Quantum mechanics summary

Humberto Gilmer

Foreword
This summary sheet was compiled over the course of the 2014-2015 school year from the Physics 7501/7502 courses taught at
the Ohio State University by Dr. Ulrich Heinz, as well as from the lecture notes from the Physics 312 course taught at Rice
University in Spring 2013 by Dr. Paul M. Stevenson. Most of the formulae were pulled from Modern Quantum Mechanics by
J.J. Sakurai and Principles of Quantum Mechanics by R. Shankar. The explanations are my own and are meant to provide
illuminations of subtle points.

Math
Gamma function:

tn1 et dt = (n) = (n 1)!

Gaussian integral:
Z

x2

r
dx =

Virial theorem with V x

n :
D E
D E
2 T = n V
Gradient:

x
+
z +
z
x
z
z

1
1

=
r +
+

r
r
r sin
=

Laplacian:
2
2
2
+ 2+ 2
2 =
2
x
z
z




1
1

1
2

2 = 2
r2
+ 2
sin
+ 2 2
r r
r
r

r sin 2
Unitary operator:
1 = U

U
Hermitian operator:
=H

H
Commutators:
h
i h
i
h
i
B
C = A,
B
C + B
A,
C
A,
h
i h
i h
i
B
C + C,
AB
+ B,
C A = 0
A,
Baker-Hausdorff lemma:

iG

 n nh h h
h
i  i2 2  h h
ii
h
iii
i
i 

iG

G,
G,
. . . G,
A
Ae
= A + i G, A +
G, G, A + +
G,
... + ...
2!
n!

Constants:
R=

me4
2~2

e2
~c

a0 =

~2
me2

General formalism
Time-independent Schr
odinger equation:
2
= p + V
H
2m

|i = E |i
H
Time-dependent Schr
odinger equation:

|i = i~ |i
H
t
Time evolution of a state:

(t, t0 ) |(t0 )i = e ~ H(tt ) |(t0 )i

|(t)i = U
Time evolution of energy eigenstates:
i

|n (t)i = e ~ En t |n i
Connection between Schr
odinger and Heisenberg pictures:
i

|iH = e ~ Ht |(t)iS

i

H = e ~i Ht

S e ~ Ht

H = H
S
H

Heisenberg equation:
H
d
~h i
=
H, H +
dt
i

!
H

Completeness and orthogonality (for a set of discrete eigenstates {|ni} or continuous eigenstates {|xi}):
X
|ni hn| = 1
hm | ni = mn
n

hx0 | xi = (x x0 )

|xi hx| dx = 1

Born rule:
Pif = |cf i |2

cf i = hf | ii

Probability conservation:
2

(~x, t) = | (~x, t)| = |h~x | (t)i|

~ (~x, t) =

i~
2m

( () )

+ ~ = 0
t
Spherical harmonics:
Hellmann-Feynman theorem:
dEn
=
d


 H

n () 



+


 n ()


Hydrogen atom
Potential:

2
2
= p e
H
2m
r


2
~

1
e2

2
2
|i 7
H
r

2mr2 r
r
2mr2
r

Energy spectrum:
En =

R
me4
mc2 2
=

n2
2~2 n2
2n2

Spatial wavefunctions:
Expectation values:
h
ri = 0
 
1
1
=
r
a 0 n2
 
1
1


= 2 3
r2
a0 n l + 12
 
1
1
1


= 3 3
r3
a0 n l l + 21 (l + 1)

Quantum harmonic oscillator

1D
Raising and lowering operators:
r
r




m
i
m
i

a
=
x

p
a
=
x
+
p
n
=a
a

2~
m
2~
m

a
|ni = n + 1 |n + 1i
a
|ni = n |n 1i
n
|ni = n |ni




a
, a
= 1
n
, a
= a

[
n, a
] =
a
r
r




~
m~
x
=
a
+ a

p = i
a
a

2m
2
3

Potential:

2
= p + 1 m 2 x
H
2
2m 2



1

H = ~ n
+
2

Energy spectrum:


1
En = ~ n +
2

Spatial wavefunctions:

3D
Potential:
Energy spectrum:
Spatial wavefunctions:

Symmetries
The expectation value of an operator must behave like the classical counterpart of the operator. In other words, on the
average, quantum mechanics approaches classical mechanics. This constraint can be placed on either the states or the
operators, depending on whether youre in the active or passive picture.
Translations:
i

|i
|i 7 T() |i = e ~ p
 E
D 


xT()  = h | x
| i +  = h
xi + 
h
xi = h | x
| i 7  T ()

under a unitary transform U

:
Invariance of a scalar operator

U
=

i
U
=0
,

the above implies:

For an infinitesimal unitary transform generated by an operator G,
h
i
i

() = e ~i G
G
=0
U
= 1 G
+ ...
,
~
:
Invariance of a vector operator Vi under a unitary transform U
Vi U
= Rij Vj
U
Generalized Ehrenfest theorem:

D E

d
dt

~ Dh iE
=
H, +
i

h
i
the symmetry is encoded in the commutator ,
G
= 0, and the generalized
For an infinitesimal translation generated by G,

Ehrenfest theorem implies a conservation law of the corresponding quantity generated by G.

