You are on page 1of 42

Chapter 2

The Harmonic Oscillator

.
The Classical H.O:
It is well-known that any particle executes S.H.O if it is acted upon be a
restoring force of the form
F  x   kx (1) k is the force constant
The potential energy of such a problem is given by
 
V  x   12 kx 2 , where F  V (2)
Now applying Newton’s 2nd law for such a problem we get
k
F  mx  x  x0 (3)
m
The solution of the above differential equation is
x  A cos t  B sin t ( 4)
with   k m , and A & B are arbitary constants
If, for instance, at t=0, x=A and v=0 then
x  A cos t with A stands for the amplitude
The total energy of the Oscillator is
E V  K  12 kx 2  12 mx 2

Knowing that x   A sin t 

 
E  12 kA2 cos 2 t  12 m 2 A2 sin 2 t  12 kA2 cos 2 t  sin 2 t  12 kA2

or E  12 kx 2  12 mx 2  12 kA2  constant (5)


From Eq.(5) we conclude that E has continuous values.

For the velocity we have from Eq.(5)

or x    A2  x 2 (6)
It is clear now, from Eq.(6), that the particle can’t exceeds the point x=A,
otherwise we will have an imaginary speed.
If the probability of finding the particle in a region of length x to be P(x) x, and
let t be the time required for the particle to cross x.

Since each particle crosses x twice during each complete oscillation, so we


have.
2t P x  
2 2
P x x  , with T is the period  
T T x t  T x

x T 2
Note that the probability is normalized, i.e.,  P x dx   2 T dt  1
0 0
This means that the probability density is inversely proportional to the speed.
Clearly, you are more likely to find the particle in regions where it is moving
slowly, and vice versa.
Using Eq.(6) and the fact that T=2/ we get
for the classical probability density
2 1
P x    (7)
2 A2  x 2  A2  x 2
It is clear now, from Eq.(7), that the
probability density is minimum at x=0 and
increases with increases the
displacement. It goes asymptotically to
infinity when x=A.
The Quantum H.O:
The potential energy of the linear Harmonic Oscillator, from Eq.(2), is given by

V  x   12  2 x 2

The Schrodinger equation now becomes

 2 d 2 1 
   2
x 2   x   E   x  (8)
 2 dx 2 2 
 
Let us now make Eq.(8) a dimensionless equation. To do that we introduce a
dimensionless parameter  as
  x  Here must has a dimension of m-1
d d d d d2 d 2
d  dx       2
dx dx d d dx 2 d 2
Eq.(8) now becomes

 2 2 d 2   1  2 2
  2 2     E   
2 d 2

 2 2 d 2   1  2 2
  2 2     E   
2 d 2

For the above equation to be dimensionally correct we must have

 2 2  2 
 2    
  
The equation can now be simplified to

 d 2    2
      E  
2 d 2 2

d 2   2
        ,  2E
  is also dimensionl ess 
d 2 

d 2  
d 2
    2    0 (9) Dimensionless equation
The direct series solution of Eq.(9) is given by

   ak  k
k 0
Substituting back in Eq.(9) we get
  
 k k  1ak  k 2
   a k    ak  k  2  0
k
k 0 k 0 k 0

Equating the coefficient of  k to zero 

k  2kak 2   ak  ak 2  0
Which is not a 2-terms recurrence relation, i.e., Eq.(9) doesn't lead to a two-
terms recurrence relation.
To solve it we first note that as    Eq.(9) approximated to
d 2  
  2    0 
d 2
 2 2 2 2
    e e

But the second term violates the boundary condition     0 
   
So we propose a solution for Eq.(9) of the form

 2 2
    e v  (10)

d    2 2  2 2 dv 
Now  e v   e
d d

d 2    2 2 2  2 2  2 2 dv   2 2 d
2
v 
 e v    e v   2e e
d 2 d d 2
Substitute back in Eq.(9) we get

d 2v  dv 
 2    1v   0 (11)
d 2 d
Equation is called Hermit's differential equation where its solution is
represented by infinite series of the form

v    ak  k 
k 0
dv   k 1 d 2
v   
k 2
  kak  and   k k  1a 
d k 0 d 2 k 0
k

Substituting back into equation (11) we get


  
 k k  1ak  k 2
 2  kak     1  ak  k  0
k
k 0 k 0 k 0
Equating the coefficient of  k to zero 

k  2k  1ak 2  2kak    1ak  0 


   1  2k  
ak 2   ak (12)
k  2k  1
If we set a1 =0 and choosing ao arbitrary we generate the even solution.

