Professional Documents
Culture Documents
Klein Gordan Equation
Klein Gordan Equation
k =
w
c
?~k
(3.1)
(3.2)
"
a ?k
or
k
mc 2#
= 0
+ h
2
k2 = kk = mc
h
40
(3.3)
(3.4)
mc 2
2
~
(k ) = k + h
s
mc 2
0
2
~
k = k + h
0 2
(3.5)
This means that the Klein-Gordon equation allows negative energies as solution. Formally,
one can see that from the square of Eq. (2.60) the information about the sign is lost.
However, when starting from Eq. (2.71) all solutions have to be considered, and there is
the problem of the physical interpretation of negative energies.
The second problem with the Klein-Gordon equation is less obvious. It occurs when interpreting the function (x) as probability amplitude. Interpretation of (x) as probability
amplitude is only possible if there exists a probability density (x) and a current ~j (x)
that fulll a continuity equation
@ +r
~ ~j = 0 ;
@t
(3.6)
j 0(x) := c(x) !
0
j (x) := ~jj ((xx))
and obtain the covariant form
(3.7)
@ j = @ j = 0 :
@x
(3.8)
Eqs. (3.6) and (3.7) correspond in form and content the charge conservation in electrodynamics.
Non-relativistically one has
NR =
~jNR = h [
(2mi)
41
$
r
(3.9)
and thus one expects in the relativistic case also bilinear expressions in for and ~j . If
one denes a density according to (3.9) with the solution (3.1), it is easy to show that
this density does not fulll a continuity equation. This has to be expected since j has
to be a four-vector so that (3.7) is valid in all Lorentz systems. Thus, it is obvious to
generalize (3.9) to
h
j := 2im
where
@$
(3.10)
A @ B := A (@ B ) ? (@ A)B :
(3.11)
Consider
h @ @$
@ j = 2im
h [(@ )(@ ) + (@ @ ) ? (@ @ ) ? (@ )(@ )]
= 2im
h [ (2 ) ? (2 ) ]
(3.12)
= 2im
If fullls the Klein-Gordon equation, the right-hand side of (3.12) vanishes, and the
continuity equation (3.7) holds. However, the four-vector dened in (3.10) contains the
second problem:
= 1c j 0
h @$0
= 2imc
"
#
i
h
@
@
= 2mc2
(3.13)
@t ? @t
can be positive or negative, depending on the values of and
@ .
@t
Since the Klein-Gordon equation denotes a partial dierential equation (2nd order) of
hyperbolic type, one has the option to arbitrarily choose the functions
@ (~x; t = 0)
(~x; t = 0) and @t
(3.14)
at the starting time (t = 0), and thus obtain, e.g., negative values for (~x; t = 0).
An interpretation of as probability density would mean that the theory allows negative
probabilities. This is the problem of the indenite probability density.
42
The solution of both problems has an interesting historical development, which can brie
y
be summarized as follows:
1928: Dirac "invents" the Dirac equation. The probability density is positive;
however, negative energies are allowed (Proc. Roy. Soc. A117, 610-628).
1930: Dirac solves the problem of negative energies via the "hole" theory. Antiparticles are related to negative energy eigenstates (Proc. Cambridge Phil. Soc.
26, 376-381).
1934:
In the following, we concentrate on the rst two topics, the last two are subjects of
quantum eld theory.
43
E 0 = E ? mc2
(3.17)
(3.18)
Now consider
!
@ = @' ? i mc2 ' exp(? i mc2 t) ?i mc2 ' exp(? i mc2 t)
@t
@t
h
h
h
! h
2
2
@ = @ @' ? i mc ' exp(? i mc2t)
@t2
h
h
#
"@t @t2
2
2 4
mc
m
c
@'
@'
mc
?i h @t ? i h @t ? h 2 ' exp(? hi mc2t)
"
#
2
2 4
mc
m
c
@'
= ? 2i h @t + 2 ' exp(? hi mc2 t)
h
Inserting this result in (3.15) gives
"
"
(3.19)
2
@' + m2 c2 ' ' exp(? i mc2 t) = 4 ? m2 c4 exp(? i mc2t)
(3.20)
? c12 i 2mc
h @t h 2
h
h
h 2
from which follows
2
h
@'
(3.21)
ih @t = ? 2m r2':
This is the free Schrodinger equation for spinless particles. Since the type of particle
described by a wave equation does not depend upon whether the particle is relativistic or
nonrelativistic, the Klein-Gordon equation describes spin-0 particles.