Ehrenfest theorem (for position and momentum):
D E
D E

D
E
D E
d p~
d ~x
~
= V
m
= p~
dt
dt

Uncertainties
be two Hermitian operators
For the following section, let A and B
Definitions:
D E
D E

B
1
A = A A 1
B
=B
Uncertainty:
A =

rD

E D E2
A2 A
4

Generalized Heisenberg uncertainty relation:

2

1 Dh iE2 1 Dn oE2
A, B
+
A, B
4
4
Dh
iE
2
1
2
2
B

A,
(A) (B)
4
2

(A) (B)

Some examples:
~
2
~
Et
2
~
Jx Jy Jz
2
xp

Heisenberg uncertainty relation

Energy-time
Angular momentum

Angular momentum
General relations
Definitions:

~ ~
J~ = L
+S

J = Jx iJy

J2 = Jx2 + Jy2 + Jz2

Eigenvalues:
J2 |j, mi = j(j + 1)~ |j, mi
Jz |j, mi = m~ |j, mi
p
J |j, mi = ~ j(j + 1) m(m 1) |j, m 1i
Total commutation relations:
h

i
j = i~ijk
k
Ji ,
h
i
Jz , J = J


p~
i is an operator
i = A
where
xi + B pi + C ~x
i
Orbital commutation relations:
h
i
i,
j = i~ijk
k
L

h
i
Ji , Jj = i~ijk Jk
h
i
J2 , Ji = 0

h
i
i, L
j = i~ijk L
k
L

Spin commutation relations:

h
i
j = 0
Si ,

i
Si , Sj = i~ijk Sk

is evidence that they live in a different Hilbert space than the orbital
Note the fact that spin operators commute with
operators. The fact that the spin operators still obey the defining commutation relations is evidence that they are still
angular momentum operators, regardless.

Spin- 12
Spin operators:

Pauli matrices:


x =

0
1

1
0

~
Si 7 i
2


0 i
y =
i 0


z =

Pauli matrix relations:

[i , j ] = 2iijk k
{i , j } = 2ij 1
i j = ik

and cyclic permutations

2

(
n ~ ) = 1

 



~ ~ B
~ ~ = A
~B
~ 1+i A
~B
~ ~
A

Rotation operator:
 
 
 
~ 7 cos 1 i sin ~
U
2
2
5

1
0

0
1

Angular momentum addition/Clebsch-Gordan coefficients




CG coefficients link vectors in the spin product space eigenstates of S12 , S1,z , S22 and S2,z to the total spin space eigenstates of S2 , Sz
|j, m; j1 , j2 i =

|j1 , m1 ; j2 , m2 i hj1 , m1 ; j2 , m2 | j, m; j1 , j2 i

m1 ,m2

with |j1 j2 | j j1 + j2 , j m j and m = m1 + m2

m
= hj1 , m1 ; j2 , m2 | j, m; j1 , j2 i is the Clebsch-Gordan coefficient. Product-total space equivalencies:
where Cjj
1 j2

j1 j2 = (j1 + j2 ) (j1 + j2 1) |j1 j2 + 1| |j1 j2 |

Two spin- 21 particles:



1 1
|1, 1i =  ;
2 2

1 1
=10



2 2  
 1 1
 1 1
1  1 1
 ;
 ;
|1, 0i =
;

+
|1,
1i
=
 2 2
 2 2
2 2 2

 

 1 1
1 1  1 1
|0, 0i =
;
 ;

2 2
2 2 2 2

Orbital and spin- 21 addition (spin-angular functions):

s
s






1


l
+

m
l + 12 m 
1
1
1
1
1
1 1 1
2
j = l , m; l,

=
l, m ; ,
+
l, m + ; ,

2
2
2l + 1 
2 2 2
2l + 1 
2 2 2

Wigner-Eckart theorem
Spherical tensor operator defintion:
i
h
Jz , Tq = q~Tq
k

Theorem:

i
h
p
J , Tq = ~ k(k + 1) q(q 1)Tq1
k

 
E
E
D
D

 
0 , j 0 , m0  Tkq  , j, m = hj 0 , m0 ; j, k | j, m; k, qi 0 , j 0 Tk , j

The Wigner-Eckart theorem allows matrix elements to be written as a product of a CG coefficient (which depends only on
geometry) and the reduced matrix element which is completely independent of geometry (including rotational invariance).