In the other hand if we set a0 =0 and choosing a1 arbitrary we generate the


odd solution. 
 2 2
    e veven    vodd   (13)
ak  2 2
It is clear from Eq.(12) that 
ak k  k
 2j  k
2
e    
k 0  k !
Now let us examine the series j!
j 0
k  2
 !

ak  2
  2

1

2
ak  k   k  k  k
  1!   1
2  2 
The function v   and the function e behave alike as   
2

both even and odd series violate the boundary condition     0 
   
To solve this dilemma we have to terminate the series after a finite number of
terms, say n, 
an 2  0  from Eq.(12) we get

 
2E
 1  2n  (14)

If n is even  an  2 and all higher terms vanish and since the odd series get
unacceptable solution we set a1 =0

If n is odd we retain only the odd series by letting a0 =0 .  the solution exits
for integer n .
from Eq.(14) we get

En  n  12  (15)

 
It is clear from Eq.(15) that En1  En  n  1  12  n  12   

The energy levels of the harmonic oscillator are evenly spaced, regardless of n.
and the Eigen-functions are
 2 2
    Ne v 

 2 x 2
or  x   Ne H n x  (16)
Where Hn are the Hermit's polynomials with the following properties:

 tn
t 2  2tx
g  x, t   e   H n x generating function
n 0 n!

n x2 d n  x2
H n  x    1 e
Rodrigues formula
n
e
dx
Parity
H n  x    1 H n  x n

H n  x   2 xH n  x   2nH n  x   0
Differential equation

 x2
e H n  x H m  x dx  2n  n! nm Orthogonality


H n1 x   2 xH n  x   2nH n1 x   0 



H n  x   2nH n1 x   0 
Recurrence relations

H n  x   2 xH n  x   2nH n  x   0 

Now using the Rodrigues formula one can prove that

H0 x  1 H1 x   2 x

H 2 x  4 x2  2 H 3  x   8 x3  12 x
Now since the wave function given by Eq.(16) must be normalized we have
 
  2 x 2 
   x  x dx  N H n x H n x dx  1
2
e with  
  
Using the orthogonally property we get
1
N 2     4 1
2 n
 n! 1  N n  
 2  n!    2n n!
The normalized wave-functions of the Harmonic oscillator is given by
1 
   4 1  x2   
n  x     e 2 Hn x
   n
2 n!   

1 
   4  x2
0  x     e 2 Eo  12 
  
1 
 4  3 3 4  x2 E1  32 
1 x     xe 2
  
3
 

E2  52 
1
    2  4  x2
2  x      x  1e 2
 4   4 
Let us draw the wave function and the probability density for n=0,1,2 and make
correspondence with the classical harmonic oscillator.
Recall that the classical harmonic oscillator has an energy given by

Eclassical  12 kA2

That is the classical energy is continuous: it allows all values between –A & A.

But from Eq.(15) it is clear that the energy of the quantum mechanically H.O is
quantized: it has discrete values.

Also in classical case the probability is inversely proportional to the speed, i.e.,
it is minimum about x = 0 and high at the turning points.

In quantum case the situation is the opposite for lower states and approaches
the classical limit as n→∞.

The ground state for the classical oscillator has zero energy (and zero motion),
whereas the quantum oscillator in the ground state has an energy of E0  2 
1

This quantum fact is a consequence of the Heisenberg uncertainty principle.


The probability density for the quantum
oscillator “leaks out” beyond the x = ±A
classical limits (tunneling).
In classical case the particle can never
exceeds these limits, since if it did it
would have more potential energy than
the total energy.
The region of non-zero probability
outside the classical limits drops very
quickly for high energies, so that
approaches the classical limit as n→∞.

the quantum probability density will have n+1 maxima and n minima. These
minima correspond to zero probability! This means that for a particular quantum
state n there will be exactly n forbidden locations where the wave function goes
to zero (nodes).
This is very different from the classical case, where the mass can be found at
any location within the limits −A < x < A.
Exercise: Prove that

(1) x kn 

 n k ,n1  n  1 k ,n1 
2

(2) p kn  i  n  1 k ,n1  n k ,n1 
2

(3) x2 


n  12 

(4)  
p 2   n  12

(5) x p   n  12 


(6) Estimate the ground state energy of a H.O. by minimizing
p 2
E  12  2 x 2
2

Subject to the uncertainty restriction x p   12 


The Dirac Notation Method
Let n be the eigenket of the system 

H n  En n (1)
p2 1
with H  2  2 x 2 (2)
2

Let us introduce the dimensionless coordinate and momenta operators


 1
x  x and p  p
2 2 
The Hamiltonian of Eq.(2) becomes


H   p  2  x  2  
H   x  ip x  ip H   x  ip x  ip  ix, p

1 i
but x, p   x, p   
2 2
  
H   p2  x2    x  ip x  ip  12 
Letting x  ip  a (3)

and x  ip  a† (4)