44
(3.22)
?!
@
@
ie
h
?
(3.23)
= 2mc2
@t ? @t
denes the charge density, and
j0 = ie2mh ( ? r ? r ? )
(3.24)
denes the charge-current density. The charge density (3.23) is allowed to be positive,
zero, or negative. This is consistent with the existence of particles and antiparticles in
the theory.
0
Let us now calculate the solution for free particles. A possible ansatz for the solution
for a free wave is given in (3.1)
(x) = a exp ? hi px
(3.25)
where 1=h p = k . We had seen that because of (3.7) there exist two possible solutions
for a given momentum p, one with positive and one with negative energy. Consequently
i (p x jE jt));
(3.26)
p
(x) = a exp( h
p
where Ep = c p2 + m2 c2. Inserting this into the expression for the density (3.23) gives
jEpj
= emc
2
:
(3.27)
This suggests the following interpretation: + species particles with charge +e and ?
species particles with the same mass, but with charge ?e. The general solution of the
wave equation will always be a superposition of both types of functions.
45
Let us now discretize the continuous plane wave by conning the wave to a cubic box
(normalization box) with box length L and demand periodic boundary conditions at the
box walls. This leads to
i
(3.28)
n() = An() exp (pn x Epnt) ;
h
where
(3.29)
pn = 2L n; n = (n1 ; n2; n3) ; ni 2 N
and
q
(3.30)
Epn = c p2n + m2 c2 En :
Here n is a (discrete) vector in the lattice space with axes n1 ; n2; n3. With (3.27) the
normalization factors An() are determined by the requirement that
eEn jA j2L3 :
3
d
x
(
x
)
=
3
mc2 n()
(3.31)
2
An() = Lmc
3E
n
and thus
(3.32)
For neutral particles the Klein-Gordon eld has to be real, see (3.23), i.e.
Thus the wave for a neutral particle can be constructed as
1
p
(
p
)
+
(
?
p
)
=
n
n(+) n
n(?)
n(0)
s2 2
?i
i
mc
= 2L3 E exp h (pn x ? Ent) + exp h (pn x + Ent)
s 2n
= 2Lmc3 E 2 cos h1 (pn x ? Ent) :
n
46
= .
(3.35)
Thus, with n(0) = n?(0) follows 0 = 0 and j0 (x; t) = 0. Consequently, in this case there
is no conservation law.
The previous showed that relativistic quantum theory leads to new degrees of freedom
of particles. In a nonrelativistic theory, free spinless particles can propagate freely with
a well dened momentum p. In the relativistic case of free spinless particles, there exist
three solutions, which correspond to the electric charge (+,-,0) of the particles, for every
momentum p.
The coupled equations can be written in a more compact form. For this we introduce
!
'
=
(3.45)
48
and make use of the Pauli matrices, ^i , 1. However, here they do not act in spin-space,
but rather in the space introduced by (3.45). With this the coupled equations (3.41) can
be combined in a Schrodinger type equation
ih @@t = H^ f ;
(3.46)
where the Hamiltonian H^ f for free particles is given by
2
H^ f = (^3 + i^2 ) 2p^m + ^3 mc2
!
! 2
p
^
1
0
1
1
+ 0 ? 1 mc2 :
(3.47)
= ?1 ? 1
2m
Each component of the vector individually satises the Klein-Gordon equation. From
(3.46) follows
!
!
@
@
ih @t + H^ f ih @t ? H^ f = 0
!
2
@
?h 2 @t2 ? H^ f2 = 0
(3.48)
!
2
@
2 2
2 4
(3.49)
?h @t2 ? h c 4 ? m c = 0;
which is just the Klein-Gordon equation valid for each component of . With this representation, the expression for the density becomes especially simple. From (3.23) using
(3.40) one obtains
?!
ie
h
@
@
0
?
= 2mc2
@t ? @t
2
?
?
?
= 2emc
mc2 ( (' ? ) + (' ? ))
= 2e (('? + ?)(' ? ) + (' + )('? ? ? ))
= 2e ('?' ? ?)
= e y^3 :
(3.50)
Similarly, the current vector is given in Schrodinger representation as
h y^ (^ + i^ )r ? (r y)^ (^ + i^ ) :
(3.51)
j0 = 2emi
3 3
2
3 3
2
2
49
d
Z
x y)^
3
d3x('? ' ? ? ) = 1
'0 and 0 are therefore determined by the solution of the coupled equations
!