Time-independent perturbation theory

Time-independent perturbation theory may be employed in cases where the exact eigenvalue problem with some Hamiltonian

H
|ni = En |ni
H
0 that may be tuned by some small parameter
cannot be solved analytically. However, if the Hamiltonian contains a term H
, so that
=H
0 + H
0
H
0 is the Hamiltonian of some system that may be solved exactly
where H


E
E
0 n(0) = En(0) n(0)
H
n
o


0 and En(0) is the set of corresponding energy eigenvalues. In this language, H
0
where n(0) is the set of eigenstates of H
0 . The key to perturbation theory lies in the fact that both the exact eigenstates {|ni}
is the perturbation to the system H
and the exact energy eigenvalues {E} may be expanded in a power series in the tuning parameter :



E 
E
E
E
 [k]



En[k] = En(0) + En(1) + + k En(k)
= n(0) +  n(1) + + n n(k)
n
Perturbation theory is divided into two kinds of problems, those where the unperturbed energy spectrum is non-degenerate
and where those where it is degenerate.

Non-degenerate perturbation theory

First order energy correction:


E
D
 0  (0)
En(1) = n(0)  H
n
First order state correction:



E
D
0  n(0)


E m(0)  H
E
X
 (0)
 (1)
=
m
n
(0)
(0)
En Em
m6=n

Second order energy correction:

En(2) =



D
E
 0  (0) 2


m
X  n(0)  H
(0)

m6=n

(0)

En Em

Degenerate perturbation theory



Hamiltonian matrix in subspace of m degenerate eigenstates n(0) ; m


E
D
 0  (0)
H = n(0) ; m0  H
n ;m
The eigenvectors of the above matrix are the first order energy states of the complete system. These states represent the
unique mixtures (to first order) of the unperturbed energy eigenstates with degenerate unperturbed energies whose degeneracy
0 . As such, the energy eigenvalues of the matrix corresponding to the eigenvectors are the first-order energy
is lifted by H
corrections. Use these states and energies to compute the first order state corrections, second order energy corrections, etc.
However, diagonalizing the matrix is generally a non-trivial task. This task can be made easier by employing selection rules
to knock down the number of states with non-trivial matrix elements between them. This is equivalent to saying that the
perturbation does not lift the degeneracy between those particular states.

Selection rules


h
i
D
E


be a Hermitian operator such that
n(0) , = |n, i. If H
0,
= 0, then m, 0  H
0  n, =
Operator selection rules: Let
0 unless = 0 .
Angular momentum selection rules: Let Tkq be a spherical tensor operator. Then
 
E
D
 
0 , j 0 , m0  Tkq  , j, m = 0 unless |j k| j 0 (j + k) m0 = m + q

Some examples with hydrogen

Spin-orbit coupling:

(1)

Ejjz =

~ S
~
e2 L
0
S.O.
H
=
2 2
3
 2m c r

2j+1
eB~jz

for j = l +

2j
2mc



eB~jz 2j+1
2mc

2j+2

for j = l

1
2

1
2

Relativistic correction:

(1)

Enl

p4
0
relativistic
H
= 3 2
8m c
!
1 2 4
3
1


= mc
3
2
4n4
n l + 12

Zeeman effect:

~ 
ekBk
0
Zeeman
z + 2Sz
H
=
L
2me c


~
e~k
Bk
1
(1)
Elm =
m 1
2me c
2l + 1
7

Stark effect
0
H

Stark = eE x
En(1) = 0

due to parity

Miscellaneous
Calculating expectation values using perturbation theory (see Shankar, Problem 17.3.4 for examples):
of which we wish to find the expectation value in a given system characterized by some HamilSuppose there is operator
0 . If H
0 has some component that could be combined with the operator and the resulting Hamiltonian H
solved
tonian H
0 and H
must be attributable to the perturbing term. As an example, consider
exactly, the differences in energies between H
0 = p2 e2 :
= and H

r
2m
r
2
2
= p e +
H
2m
r
r
p2
e2 +
=

2m
r
p2
e2 ()
=

2m
r

let
e2 () = e2 +

This is a modified hydrogen atom, with energies

En () =

2


m
m
me4 ()
= 2 2 e2 + = 2 2 e4 + 2e2 + 2
2
2
2~ n
2~ n
2~ n

Variational method
the hermiticity of H
guarantees that the spectrum of
Although the full eigenvalue problem may not be solvable for H,
basis states {|ni} are complete and orthogonal. This means that any arbitrary state |i may be decomposed as a linear
combination of these (unknown) basis states. In other words, the basis can still form a resolution of the identity. First note
that for any state |i, the TISE gives
D   E
 
H

|i = E |i E =
H
h | i
This may be used to provide an upper bound to the ground state energy (E0 ):
P

E0 = E0 P

|hn | i|

|hn | i|

E0

En |hn | i|
=
P
2
n |hn | i|
n

D   E
 
H

h | i

En h | ni hn | i
h | ni hn | i
P D   E
n  H  n hn | i
= P
h | ni hn | i
D n E
 
nH n
=
h | i
n

=E

Employing the physical properties of the potential of interest (such as parity, number of nodes, length scales, etc.) it is
possible to choose a test state |()i where is some tunable parameter which may be manipulated to minimize E and
therefore provide as a low an estimate of the ground state energy. To do this, calculate the energy quotient as shown above
and differentiate with respect to the tunable parameter(s), solve for the minima (make sure to check the stationary points!)
and evaluate the energy quotient at the respective point. This procedure may also be used to calculate higher-order energies
(although the strict upper bound cant be guaranteed). This may be done by choosing another test function that in addition
to satisfying some family of parameters (parity, number of nodes, etc.) is also orthogonal to the test function for the ground
state.

