 1
=
* *
dx U ( x ) U ( x ) dx ( x ) U U ( x )
−  1  −
 ( x ) U
* †

On the other hand, by using the eigenvalue equation we have

 

 dx U ( x) U ( x)  =  *  dx  ( x)  ( x) = 

* * 2

− −
 * * ( x )  ( x )
1
i
These two results give  = 1   = e where  is a real number.
2

10
Prb 11 (Gasiorowicz):

Prove the following

a) If A and B Hermitian, then i  A, B is also Hermitian.
b)  AB, C = A  B, C +  A, C B .
c)  A  B, C +  B, C, A  + C,  A, B = 0

a. If A and B Hermitian, A† = A and B† = B . We calculate

(i  A, B)† = (iAB − iBA)† = −i( AB)† + i ( BA)† = −iB† A† + iA† B† = −iBA + iAB = i  A, B 
(i  A, B)† = i  A, B   i  A, B is Hermitian

b.  AB, C = ABC − CAB = ABC − ACB + ACB − CAB = A B, C +  A, C B

A( BC −CB ) ( AC −CA ) B
 B ,C   A ,C 

c.
 A  B, C +  B,  C, A  +  C,  A, B =  A( BC − CB )  +  B, (CA − AC )  + C , ( AB − BA) 
= A( BC − CB) − ( BC − CB) A + B (CA − AC ) − (CA − AC ) B + C ( AB − BA) + ( AB − BA) C
= ABC − ACB − BCA + CBA + BCA − BAC − CAB + ACB + CAB − CBA − ABC + BA C
= ( ABC − ABC ) + (− ACB + ACB) + (− BCA + BCA) + (CBA − CBA)
0 0 0 0

+ (− BAC + BA C) + (−CAB + CAB )

0 0

 A  B, C +  B,  C, A  +  C,  A, B = 0

Example:
Evaluate the commutator  px2 , x  .
 px2 , x  =  px px , x  = px  px , x  +  px , x  px = −2i px
p x  p x , x  + p x , x  p x −i −i

where
 px , x  f ( x) = −i
d  d d   d d 
, x  f ( x) = ( −i )  ( xf ) − x f  = ( −i )  f + x f −x f  = −i
 dx   dx dx   dx dx 

Example:
Evaluate the commutator  p x , x 2  .
 p x , x 2  =  p x , xx  = x  p x , x  +  p x , x  x = −2i x
x p x ,x + p x ,x  x −i −i

11
Time Dependence and Classical Limit

The expectation value of an operator may change with time.


At=  dx 
*
( x, t ) Aˆ  ( x, t )   ( x, t ) Aˆ  ( x, t )
−

d d
At=  ( x, t ) Aˆ  ( x, t )
dt dt
  
=  ( x, t ) Aˆ  ( x, t ) +  ( x, t ) Aˆ  ( x, t ) +  ( x, t ) Aˆ  ( x, t )
t t t

1   1 
= i  ( x, t ) Aˆ  ( x, t ) +  ( x, t ) Aˆ  ( x, t ) +  ( x, t ) Aˆ i  ( x, t )
−i t t i t
H ( x ,t ) H ( x ,t )

1  ˆ 1
=− H ( x, t ) Aˆ  ( x, t ) +  ( x, t ) A  ( x, t ) +  ( x, t ) Aˆ H ( x, t )
i t i
 ( x ,t ) H Aˆ  ( x ,t )
† ˆ  ( x ,t )
 ( x ,t ) AH
 ( x ,t ) HAˆ  ( x ,t ) since H † = H

ˆ
=−
1
 ( x, t ) HAˆ − AH ˆ  ( x, t )  ( x, t ) A  ( x, t )
i  H , Aˆ 
t
 

d i Aˆ
A t =  H , Aˆ  +
dt t t t

Aˆ
If  has no explicit time dependence, = 0 , then
t
d i
A t =  H , Aˆ 
dt t

The observable A is a constant of motion if the operator commutes with H.

12
Example:
Consider successively Aˆ = x and Aˆ = pˆ x

d i
xt=  H , x t
dt
 
 
i p 2
i 1  i 1
=  + V(x), x  = 
x
 px2 , x  +  V(x), x   = −2i px
 2m   2m =0  2m
px  px ,x + px ,x  px
 −i −i

d p
xt= x
dt m

d i
pˆ x t
=  H , pˆ x  t
dt
 
i p 2
 i  1  i V
=  + V(x),x
ˆ
p x =   px2 , pˆ x  +  V(x), pˆ x  = i
 2m   2m  x
V
 
0 i
x

d dV
pˆ x t = −
dt dx

d p d2 d d2 dV
x t = x m 2 x = px  m 2 x t = −
dt m dt dt dt dx t
dV
−
dx

This looks very much like the equation of a classical particle in V(x) as
d2 dV d 2 xcl dV ( xcl )
m 2 x t =− →m 2 =− "Ehrenfest Theorem"
dt dx t dt dxcl
dV d
where xcl = x and  V ( x ).
dx d x

Consider Taylor expansion of a function F(x) around x as

1
F ( x) = F ( x ) + ( x − x ) F '( x ) +
( x − x ) 2 F ''( x ) + ...
2!
If the uncertainty ( x) = ( x − x ) is small, the higher order terms in the expansion can
2 2

be neglected,
F ( x)  F ( x ) + ( x − x ) F '( x ) F( x )
For macroscopic fields, this is a good approximation and allows us to describe e- or p
orbits by means of classical equation of motion.

13
Prb 17 (Gasiorowicz):

An electron in an oscillating electric field is described by the Hamiltonian operator

p2
H= − (eE 0 cos wt ) x
2m
Calculate expressions for the time dependence of x , p and H .

d i Aˆ
A t =  H , Aˆ  +
dt t t t

d i x
xt= H , x t +
dt t t
0

 
 
i p 2
 i 1 
=  2m − (eE 0 cos wt ) x, x  =  2m  p 2 , x  − (eE 0 cos wt )  x, x  
   
p p ,x + p ,x  p 0
 −i −i

d p
xt=
dt m

d i
pˆ x t
=  H , pˆ x  t
dt
 
i  p2  i 1  i V
=  2m − (eE 0 cos wt ) x, pˆ x  =  2m  p , pˆ  + −(eE 0 cos wt )  x, pˆ  = i
2

    x
0 i

d
pˆ x t
= eE 0 cos wt
dt

d i H   p2 
H = H , H  + =  − (eE 0 cos wt ) x 
dt t
t t t  2m  t
0 t


= 0 − eE 0 x cos wt = 0 + eE 0 xw sin wt
t
= eE 0 w sin wt x

14
Prb 18 (Gasiorowicz):

Solve the equations of motion obtained in Prb 17.

For momentum operator,

pˆ x t t
d
dt
pˆ x t
= eE 0 cos wt  
pˆ x
d pˆ x t
= eE 0  dt cos wt
0
0
1 1
pˆ x t − pˆ x 0
t
sin wt 0 = sin wt
w w

eE 0
pˆ x t
= pˆ x 0
+ sin wt
w

For coordinate,

d p 1 eE 
xt= =  pˆ 0 + 0 sin wt 
dt m t m w 
x t t t
1 eE 0 1 eE 1
x
 d xt=
m
pˆ 0 
0
dt +
mw  dtsin wt
0
=
m
pˆ 0 t − 0 (cos wt − 1)
mw w
0
t 1 1
x t− x 0
t
− cos wt 0 =− (cos wt − cos 0)
w w

1 eE
x t = x 0+ pˆ 0 t − 02 (cos wt − 1)
m mw

15