2
2
p
^
E ? 2m ? mc2 '0 ? 2p^m 0 = 0
!
p^2 ' + E + p^2 + mc2 = 0
0
2m 0
2m
(3.52)
(3.53)
(3.54)
(3.55)
(3.56)
Since the determinant of the system (3.56) needs to vanish, it follows that
!2
!
2
2 2
p
^
p
^
E 2 ? 2m + mc2 + 2m = 0;
from which one recoveries the relativistic energy momentum relation
q
E = c p2 + mc2 Ep :
(3.57)
(3.58)
!
!
(+)
i
'(+)
'
(
p
)
0
exp h (p x ? Ept) (+) (p)
(+)
0
50
(3.59)
(3.60)
!
!
2
mc
+
E
'(+)
p
0
= mc2 ? E :
(3.61)
p
(+)
0
Eq. (3.52) allows to calculate the normalization constant A(+) from
Z
? '(+) ? (+)? (+) ) = jA j2 L3 h(mc2 + E )2 ? (mc2 ? E )2 i = 1 (3.62)
2
jA(+) j d3x('(+)
p
p
(+)
0
0
0
0
If the phase is chosen to be real, then
1q
:
4mc2 L3 Ep
A(+) = p
(3.63)
In the other case, E = ?Ep , proceeding similarly leads to the wave function
(p) = A(?)
(?)
with
!
!
'(0?) exp i (p x + E t) '(?) (p)
p
(?) (p)
h
(0?)
!
!
'(0?) = mc2 ? Ep
mc2 + Ep
(0?)
A(+) '(+)
0
A(+) (+)
0
Similarly
(3.65)
(mc2 + Ep)=q4Epmc2 A
= p1 3 @
L (mc2 ? Ep)= 4Epmc2
!
2
2
1
2
mc
=
2
mc
p 3 (? p2 )=2mc2
L
2m
!
!
1
v=c!0 1
1
1
= p 3 ?1=4(v=c)2 ?! p 3 0
L
L
!
!
!
!0 1
A(?) '(0?) p1 ?1=4(v=c)2 v=c
0
?! p 3 1 :
1
A(?) (0?)
L3
L
51
(3.64)
(3.66)
(3.67)
Thus, we see that in the nonrelativistic limit for states with positive charge the upper
component is large, and the lower one is small. For negative energies, we nd the reverse
being true.
Charge conjugation follows in this case from comparing the expressions for the wave
functions in 3.59 for (+) (p) with the corresponding expression for (?) (p), 3.64.
!
!
(?)
(+)?
'
(
?
p
)
(
p
)
(p = (?) (?p) = '(?)? (p) = 1 (+)? (p):
(?)
Thus, if
'
represents a positive charge, then the state
?
C^ C^ ?1 C = 1 ? = '?
(3.68)
(3.69)
(3.70)
! (0?)
'(+)
0
(+)
! '0(?)
0
p ! ?p
+Ep ! ?Ep
(3.72)
(3.74)
so that
2 = 1 ; = 1:
(3.75)
Thus there exist two dierent kinds of neutral particles. To resolve this, one introduces a
new quantum number, the charge parity . One has then
(a) Neutral particles with positive charge parity ( = +1)
C = 1 ? = or '? =
(b) Neutral particles with negative charge parity ( = ?1)
C = 1 ? = ? or '? = ?
First, consider the eect of U^ applied on the states (+)(p) of (3.59) and (?) (p) of (3.64)
(+) (p) = U^ (+) (p)
!
2
1
1
i
mc
+
E
p
^
= Up 3q 2
exp h (p x ? Ept)
L 4mc Ep mc2 ? Ep
!
!
2
2
2
i
1
1
(
mc
+
E
)
?
(
mc
?
E
)
mc
+
E
p
p
p
= p 3q 2
mc2 ? Ep exp h (p x ? Ept)
L 4mc Ep ?(mc2 ? Ep) (mc2 + Ep)
!
1
1
(3.79)
= p 3 0 exp i (p x ? Ept) :
h
L
Similarly,
(3.80)
So, U^ is the desired transformation. However, U^ is not unitary in the usual sense (i.e.