Born-Oppenheimer approximation
The Born-Oppenheimer approximation is used to solve for the approximate energy levels of a molecule. The sketch of the
procedure is first solving the Schr
odinger equation assuming the positions of the nuclei are fixed. This is generally not exactly
solvable, so a good estimate can be made by means of the variational method. Next, the electrons are assumed to sit in
one of their energy eigenstates (generally the ground state) and this generates an effective potential in which the nuclei sit.
The local minima of the electronic potential may be expanded about to give approximate harmonic behavior, which finally
allows for the solution of the energy eigenstates of the entire molecule. Begin by noting down the Schrodinger equation for
a molecule with K nuclei and k electrons:



 
 
, X
+ Vnn X
|i = Tn + Te + Vne ~x
~ + Vee ~x
~ |i
H

K,k
k
K
2
2
2
X
X
X
X
X
1
1
Z Z e
e
Z Z e
1 2
1 2

+

N n +
M N |i
P +
Pn
E |i =

2MN N
2m
2
2
n ~x
m
~N
~N X
~ M
x
xn X
n

N,n ~
N
n6=m ~
N 6=M X
Ideally, we could solve for the energy associated with the state |i, which would be the molecular energies. In accordance
with the procedure sketched above, since the nuclear masses are so much larger than the electron masses, this effectively
makes the nuclear kinetic term a perturbation (although it cant
beE treated with perturbation theory!), so solve the reduced


Schro
odinger equation for the electronic energy eigenstates a,X~ , which are labeled by the energy eigenvalue a and the
~ by dropping the nuclear kinetic energy operator Tn :
nuclear positions (which are nailed down, X),


E
E

 

 
  
, X
+ Vnn X
~  ~
~ + Vee ~x
~  ~ = Ea X
Te + Vne ~x
a,X
a,X

The normalization of these states is


D
E



3 ~
0
~N
a,X~ 0  b,X~ = ab K
XN X
N

Now cast the energy eigenstate portion of the electronic ket above into the electronic position basis

D
E



~
~x,X~  a,X~ 0 = a ~x, X
so the reduced Schr
odinger equation becomes in the electronic position

K,k
k
X
X
~2 2
1 X
e2
1 X
Z Z e2

N n +
+
~xn


2mn
~ N 2 n6=m k~xn ~xm k 2 N 6=M
xn X
n
N,n ~

basis

  

ZM ZN e2  ~ 
~ a ~x, X
~

a ~x, X = Ea X
~
~M
XN X

The actual state |i may be expanded as a linear combination of the electronic energy eigenstates
 
E
XZ
~  ~ dX
|i =
fa X
a,X
a

 
~ are governed by
where the expansion coefficients fa X
!
K
 
 
 
X
~2
2
~
~ = Efa X
~
X~ + Ea X
fa X

N
2MN
N

which is the Born-Oppenheimer equation. Note that this equation is arrived at by taking the inner product of the full
Schr
odinger equation with the
eigenstates and then invoking the Born-Oppenheimer approximation to neglect all
 electronic

~ . In addition, note that the effect of the electronic energy is to serve as a potential well in
but the the Laplacian of fa X
which the nuclei reside. The electronic energy for a given electronic configuration a may be found by means of the variational
 
~
method applied to the reduced Schr
odinger equation. Note that this equation is also relatively difficult to solve, since Ea X
may have multiple minima. Instead, expand the electronic energy function about its minima. To first order, the potential
about each minima may be approximated as a quantum harmonic oscillator. Thus the molecular energies to first order are
simply those of a quantum harmonic oscillator. There exists a fine structure corresponding to molecular rotation. Finally,
each vibrational-rotational spectrum is different corresponding to each different electronic configuration, a.

WKB approximation
General relations
In free space (or space where the potential varies so slowly, its essentially constant), the wavefunction should exhibit oscillatory
behavior, although the frequency and amplitude are not specified. Therefore, let (x) be defined such that
(x) = A(x)ei(x)
where A(x) and (x) are the amplitude and frequency, respectively, both as functions of x. Plugging into the Schr
odinger
equation (and rewriting slightly) gives
 0

 00
A
A
02
00
+i 2 +
+ k2 = 0
A
A
Taking the real and imaginary parts gives the following relations
A00
02 + k 2 = 0
A

A0
+ 00 = 0
A

The WKB approximation consists of noting that for a very slowly varying potential, A(x) in turn also varies slowly so
and so the term is negligible. Thus the solutions for A(x) and (x):
Z
C
C
=p
(x) = k(x)dx
A(x) = p
0
(x)
k(x)

