Dong S.-H. Wave Equations in Higher Dimensions

Подождите немного. Документ загружается.

214 15 The Levinson Theorem for Dirac Equation

If α

< 0, there is an inﬁnite number of bound states. We will not discuss this case

as well as the case with α = 0 here. When α

> 0, we take α>0 for convenience.

Some formulas given in the previous sections will be changed.

When E ≤ M we have

(r, λ) = e

i(α+1)π/2

2M(πκr/2)

1/2

(1)

(iκr),

(r, λ) = e

i(α+1)π/2

κ(πκr/2)

1/2



−

d(κr)

(1)

(iκr) +

K−1/2

κr

(1)

(iκr)



(15.45)

Hence, the ratio at r =r

for E =M is

(r, λ)



r=r

2Mr

K +α −1/2

,E=M. (15.46)

When E>Mwe have

(r, λ) = 2M(πkr/2)

1/2

{cosη

(E, λ)J

(kr) −sinη

(E, λ)N

(kr)},

(r, λ) = k(πkr/2)

1/2



cosη

(E, λ)



−

d(kr)

(kr) +

K −1/2

(kr)



−sin η

(E, λ)



−

d(kr)

(kr) +

K −1/2

(kr)



(15.47)

When (kr) tends to inﬁnity, the asymptotic forms of the solutions are given by

(r, λ) ∼2M cos(kr −απ/2 −π/4 +η

(E, λ)),

(r, λ) ∼k sin(kr −απ/2 −π/4 +η

(E, λ)).

(15.48)

In comparison with the solution (15.22) we obtain the phase shift δ

(E, λ) for

E>M:

(E, λ) =η

(E, λ) +(K −α −1/2)π/2,E>M. (15.49)

From the matching condition (15.8), for the sufﬁciently small k we obtain

tanη

(E, λ) ∼

−π(kr

/2)

2α

(α +1)(α)



K −α −1/2

K +α −1/2



(M, λ) −2Mr

/(K −α −1/2)

(M, λ) −2Mr

/(K +α −1/2)

. (15.50)

Therefore, as λ increases from zero to one, each time the A

(M, λ) decreases from

near and larger than the value 2Mr

/(K +α −1/2) to smaller than that value, the

denominator in Eq. (15.50) changes sign from positive to negative and the rest factor

keeps positive, so that η

(M, λ) jumps by π. Simultaneously, from Eq. (15.46)a

new overlap between the variant ranges of the ratio at two sides of r

appears such

that a scattering state of a positive energy becomes a bound state. Conversely, each

time the A

(M, λ) increases across that value, η

(M, λ) jumps by −π, and a bound

state becomes a scattering state of a positive energy.

5 Discussions on General Case 215

Second, we calculate the solutions with the energy E ∼−M.Let

β =(K

+K −2Mb +1/4)

1/2

=K +

. (15.51)

Similarly, we only discuss the cases with β

> 0, and take β>0.

When E ≥−M we have

(r, λ) =−e

i(β+1)π/2

κ(πκr/2)

1/2



d(κr)

(1)

(iκr) +

K+1/2

κr

(1)

(iκr)



(r, λ) = e

i(β+1)π/2

2M(πκr/2)

1/2

(1)

(iκr).

(15.52)

Hence, the ratio at r =r

for E =−M is

(r, λ)



r=r

=−

K −β +1/2

2Mr

,E=−M. (15.53)

When E<−M we have

(r, λ) =−k(πkr/2)

1/2



cosη

(E, λ)



d(kr)

(kr) +

K +1/2

(kr)



−sin η

(E, λ)



d(kr)

(kr) +

K +1/2

(kr)



(r, λ) = 2M(πkr/2)

1/2

·{cos η

(E, λ)J

(kr) −sinη

(E, λ)N

(kr)}.

(15.54)

When (kr) tends to inﬁnity, the asymptotic forms for the solutions are:

(r, λ) ∼k sin(kr −βπ/2 −π/4 +η

(E, λ)),

(r, λ) ∼2M cos(kr −βπ/2 −π/4 +η

(E, λ)).