U^ ?1 6= U^ y), since
h 2
i
2
^ 3 = 1q 1
U^ ?1 = 3 U
(
mc
+
E
)
+
(3.81)
p
1 (mc ? Ep ) :
4mc2Ep
By explicit calculation we see that
U^ U^ ?1 = 1:
(3.82)
The normalization of the () -functions follows from the normalization of () (3.52) as
Z
1 = d3x y3
Z
= d3x(U^ ?1 )y3 (U^ ?1 )
Z
= d3xy(U^ ?1 )y3 U^ ?1
Z
= d3xy3
^ 3 and thus
Here we used that from (3.81) follows 3 U^ ?1 = U
^ 3 = U^ ?1 U
^ 3 = 3
(U^ ?1 )y3U^ ?1 = (U^ ?1 )yU
54
(3.83)
(3.84)
h j 0i
so that from (3.83) follows
d3x y3
(3.85)
h j 0i = hj0i;
(3.86)
i.e. the generalized scalar product is invariant under the transformation U^ . It seems
natural to call an operator A^ with the property
^ jA
^ 0 i
h j 0i = hA
(3.87)
(3.93)
(3.94)
and thus
Integration gives
(3.95)
(3.96)
!
!
@
1
1
ih @t = Ep ?
2
2
(3.97)
1
ih @
@t = Ep1
2
ih @
@t = ?Ep2 :
(3.98)
1 = N1 exp ? hi Ept
2 = N2 exp hi Ept :
(3.99)
Here N1 and N2 are normalization constants which are determined by the normalization
condition (3.83) as
Z
Z
3
y
d x 3 = d3xy3 = 1
(3.100)
yielding
(3.101)
jN1 j2 ? jN2j2 = V1 :
This lead to the two independent solutions
!
1
i
1
(+)
= p 0 exp h p x ? Ept
V
!
1
0
(?)
= p 1 exp hi p x + Ept
V
56
charge
+1
charge
?1
(3.102)
n1 (+) + n2 (?)
(3.103)
with jn1j2 ? jn2j2 = 1 being a normalized eigenfunction of the momentum p with charge
+1 and each linear combination with jn1j2 ? jn2j2 = ?1 being a normalized solution with
charge -1.
ih @@t = H^ f
with
(~x; t) =
(~x; t)
(~x; t)
(3.104)
(3.105)
(3.106)
With the help of (3.104) the vector (~x; t) may be evaluated at any later time t, if the
values (~x; 0) are known at t = 0. This can be expressed by the transformation
with
!
!2 2
i
^
^
i
H
i
H
t =
f
f
U^ (t) = exp ? h H^ f t = 1 + ? h t + ? h
2!
(3.107)
(3.108)
(Heisenberg picture). The change from the Schrodinger to Heisenberg picture is performed
by the transformations
H (~x) = U^ (t) (~x; t)
F^H (t) = U^ ?1 (t) F^ (0) U^ (t)
which leave the scalar product invariant.
h (~x; t) jF^ (0)j 0(~x; t))i = hU^ (t) H (~x jF^ (0)j U^ (t) 0H (^x)i
= h H (^x) jU^ ?1 F^ (0)j 0H (^x)i
= h H (^x) jF^H (t)j 0H x^)i :
(3.109)
(3.110)
^
ih ddtF = ih dtd eiH^ f t=h F^ (0) e?iH^ f t=h
^ H^ f ]
= ?H^ f F^ + F^ H^ f = [F;
(3.111)
in analogy to non-relativistic quantum mechanics. From (3.112) follows that the physical
observation F whose corresponding operator F^ commutes with H^ f , are constants of motion. This means that the expectation values of these operators are constant in time. One
of the basic postulates in non-relativistic quantum mechanics states that the eigenvalues
of an operator describe the measurable values of the corresponding physical quantity in
a state of the system. To satisfy this postulate in the relativistic theory, we must modify
the denitions of some of the operators.
Let us consider the energy of a system. The eigenvalues and eigenstates of the operator
H^ f (3.106) are determined (in case of the free motion with momentum ~p) by
H^ f = E :
The free motion had two solutions (see previous section)
58
(3.112)
!