A00
A

1

where the wavenumber k(x) is proportional to the momentum of the particle and is given by
r
2m
(E V (x))
for E > V (x)
k(x) =
~2
r
2m
(x) = ik(x) =
(V (x) E)
for E < V (x)
~2
Putting the solutions together gives the wavefunction in each region
Classical region:
R
R
C
D
(x) = p
ei k(x)dx + p
ei k(x)dx
k(x)
k(x)
Note that in the region where k(x) 0, (x) . In other words, in a region where the particle classically slows down
(momentum decreases, as near a turning point), it is more likely to be found.
Classically-forbidden region:
R
R
C
D
(x) = p
e (x)dx + p
e (x)dx
(x)
(x)

Linear potential/Airy functions

In the neighborhood of a turning point x = a, it is sufficient to examine the potential in a neighborhood where it is
approximately linear:

dV
= E + V 0 (a)(x a)
V (x) = V (a) + (x a)
dx x=a
This requires solving a form of the Schr
odinger equation with a linear potential

~2 d2 (x)
+ V (x)(x) = E(x)
2m dx2

~2 d2 (x)
+ (E + V 0 (a)(x a)) (x) = E(x)
2m dx2
~2 d2 (x)

+ V 0 (a)(x a)(x) = 0
2m dx2
d2 (x)
= (x a)3 (x)
dx2
d2 (z)
= z(z)
dz 2
10

where =

2m 0
~2 V (a)

13

and z = (x a). The solutions of this ODE are the Airy functions Ai(x) and Bi(x)
(z) = aAi(z) + bBi(z)

The properties of the Airy functions are

Integral representations
1
Ai(z) =


cos


s3
+ sz ds
3

1
Bi(z) =

s3 +sz

+ sin

s3
+ sz
3


ds

Asymptotic forms
za
Ai(z)

1
1

(z) 4

za

sin

3
2

(z) 2 +
3
4


Bi(z)

1
1

(z) 4

cos

3
2

(z) 2 +
3
4

3
2
1
Bi(z) 1 e 3 z 2
2 z 4

3
2
1
Ai(z) 1 e 3 z 2
2 z 4

Connection formulas
Let a be a turning point such that V (a) = E. Let V (x) < E for x < a such that x a is a classical region and let V (x) > E
for x > a such that x > a is a classically-forbidden region. In the region x < a, the wavefunction may be written as a linear
combination of WKB wavefunctions:
(x) = p

A
k(x)

ei

k(x)dx

+p

B
k(x)

ei

k(x)dx

while in the region x > a, the wavefunction is only of the form of the decaying exponential WKB wavefunctions:
R
D
e (x)dx
(x) = p
(x)
00

However, in the immediate neighborhood of the turning point, the WKB approximation breaks down b/c AA 1 (or, put
another way, 1 blows up since the momentum of the particle approaches 0). In this neighborhood, the potential can
k(x)

be approximated linearly, which means in the immediate neighborhood of the turning point, the analysis outlined in the
previous subsection may be employed, so that the patching wavefunction in the neighborhood of the turning point is a
linear combination of Airy functions
p (x) = aAi(x a) + bBi(x a)
Now, assuming the neighborhood of the turning point is large enough such that the asymptotic forms of the Airy functions may
be employed at its extremes, and in addition the wavefunction at these boundaries is well-behaved enough to be approximated
by WKB wavefunctions (which is a somewhat strict double-constraint), the patching wavefunction becomes





3
2
1
2

1
2

2 +
3 +
a
((x

a))
+
b
cos
((x

a))
xa
1 sin
1

3
4
3
4
((xa)) 4
((xa)) 4
p (x) =
3
3
2
2
23 ((xa)) 2

3 ((xa))
+b 1
xa
a 1
1 e
1 e
2 ((xa)) 4

2 ((xa)) 4

Simultaneously, employing the fact that in this neighborhood, the potential may be approximated linearly V (x) E +
V 0 (a)(x a), and inserting this relation into the definition of k(x) gives
r
r
r
1
3
1
2m
2m
2m 0
0
k(x) =
(E V (x))
(E E V (a)(x a)) =
V (a)((x a)) 2 = 2 ((x a)) 2
2
2
2
~
~
~
and

(x) = ik(x) = 2 (x a) 2
In the x > a overlap region
Z

(x0 )dx0 = 2

(x a) 2 dx0 =

3
2 3
2 (x a) 2
3

so that the WKB wavefunction in this region becomes

(x) =

3
4

(x a)
11

1
4

e 3 2 (xa) 2

Since the patching wavefunctions may be used to smoothly map the the oscillating WKB solutions to the exponentially
decaying WKB solutions, then compare the powers of the asymptotics of the patching solutions to the asymptotics of the
WKB solutions so that
r
D
4
a
a=D
1 = 3