(15.55)

In comparison with the solution (15.22) we obtain the phase shift δ

(E, λ) for

E<−M:

(E, λ) =η

(E, λ) +(K −β +1/2)π/2,E<−M. (15.56)

From the matching condition (15.8), for the sufﬁciently small k we obtain

tanη

(E, λ) ∼

−π(kr

/2)

2β

(β +1)(β)

(−M,λ) +

K+β+1/2

2Mr

(−M,λ) +

K−β+1/2

2Mr

. (15.57)

Therefore, as λ increases from zero to one, each time the A

(−M,λ) decreases

from near and larger than the value −(K −β +1/2)/(2Mr

) to smaller than that

value, the denominator in Eq. (15.57) changes sign from positive to negative and

the rest factor keeps negative, so that η

(−M,λ) jumps by −π . Simultaneously,

from Eq. (15.53) an overlap between the variant ranges of the ratio at two sides

216 15 The Levinson Theorem for Dirac Equation

of r

disappears such that a bound state becomes a scattering state of a negative en-

ergy. Conversely, each time the A

(−M,λ) increases across that value, η

(−M,λ)

jumps by π, and a scattering state of a negative energy becomes a bound state.

In summary, we obtain the modiﬁed relativistic Levinson theorem for non-critical

cases when the potential has a tail (15.40) with n =2 at inﬁnity:

(M) +δ

(−M) =n

π +(2K −α −β)π/2. (15.58)

We will not discuss the critical cases in detail. In fact, the modiﬁed relativistic

Levinson theorem (15.58) holds for the critical cases of α>1 and β>1. When

0 <α<1or0<β<1, η

(M, 1) or η

(−M,1) in the critical case will not be

multiple of π, respectively, so that Eq. (15.58) is violated for those critical cases.

Furthermore, for the potential (15.40) with a tail at inﬁnity, when n>2, even if

it contains a logarithm factor, for any arbitrarily small positive , one can always

ﬁnd a sufﬁciently large r

such that |V(r)|</r

in the region r

<r<∞. Thus,

from Eqs. (15.44) and (15.51) we have for the sufﬁciently small 

α =(K

−K ±2M +1/4)

1/2

∼K −

β =(K

+K ∓2M +1/4)

1/2

∼K +

(15.59)

Hence, Eq. (15.58) coincides with Eq. (15.33). In this case the Levinson theorem

(15.33) still holds for the non-critical case.

6 Friedel Theorem

The Levinson theorem is closely related to the Friedel theorem [452] in three di-

mensions described by

N =

∞



l=0

(2l +1)δ

). (15.60)

It states that the change of the number of states N around a potential barrier can

be expressed as the Friedel sum rule [453], which sets up the relation between N

and the phase shifts at the Fermi energy. It provides a powerful method in calcu-

lating some of the important properties of electron structure. Recently, this theorem

has been generalized to the system of Dirac fermions [454, 455] in two and three

dimensions as well as in arbitrary dimensions [456].

AsshowninRef.[456], the generalized Friedel theorem for Dirac fermions in D

dimensions is given by

N =

⎧

⎪

⎨

⎪

⎩



±K



∞

l=0



D+l−2



·([δ

) −δ

(M) +δ

(−E



) −δ

(−M)]

+

(−1)

|K|

[sin

(−M)−sin

(M)]), D =2d +1,



±K



∞

l=0

d−1



D+l−2



·[δ

) −δ

(M) +δ

(−E



) −δ

(−M)],D=2d,

(15.61)

7 Comparison Theorem 217

where δ

)[δ

(−E



)] and δ

(±M) are the phase shifts of scattering states at

Fermi energy E

=−E



) for Dirac fermions (antifermions) and the crit-

ical points E

=±M of zero momentum with ±M being effective mass of fermion

(antifermion). The notation



±K

denotes the summation over the quantum num-

bers K =±|K|, and 

=+1 (−1) for K>0 (K < 0). The total angular momen-

tum K =±(2l +D −1)/2.