'0 exp i (p~ ~x) ; = 1
0
h
(~x) = p1 3
2
(3.113)
q
E = Ep = c p2 + mc2 :
(3.114)
E?1 is negative and thus can not be interpreted as a one-particle energy, which must
always be positive. Here we need to remember the double meaning of the energy eigenvalues of the Hamiltonian in non-relativistic quantum mechanics: First, they represent
the energy of stationary states. Second, they characterize the time-evolution of the wave
function. We have already seen (3.53) that the eigenvalues E of H^ fh represent
i the time
i
dependence of the wave functions in a relativistic theory (factor exp ? h Ept ), i.e.,
i
i
^
(~x; t) = exp ? h Hf t (~x) = exp ? h Ept (~x) :
(3.115)
The energy of these states is always positive and hence independent. We can see this
from the following. The energy " of a system in a stationary state is identical with the
mean value of the energy, i.e.,
" =
d3x y ^3 H^ f :
(3.116)
R
With H^ f = E = Ep and d3x y^3 = , we have
" = Ep
d3x y ^3 = 2 Ep = Ep :
(3.117)
The energy is always positive and independent of . Thus, the problem of the energy is
solved. The dual character of the eigenvalues of H^ f , i.e., as the characteristic factor of
the time-evolution and as an energy, evolves quite naturally in the relativistic quantum
theory. We can give the correct interpretation of the energies of the states by making use
of the canonical formalism, the energy operator is not H^ f but ^3 H^ f (3.116)
In non-relativistic quantum mechanics, there is always a correspondence between a relation of operators and that of classical objects (measurable values). Example: Newton's
equation of motion corresponds to the operator equation
59
dp^ = 1 [^p; H^ ]
dt
ih
(3.118)
d~x = i [^x ; H^ ] = ~p ;
dt
ih 1
m
(3.119)
which corresponds to the classical relations between velocity and linear momentum. Since
these operators equations are of the same form as the classical equation, it is ensured that
the quantum mechanical expectation values satisfy the classical equations of motion. In
relativistic quantum theory, the situation is dierent. E.g., the expression for the "velocity
operator" of a relativistic spin-0 particle was given by
(3.120)
dx = p = c2 p = c2p ;
dt
M
Mc2
E
(3.121)
v
!
u
2
2
2
2
u
mc
mc
v
v
2
t
E = Mc = q
= q
+ 1 ? c2
1 ? vc22
1 ? vc22 c2
v
u
u
m2 v2 + m2 c2 (1 ? vc22 )
t
= c
1 ? vc22
v
!2
u
q
u
mv
t
2 c2 = c
p2 + m 2 c 2
+
m
= c
(1 ? vc22 )1=2
(3.122)
is the total energy of a free particle with rest mass m. Obviously, the right-hand side of
(3.120) is dierent from the left-hand side. Furthermore, the eigenvalues of the matrix
^3 + i^2 =
1 1
?1 ?1
60
(3.123)
are zero. This means that the eigenvalues of the velocity operator (3.120) are also zero.
Hence, we again notice that, in general within a relativistic theory, the expectation values
of a reasonably constructed operator are not the same as the values of the corresponding
classical quantity. Thus, we conclude that not all operators of the non-relativistic theory
can be transferred to a relativistic one-particle theory. The reason for this is the restriction
to the one-particle concept. In relativistic quantum mechanics, the consistency of the oneparticle description is limited. To be mathematically precise, the formulation of the
relativistic theory within a one-particle concept implies the condition that the only valid
operators are those which do not mix dierent charge states. Such operators are called
even operators or true one-particle operators.
More formally, an operator O^even is called even if
O^even = 0() ;
(3.124)
where 0() are functions with positive and negative frequencies. Similarly, an operator
O^odd () is called odd if
(3.125)
Thus, the Hamiltonian of the free Klein-Gordon equation in the Schrodinger representa~ are even operators.
tion, H^ f , and the momentum operator p^ = ?ih r
In general, an operator can be split into an even and odd part
O^ = O^even + O^odd ;
(3.126)
thus, one can separate from any given operator O^ a true one-particle operator O^even. In
matrix notation
O^ =
a^11 0
0 a^22
0 a^12
a^21 0
61
= O^even + O^odd :
(3.127)
(3.133)
Substituting the positive and negative energy solutions (3.23) into (3.133) gives
i
h
(3.134)
(Ep e)2 () (x; t) = (^p eA)2 + m2 ()(x; t)
Again, once can use (3.134) as starting point and use it with more general potentials V
and A. For example, let A = 0 and V = e, i.e. allow only a scalar potential V . Then
(3.134) gives
(E 2 + V 2 ? 2EV ) = (^p2 + m2)
(3.135)
62
Substituting the relation E 62 = k2 + m2 between energy and wave vector and using
p^ ! ?ir leads to
(r2 + k2) = (2EV ? V 2) ;
(3.136)
which looks like a Schrodinger equation with the equivalent energy dependent potential
V SE = 2EV2m? V
(3.137)
Another type of potential to consider is the Lorentz scalar, which adds to the mass, since
p p = m2 . The KG equation with coupling to the scalar potential is
i
h
(3.138)
E 2 = p^2 + (m + S )2
63