2 4
4
b=0
In the x < a overlap region
Z

a
0

k(x )dx =

3
2

((x a)) 2 dx0 =

3
2 3
2 ((x a)) 2
3

so that the WKB wavefunction in this region becomes

(x) =

2i

3
4

((x a))

1
4

e 3 2 ((xa)) 2 +

2i

3
4

((x a))

1
4

e 3 2 ((xa)) 2

The patching wavefunction (with b = 0) gives

p (x) = a

3
2

((x a)) 2 +
3
4

1 sin
((x a)) 4
 



3
3
1
i 32 ((xa)) 2 +
i 23 ((xa)) 2 +
4
4
=a
e

e
1
2i ((x a)) 4


3
3
2i
1
i
((xa)) 2
i
((xa)) 2
2i
4
3
4
3
e e
e
=a
e
1
2i ((x a)) 4

Again, comparing powers, this gives relations for A and B to a:

a
A
i
1e 4 = 3
2i 4
4

a
B
i
1e 4 = 3
2i 4
4

Inserting the previously derived relation between a and D gives

a
A
i
1e 4 = 3
2i 4
4


1
1
A
1
i
4
2 2
D(4)
= 3
e
1
2i 4
4


3

A
iei 4 D 4 = 3
4

iei 4 D = A

a
B
1 ei 4 = 3
4
2i
4



1
1
1
B
1 ei 4 D(4) 2 2 = 3
2i 4
4



3
B
iei 4 D 4 = 3
4

iei 4 D = B

Therefore the wavefunction about x = a becomes

Ra
2D sin x k(x0 )dx0 + 4
k(x)
Rx
(x) =
0
0
D e a (x )dx
(x)

x<a
x>a

By similar analysis, the wavefunction about x = b where the wavefunction crosses from E < V (x) to E > V (x) from left to
right (where b < a)

Rb
0
0
C e x (x )dx
x<b
(x)


R
(x) =
2C sin bx k(x0 )dx0 4
x>b
k(x)

12

Bound states
The WKB approximation may be used in the case of a well to calculate the energy levels for a given potential without ever
calculating the wavefunctions (!). This is done by noting that the wavefunction may be written either of ways listed above,
which induces a consistency relation between the phases of the different ways the phase of the wavefunction may be written:
Z a
Z x

2 =
k(x0 )dx0
1 =
k(x0 )dx0 +
4
4
x
b
The two above phases must (up to an integer number of ) be the same:
2 = 1 + n


a
1
k(x0 )dx0 = n +

2
b

which is very similar to the Bohr-Sommerfeld quantization rule. Once the integral is performed, the result may be solved
for E(x) (contained in the k(x)) to derive the energy spectrum. Note that when the system involves a central potential, the
WKB approximation may be used to calculate the energy levels, but the condition changes slightly:


Z rmax r
3
2m
(E V (r))dr = n +
~2
4
0

Tunneling

Time-dependent perturbation theory

There are two principal approaches to time-dependent perturbation theory. The first is Diracs approach, which involves
writing the states as a linear decomposition of basis states, where the coefficients are allowed to evolve in time. The other
involves writing a formal solution to the Schwinger-Tomonaga equation and perturbatively expanding this formal solution.

Dirac approach

(0) + H
0 (t) where all time-dependence may be shunted onto this perturbing
For a system described by a Hamiltonian
H(t)
=H
 (0) 
0


Hamiltonian H (t). Suppose n
constitute a complete set of energy eigenstates to the exactly solvable system described
(0)

by H


E
E
(0) n(0) = En(0) n(0)
H
and thus an arbitrary state |i may be written as a linear combination of these states:
E
X 
|i =
cn n(0)
n

0 (t) = 0, this is it. Recall that the time evolution operator (when H
(0) is independent of time) is
For the system where H
given by

t = e ~i Ht
U
and thus the time evolution of an arbitrary state |(t)i becomes
(t) |i =
|(t)i = U

(0) t

cn e ~ H



E X
E
(0) 
i
 (0)
=
cn e ~ En n(0)
n
n

0 (t) is introduced. Thus, time-evolution in addition to that due to H

(0) must also be introduced. This
Now suppose that H
may be done by allowing the parameters cn to vary in time. Thus

E
X
(0) 
i
|(t)i =
cn (t)e ~ En t n(0)
n

13

Inserting this into the Schr

odinger equation gives
d
|(t)i
|(t)i = H(t)
dt


E 
X
E
(0) 
(0) 
i
i
d X
(0) + H
0 (t)
cn (t)e ~ En t n(0)
i~
cn (t)e ~ En t n(0) = H
dt n
n



E

E

X
X
(0) 
(0)
i
i
i
0 (t)cn (t)e ~i En(0) t n(0)
i~
cn (t) En(0) cn (t) e ~ En t n(0) =
cn (t)e ~ En t En(0) + H
~
n
n