7 Comparison Theorem

Before ending this Chapter, we are going to give a brief sketch of the comparison

theorem for Dirac equation in arbitrary dimensions due to its interest. For the Dirac

problems, this theorem has been discussed by some authors [457–461].

We consider two different attractive potentials V

(r) and V

(r) satisfying

(r) < V

(r). If we write the corresponding pairs of radial wavefunction as

} and {F

}, then in terms of Eqs. (15.1)wehave

(r)

(r) =[E

−V

(r) −M]F

(r), (15.62)

−

(r)

(r) =[E

−V

(r) +M]G

(r), (15.63)

(r)

(r) =[E

−V

(r) −M]F

(r), (15.64)

−

(r)

(r) =[E

−V

(r) +M]G

(r). (15.65)

The difference between Eq. (15.62) multiplied by F

(r) and Eq. (15.65) multiplied

by G

(r) leads to the following formula

(r)F

(r)]=[−M −V

(r) +E

(r)F

(r)

+[−M +V

(r) −E

(r)G

(r). (15.66)

In a similar way, the difference between Eq. (15.64) multiplied by F

(r) and

Eq. (15.63) multiplied by G

(r) allows us to obtain

(r)G

(r)]=[−M −V

(r) +E

(r)F

(r)

+[−M +V

(r) −E

(r)G

(r). (15.67)

In terms of Eqs. (15.66) and (15.67)wehave

[(E

−E

) −(V

(r) −V

(r))][F

(r)F

(r) +G

(r)G

(r)]

=−

(r)G

(r) −G

(r)F

(r)]. (15.68)

By integrating this equation with respect to argument r ∈[0, ∞) and considering

the boundary conditions F(0) =G(0) =0 and F(∞) =G(∞) = 0, we have

218 15 The Levinson Theorem for Dirac Equation



∞

−E

)[F

(r)F

(r) +G

(r)G

(r)]dr



∞

(r) −V

(r))[F

(r)F

(r) +G

(r)G

(r)]dr. (15.69)

Therefore, two integrals have the same sign for the node-free wavefunctions

{F(r),G(r)}. It should be emphasized that the potentials and eigenvalues are both

real. On the contrary, if the potential parameters stray into a region such that the

corresponding energy level becomes complex, then Eq. (15.69) will no longer lead

to the expected result V

(r) < V

(r) ⇒ E

since the complex energy levels

cannot be well ordered.

Recently, Hall has supposed that the potential V =V(r,a)depends smoothly on

a parameter a and restated this theorem [460, 461]. For the real attractive central

potential V(r,a) and the corresponding discrete Dirac eigenvalue E(a) = E

(a),

one has



(a) ≥0, for

∂V

∂a

≥0,



(a) ≤0, for

∂V

∂a

≤0.

(15.70)

Moreover, if we suppose that E

and E

are Dirac eigenvalues corresponding

to two distinct attractive central potentials V

(r) and V

(r), then we have

(a)

≤E

(b)

, for V

(r) ≤V

(r). (15.71)

Finally, it should be pointed out that those attractive central potentials V

(r) and

(r) are time-independent.

8 Concluding Remarks

In this Chapter we have uniformly studied the Levinson theorem for the Dirac equa-

tion in (D + 1) dimensions by the Sturm-Liouville theorem. Thus, the Levinson

theorem have been established by proving the number of bound states to be equal

to the sum of the phase shifts of the scattering states at E =±M. The general case

where the potential V(r) has a tail at r>r

has also been analyzed. Finally, we

have discussed the Friedel and comparison theorems for the radial Dirac equation

in arbitrary dimensions D.

Chapter 16

Generalized Hypervirial Theorem for Dirac

Equation

1 Introduction

The recurrence relations for matrix elements are very useful in quantum calcula-

tions, but their direct computation are generally very cumbersome. An interest-

ing and useful recurrence formula is the Blanchard’s relation for arbitrary non-

relativistic matrix elements of the form n

, where the symbol |nl stands

for non-relativistic hydrogenic radial eigenstate.