 H   a†a  12  (5)

Substituting back in Eq. (1) 

 
 a†a  12 n  En n 

a†a n  12  n  En n (6)

Letting a†a n  n n (7)

Eq.(6) becomes 
 
 n  12 n  En n

or En   n  12  (8)

To find the values of n we operate on Eq.(7) by a we get

aa†a n  n a n (9)

From the definition of Eqs.(3&4) it is easy to prove that

a, a†  1 (10)

 aa†  1 a†a
Substituting back in eq.(9) we get

1  a†a a n   na n 


Letting a†a n  n n (7)

a†a a n   n  1a n  (11)

Comparing eqs (7) and Eq.(11) we conclude that

n and a n  are eigenkets for the operator a†a 


The first with eigenvales of n  While the second with eigenvales of n  1

a n  Cn n  1 (12)

By operating on Eq.(7) by a † one can prove that

a† n  Cn n  1 (13)

Now multiplying Eq.( 12) by its bra form 


† 2
na an  Cn n 1 n 1 (14)

Recalling Eq.(7) a †a n  n n (7 )

Now, from Eq.(7) and Eq.( 14) 

2
n  Cn (15)

Similarly, multiplying Eq.( 13) by its bra form 

† 2
n aa n  Cn n 1 n 1

But from Eq.(10) we have aa †  1 a†a

The last equation becomes

 †
 2
n 1  a a n  Cn n  1 n  1
Using Eq.(7) we get

2
1  n  Cn (16)

From Eq.(15) we conclude that n  0

 there must be a nonnegative minimum value such that , from Eq.(12)

a n  Cn n  1 contradiction
(12)


a nmin  Cmin nmin  1 
Cmin 0 
From Eq.(15) we conclude that
min  0
By applying the operators a & a† repeatedly on Eq.(7) we can generate from
any given eigenket n new eigenkets with different eigen values that is
integrally spaced.

 n composed of a set of positive integers and zero.


Setting n  n and using Eq.(8) we obtain


En   n  12  (17)

Now from Eqs.(15 & 16) we have

Cn  n Cn  n  1

Eqs.(12 & 13) now read

a n  n n 1 (18)

a† n  n  1 n  1 (19)

Equations (18) and (19) specify completely the operators a & a† as the
lowering (annihilation) and raising (creation) operators, respectively.
Recalling Eqs.(3&4) we have

x  i p   a x  ip  a †
 1  1
 xi pa & x i p  a†
2 2  2 2

Adding and subtracting the above two equations we get, respectively

2 2
x  a  a† i p  a  a†
 

x

2
a  a†  (20)

pi

2
a†  a  (21)
Matrix Representation of Some Operators
Let is now evaluate the matrix elements of some operators

X mn  m x n 

2

m a  a† n  

X mn 

2
 m a n  m a† n   X mn 

2
 n m n 1  n  1 m n  1 

X mn 

 n m,n1  n  1 m,n1 
2
 0 1 0 0 
  a n  n n 1
 1 0 2 0 
 
x 0 2 0 3  a† n  n  1 n  1
2  
 0 0 3 0 
 
     
For the lowering operator we have

amn  m a n  n m n  1  amn  n m,n 1



0 1 0 0 
 
0 0 2 0 
a  0 0 0 3  a n  n n 1
 
0 0 0 0 
  a† n  n  1 n  1
    

For the raising operator we have a†

a†mn  n  1 m,n1
a †mn  m a † n  n  1 m n  1  
 0 0 0 0 
 
 1 0 0 0 
a†   0 2 0 0 
 
 0 0 3 0 
  
    
Now for the Hamiltonian operator we have

   
H mn  m H n   m a† a  12 n   m a† a n  12  m n 

H mn   n m a† n  1  12 mn   n  12  mn

1 0 0 0 
 
0 3 0 0  a n  n n 1
 
H 0 0 5 0 
2   a† n  n  1 n  1
0 0 0 7 
  
   
Let us generate some eigen functions:

Using Eq.(17) we have a0 0 


Expressing a in terms of x and p 
  1 
 xi p 0  0
2  

 2

   d   x  1 d    0
 x 0  0
2  dx 
 

 0
 dx  
 2
d d0
dx
0   2 x0  0  0
  2 xdx 
1  2 x2
  4 
0    e 2
 
To generate the first excited state we have

a† 0  1 

  1   2 1 d 
 x i p o  1   x o  1
 2 2    2 2 2 dx 
 
1  2 x2
  4  d
0   2 x0
But 0    e
 
2 and
dx 
    2 x2

2 2 1
 
 o  1  1  2xo   
4
x 2xe 2
 2 2   
 
Exercise: Prove that for the H.O x p   n  12 
2
where A  A2  A
The Motion of the Wave Packets:
The evolution of the classical Harmonic oscillator is governed by Eq.(3), which
is by Newton’s second law as
k
x  x  0 (3)
m
Which has a solution given by Eq.(4), i.e.,

x  A cos t (4)

with E  12 kA2  12 m 2 A2 (5)