E X
E
X
(0) 
(0)
~i En
t
0 (t)cn (t)e ~i En(0) t n(0)
=
H
i~
cn (t)e
n
i~

Exploiting the orthogonality of the basis states yields




E X
D
D
E
X
(0)
(0)
i
i

 0  (0)
cn (t)e ~ En t m(0)  H
i~
cn (t)e ~ En t m(0)  n(0) =
(t)  n
n

i~

cn (t)e

(0)
~i En
t

mn =

(0)



E
 0  (0)
m(0)  H
(t)  n

(0)



E
 0  (0)
(t)  n
m(0)  H

cn (t)e ~ En

i~cm (t)e

(0)
~i Em
t

cn (t)e ~ En



E
(0)
(0) D
Em En
1 X
 0  (0)
t
~
cm (t) =
cn (t)ei
m(0)  H
(t)  n
i~ n


D
E
iX
 0  (0)
cn (t)eimn t m(0)  H
(t)  n
cm (t) =
~ n


(0)
(0)
was defined. Note that this defines a matrix differential equation; were it
where in the last step mn = ~1 Em En
exactly solvable, this would not be a question in perturbation theory. Alas, the equation isnt solvable so, here is where the
0
perturbation
 (0)  theory comes in. The matrix equation may solved iteratively. To first order, neglecting H (t), a system in a

will not evolve so
state n
c(0)
n = nm
(1)

Inserting this relation into the matrix equation for the cn gives the first order relation for cn :


D
E
i X (0)
 0  (0)
c(1)
cn (t)eimn t m(0)  H
(t)  n
m (t) =
~ n


D
E
iX
 0  (0)
mf eimi t m(0)  H
(t)  i
=
~ m


D
E
i
 0  (0)
(t)  i
= eif i t f (0)  H
~
Z
i t if i t0 D (0)  0 0  (0) E 0
(1)

c (t) =
e
f  H (t )  i
dt
~ t0
and thus
[1]

(0)

(1)

cf (t) = cf + cf (t) = f i

i
~

eif i t

t0



D
E
 0 0  (0)
f (0)  H
(t )  i
dt0

Interaction (Dirac) picture

See the document Pictures of Quantum Dynamics for a more complete treatment of this topic. The following is a summary.

Operator approach/Dyson series

Recall the Schwinger-Tomonaga equation:
i~

d
I0 (t)U
I (t, t0 )
UI (t, t0 ) = H
dt

which has formal solution

I (t, t0 ) = 1 i
U
~

t0

14

I0 (t0 )U
I (t0 , t0 )dt0
H

This may be solved iteratively:

i
I (t, t0 ) =
U
1
~

I0 (t0 )U
I (t0 , t0 )dt0
H

t0

!
Z t0
i
0
00
00
00
(t )U
I (t , t0 )dt dt0
1
H
I
~
t0
t0
  2 Z t Z t0
Z t
i
0 (t0 )dt0 + i
I (t0 )H
I (t00 )U
I (t00 , t0 )dt00 dt0
=
1
H
H
~ t0 I
~
t0 t0
  2 Z t Z t0
Z
i t 0 0 0
i
I (t0 )H
I (t00 )dt00 dt0 + . . .

=1
H (t )dt +
H
~ t0 I
~
t0 t0
 


Recall that the transition probability between any two states i(0) and f (0) is given by


E
D

 (0)
c(t) = f (0)  U
I (t, t0 )  i
i
=
1
~

0 (t0 )
H
I

(0) gives
Expanding both sides over a complete set of basis states of the Hamiltonian H


E
D

 (0)
c(t) = f (0)  U
I (t, t0 )  i

*
!
+
  2 Z t Z t0
Z t


i
i


0 (t0 )dt0 +
I (t0 )H
I (t00 )dt00 dt0 + . . .  i(0)
= f (0) 
1
H
H


~ t0 I
~
t0 t0
0


2Z tZ t D





E
D
E i Z tD
E
i
 0 0  (0)

0
(0)  0 0 0 00  (0)
f (0)  H
(t
)
i
dt
+
= f (0)  i(0)
f
H
(t
)
H
(t
)
i
dt00 dt0 + . . .



I
I
I
~ t0
~
t0 t0
  2 Z t Z t0 D
Z



E
E
i
i t D (0)  (0) 0
 (0)
 0 0 0 00  (0)
0 0 (0) 0
0
dt +
dt00 dt0 + . . .
f  US (t , t0 ) HS (t )US (t , t0 )  i
f (0)  H
= f i
I (t )HI (t )  i
~ t0
~
t0 t0
 2 X Z t Z t0 D
Z




E
ED
i
i t if i t0 D (0)  0 0  (0) E 0
 0 00  (0)
 0 0  (0)
dt +
dt00 dt0 + . . .
= f i
e
f  H (t )  i
n(0)  H
f (0)  H
I (t )  i
I (t )  n
~ t0
~
t0 t0
n
 2 X Z t Z t0
Z t






D
D
E
E
D
E
0
0
i
i
 0 0  (0) ini t00 (0)  0 00  (0)
 0 0  (0)
= f i
eif n t f (0)  H
eif i t f (0)  H
(t )  n
(t )  i
e
dt0 +
n  H (t )  i
dt00 dt0 + . . .
~ t0
~
t
t
0
0
n
Thus it is possible to see how this perturbation expansion generalizes the Dirac method and gives a method to find higher-order
terms.