As shown in Chap. 10, we have generalized the hypervirial theorem for the non-

relativistic equation case in arbitrary dimensions D. In this Chapter we attempt to

present the generalized hypervirial theorem for relativistic Dirac equation case.

This Chapter is organized as follows. In Sect. 2 we are going to derive the gen-

eralized hypervirial theorem for Dirac equation in off-diagonal case. Section 3 is

devoted to the diagonal case. We summarize our conclusions in Sect. 4.

2 Relativistic Recurrence Relation

Based on the study in Chap. 4, when the factor r

(1−D)/2

is built into each radial

function 

K,[j ]

(x,t) we may write out the relativistic radial Hamiltonian

=−iα





+Mβ +V(r), (16.1)

where

α ·r =



0 −1

−10



,β=



0 −1



=∓[j

+(D −2)/2]

(16.2)

Note that the ﬁrst derivative d/dr comes from the radial momentum operator p

=−i[d/dr +

(D −1)/2r] as given in Eq. (3.53). Due to the introduction of the factor r

(1−D)/2

, the second term

(D −1)/2r in p

is canceled each other.

S.-H. Dong, Wave Equations in Higher Dimensions,

DOI 10.1007/978-94-007-1917-0_16, © Springer Science+Business Media B.V. 2011

219

220 16 Generalized Hypervirial Theorem for Dirac Equation

with the property α

β =−βα

. For example, we have in three dimensions

K =κ =



−(l +1), j =l +

l, j = l −

(16.3)

and K =0 in one dimension.

As a result, the reduced radial Dirac Hamiltonian H can be written as a matrix

form



M +V(r) i(

−

)

)V(r)−M



(16.4)

with the property

ψ(r)=E

ψ(r), (16.5)

where we have introduced purely radial wavefunction

ψ(r)=



G(r)

iF(r)



. (16.6)

For bound states, they may be normalized by the relation

(ψ(r), ψ(r)) =



∞

(r) +F

(r)]dr =1. (16.7)

We use inner products without the radial measure r

D−1

because the factor r

(1−D)/2

is already built into each radial function as mentioned above.

The corresponding radial equations are written as

dG(r)

G(r) =[E −V(r)+M]F(r),

−

dF(r)

F(r)=[E −V(r)−M]G(r).

(16.8)

It is found that this set of radial equations are slightly different from those given in

Chap. 4 due to the exchange between M and −M. Such a modiﬁcation does not

affect the ﬁnal result at all.

By using the deﬁnition of the Hamiltonian H

given in Eq. (16.1), we have

f(r)−f(r)H

=−iα





(r) +



−

f(r)



, (16.9)

where 

∓

∓K

. Thus, acting this relation on the eigenfunctions yields

−E

)n

|f(r)|n

=−iα

n



(r) +



−

f(r)|n

, (16.10)

where the matrix elements of radial functions are calculated as follows:

n

|f(r)|n

=



f(r)[F

∗

(r)F

(r) +G

∗

(r)G

(r)]dr,

n

|βf (r )|n

=



f(r)[F

∗

(r)F

(r) −G

∗

(r)G

(r)]dr.

(16.11)

In a similar way, one is able to obtain the following relations

2 Relativistic Recurrence Relation 221

ξ =−





(r) +f



(r)

−



−



−







−







−

βf (r )



−iα

(V (r) −Mβ)





(r) +



−

βf (r )



(16.12)

and

ξH

=−





(r) −



−

βf (r )



−

βK





(r) −



−

βf (r )



−iα





(r) +



−

βf (r )



(Mβ +V (r)), (16.13)

where

ξ =H

f(r)−f(r)H

. (16.14)

The difference between (16.12) and (16.13) leads to

ξ −ξH

=−f



(r) +



−

βf (r )

(

−

)

f(r)−

2β



−

f(r)

βf



(r) +2iα

Mβ





(r) +



−

βf (r )



, (16.15)

from which we have

−E

)

n

|f(r)|n



=n

ξ −ξH



=n

|−f



(r) +



−

βf (r ) +

(

−

)

f(r)−

2β



−

f(r)

βf



(r) +2iα

Mβ





(r) +



−

βf (r )



, (16.16)

where the ﬁrst derivative d/dr in (16.16) can be substituted by the Hamiltonian

given in (16.1).