Now we want to show how the quantum Harmonic oscillator evolves with time
using the time dependent Schrodinger's equation. Also we want to show if we
can derive a correspondent equations similar to Eq.(4) and another classical
harmonic oscillator equations.
A wave packet is a superposition of many waves with slightly different
momenta, i.e.,


  x,0   Cn n  x  (22)
n 0

Let us now show how the oscillator evolves with time. It is known that
i
  Ent
  x, t    Cne  n x (23) 
n 0

If the initial state is known, that is, If   x,0  is known we can find the state at
any time by Eq.(23) provided that you can find Cn. To find Cn we have to
multiply both sides of Eq.(22) by m  and integrate we get

   

 m ( x)
  x,0dx   Cn  m n dx   Cn mn  Cm (24)
 n 0  n 0
Substituting for En from Eq.(16) into Eq.(23) we get for the state at any time
i
 t 
  x, t  e 2  Cne int n  x  (25)
n 0

With Cn is given by Eq.(24)

Now if  x,0 is normalized we have


   
  
 x ,0   x ,0 dx    C m n  m  x n  x dx 

C  

 m 0 n 0 
 
1    Cm Cn mn 
m 0 n 0

 Cn  1
2
(26)
n 0

Probability of finding the oscillator in the nth state which is time-independent.


Let us now find the expectation value of the Hamiltonian, we have
   2 

H     x, t H  x, t dx     x, t  
 
  2  x   x, t dx
2 1 2 2

   2 
Substituting for  x, t  From Eq.(25) into the above equation we get

  2 2 1 2 2

 n  x   2   2  x n  x dx
 
H   Cn
2
n 0   
From the Schrödinger equation we have

 2 2 1 
 2   2  x n  x   En n  x 
 2 2  
 
  
H   Cn En  n  x n  x dx   Cn En

2 2
(27)
n 0  n 0

From Eqs. (26) and (27) we conclude the two facts :


(1) The only possible results of measuring the energy of the system are the
energy values En.
(2) If the system is in the state  x, t 
2
the probability of obtaining the result En is equal to Cn

As stated in postulate no. (IV).

Let us now find the expectation value of x with respect to the state  x, t 


x   n  x, t xk  x, t dx

  
  CnCk ei nk t  n  x xk  x dx
 
n 0k 0 
But we proved that

 n  x xk  x dx  2  n k ,n1  n  1 k ,n1 
 


 
x 

C 
 n k
C e i n  k t
 n k ,n1  n  1 k ,n1  
2 n0k 0

  nCnCn1eit  n  1CnCn1e it 


 
x 
2  n0

Letting n+1=n in the 2nd term we get

 n CnCn1eit  Cn1Cn e it 


 
x 
2 n0

Setting C n  C n e i n 

x 
n 
 n n1
2 n0
C C ei t n n 1 
 C C
n n 1 e i t n 1 n 

But ei t n n1   e i t n1 n   cost  n1  n  

2n 
 x   Cn Cn1 cost   n1   n  (28)
 n0
Let us test the last result at the limit n → ∞. At this limit we can assume

n1  n   and Cn1  Cn

Again for n → ∞ we can write for the energy

En   n  12   n

Then Eq.(28) now reads

2   
2  n
En  cost   
2
x   C
  n0 
  
Knowing that   Cn En   E
2

 n 0 
2E
x  cost    (29)
 2

For classical harmonic oscillator and using Eq.(4) and Eq.(5) we have

x  A cos t   

and E  12  2 A2 

2E
x  cost    (30)
 2

Which match Eq.(29).


In the first chapter we proved that

A d 1
x   x, H 
d 1
A   A, H   
dt i t dt i

 p2 1 2 2 1 p
x, H    x,  2  x   x, p p  x, p  ip
 2  2 2 
d p
 x  (31)
dt 
d 1
similarly p   p, H 
dt i
 p2 1 2 2
 p, H    p,  2  x   12  2  p, xx  12  2 x p, x   2i  2
 2 
d
 p    2 x (32)
dt
Differentiate Eq.(28) with respect to t we get

d2 1d p
x 
dt 2
 dt
Using the result of Eq.(32) we get

d2
2
x   2
x 
dt
Which is similar to Eq.(3) with again

x  x 0 cost    (33)

You might also like