Sudden approximation
(0) describes
The sudden approximation is one of two treatments for Hamiltonians with extreme time dependence. Suppose H
0 (t) with
some system with a characteristic time-scale 0 . Now suppose that the system is perturbed by some
Hamiltonian
H

(0) , n(0) . In the limit that  0 ,
proper time-scales on the order of . Suppose that the system is in an eigenstate of H
the perturbation turns on at such a rate that the system cannot
respond. The system is no longer in a stationary state of

=H
0 + H
0 ; therefore, the state i(0) may be expanded in a linear combination of the eigenstates
the total Hamiltonian H
of this combined Hamiltonian, and thus the time-evolution becomes non-negligible.

Adiabatic approximation
0 (t) where the characteristic time scales of the perturbation are
The other extreme of time-dependence is a perturbation H
much slower than that of the underlying exact system, 0  .

Harmonic perturbation
Stimulated EM radiation emission/absorption (Fermis golden rule)

Scattering
Lippmann-Schwinger equation
The Lippmann-Schwinger equation is a formal solution of the Schrodinger equation for the scattering states of a potential
V (~x). It consists of an integral over a scattering state |i along with a Green function which is essentially an inverted
15

Schr
odiner equation:
1
+ i
EH
As with the Laplace equation, any general solution with this Green function is
 E 2m Z
1

|i = ~k + 2
V (~x0 ) |i d3 x0
+ i
~
EH
 E

Note that ~k is a state which is a solution of the homogeneous Schrodinger equation and thus may be added to any general
solution without consequence. Now project the above equation onto the position basis:
 
D  E 2m Z  

1
~
0 


~x 
h~x | i = ~x  k + 2
V (~x ) 

~
E H + i
 =
G

Born approximation
The Lippmann-Schwinger equation is really only a formal solution since it contains the scattering wavefunction within the
kernel. The Born approximation is a first order perturbative solution to this equation. It consists of replacing the scattering
state with the incident state ~k :

Partial wave expansion

Phase shifts
Optical theorem
Resonances

Dirac equation
Path integrals
Basic relations
(tf , ti ) or propagator. Recall
The most important operator (beside the Hamiltonian) is the unitary time evolution operator U

that for a Hamiltonian, H(t), this may be written as

(tf , ti ) = e ~i
U

R tf
ti

0 )dt0
H(t

to write the propagator as a function:

It is possible, given a complete set of basis states {|ni} of the Hamiltonian H,
X
i
U (tf , ti ) =
|ni hn| e ~ En (tf ti )
n

However, this may not always be tenable, since solving the Schrodinger equation for its eigenstates may be difficult (if
not outright impossible, as with most time-independent Hamiltonians). Another approach is to obviate the need for the
eigenstates and calculate the propagator directly with the path integral formulation. This formulation focuses on discretizing
the path between two points (whether that be in position, or configuration space, phase space, etc., depends on the type of
path integral), calculating the propagator between the discrete points, and then taking the limit as the number of discrete
intermediate points approaches infinity.

Path integration measures

There are three different integration measures to describe the sums-over-histories, each related to different bases over which
the propagator may be expanded. These are the configuration space measure, the phase space measure and the coherent
states space measure.
Configuration space:
Z
Z
 m  12
 m  12
N 1
( ) Dx = lim
n=1
( )
dxn
N 2i~
2i~

16

Note that the number of integrals increases with N . The propagator written with this measure becomes
Z
Rt
0
i

0
Dx
U (x, t; x , 0) = e ~ 0 L(x,x)dt
This may be interpreted as the propagator written as an integral of the exponential of the classical action (integral of the
Lagrangian over time) over the space of all possible paths linking points x0 and x.
Phase space:
Z
Z
Z
dpn
N 1
N
( )
(. . . ) DxDp = lim n=1 m=1
...
dxm
N
2~

An important note is that there are 2N 1 integrals, N dp integrals and N 1 dx integrals. The propagator written with
this measure becomes
Z
Rt
0
i

x))dt

DxDp
U (x, t; x0 , 0) = e ~ 0 (xpH(x,
Note that the action is written as the integral over time of the Lagrangian written in terms of the Hamiltonian. For this
reason, the phase space integral is also known as the Hamiltonian path integral

Coherent states
Coherent state space:
Some examples
Free space propagator:
U (x, t; x0 , 0) =

 m  12 im(xx0 )2
e 2~t
2~it

Harmonic oscillator propagator:

2

U (x, t; x0 , 0) = A(t)e 2~ sin t ((x

im

Imaginary time path integrals/Partition functions

Non-classical path integrals/Grassmann variables

17

+x02 ) cos t2xx0 )