On the other hand, the addition between (16.12) and (16.13) allows us to obtain

the following off-diagonal matrix elements

−E

)n

|f(r)|n



=n

ξ +ξH



222 16 Generalized Hypervirial Theorem for Dirac Equation

=n

|−f



(r) −2f



(r)



−

βf (r ) +



−



f(r)

−2iα





(r) +



−

βf (r )



V(r)|n

. (16.17)

Due to the presence of Dirac matrices, we also need to calculate off-diagonal

matrix elements of expressions involving α

f(r)and βf (r ), i.e.,

(−iα

f(r))=−





(r) +f(r)



βf (r ) +iα

(Mβ −V (r))f (r) (16.18)

and

(−iα

f (r))H

=−f(r)





−iα

(Mβ +V (r))f (r). (16.19)

Summing and subtracting them lead to

)n

|−iα

f(r)|n



=n

(−iα

f(r))−iα

f(r)H



=n

|−f



(r) −2f(r)



−

βf (r ) −2iα

V(r)f(r)|n

,

(16.20)

−E

)n

|−iα

f(r)|n



=n

(−iα

f(r))+iα

f(r)H



=n

|−f



(r) +



βf (r ) +2iα

βM(r)f (r)|n

. (16.21)

In addition, one has

βf (r ) +βf (r)H

=−iα



−

f(r)+βf



(r)



+2(M +βV (r))f (r), (16.22)

from which we have

)n

|βf (r )|n



=n

|−iα



−

f(r)+βf



(r)



+2(M +βV (r))f (r)|n

. (16.23)

Equations (16.10)–(16.23) are the basic results of our problem. Even though we

might consider, as in the non-relativistic case, radial functions of the form f(r)=r

and insert the explicit expression for the Coulomb potential, we do not show them

for simplicity.

3 Diagonal Case 223

3 Diagonal Case

Let us study the diagonal case 

−

=0, i.e., K

. In this case, it is shown from

Eq. (16.10) that

−E

)n

|f(r)|n

=−iα

n



(r)|n

. (16.24)

Similarly, Eqs. (16.16) and (16.17) are simpliﬁed as

−E

)

n

|f(r)|n



=n

ξ −ξH



=n

|−f



(r) +

βf



(r) +2iα

Mβf



(r)|n

, (16.25)

−E

)n

|f(r)|n



=n

ξ +ξH



=n

|−f



(r) −2f



(r)

−2iα



(r)V (r)|n

. (16.26)

Likewise, Eqs. (16.20) and (16.21) become

)n

|−iα

f(r)|n



=n

|−f



(r) −2f(r)

−2iα

V(r)f(r)|n

, (16.27)

−E

)n

|−iα

f(r)|n



=n

|−f



(r) +

βf (r ) +2iα

βM(r)f (r)|n

, (16.28)

)n

|βf (r )|n



=n

|−iα

βf



(r) +2(M +βV (r))f (r)|n

. (16.29)

Finally, let us consider a very special case f(r)=1 and V(r)=−Zα/r.Ifso,

we have

)n

|β|n

=



M −β

Zα



. (16.30)

Before ending this section, we want to give a few useful remarks on the radial

Hamiltonian H given in (16.4). First, it is found that this Hamiltonian is hermitian

if we consider (d/dr)

†

=−d/dr, i.e., H

†

= H . Therefore, the eigenvalues of the

Hamiltonian are reals. Second, it might be noted that there are other two forms of

the radial Hamiltonian. The ﬁrst one is given by

H =−iα





+Mβ +V(r), α

=σ



0 −i

i 0



, (16.31)

whose matrix form becomes

H =



M +V(r) −(

−

)

(

)V(r)−M



(16.32)