Feb 2, 2002
Summary of perturbation theory equations
Summary of perturbation theory equations
[0][0]
Time independent perturbation expansion
Suppose we have a reference Hamiltonian H0 for which we know
all of the eigenvalues and eigenfunctions:
Now we want to
approximate the eigenvalues En and eigenfunctions Fn of total
Hamiltonian H º H0 + H1, where the second
term is small compared to the reference Hamiltonian term. If the nth
zero-order eigenstate (E0n) is not degenerate, then we can make the
following expansion. We will use the
shorthand notation áF0k|H1|F0mñ º Vkm.
En » E0n + Vnn + |
å
m ¹ n
|
|
|Vmn|2 E0n-E0m
|
+ O(V3). |
| (2) |
Fn » F0n + |
å
m ¹ n
|
F0m |
Vmn E0n-E0m
|
+ O(V2). |
| (3) |
If, on the other hand, the zero-order eigenstate (E0n) is degenerate
with one or more other eigenstates, another method must be used. Suppose
there are N such degenerate states which we will label {F0ni},
where i = 1,2,¼N. We suppose that we can find N new zero-order states
{F0a} from linear combinations of the original states, by diagonalizing
the following N ×N matrix:
|
æ ç ç
ç ç è
|
|
|
|
ö ÷ ÷
÷ ÷ ø
|
|
æ ç ç
ç ç è
|
|
|
ö ÷ ÷
÷ ÷ ø
|
= Ea |
æ ç ç
ç ç è
|
|
|
ö ÷ ÷
÷ ÷ ø
|
|
| (4) |
The energy eigenvalues {Ea} correspond to corrections up to
first order in the perturbation for this system. Each eigenvalue Ea
corresponds to a linear combination of the zero order eigenfunctions in terms
of the coefficients {Cani}:
F0a = |
N å
i = 1
|
Cani F0ni. |
| (5) |
Time dependent perturbation expansion
Now suppose that the perturbation depends on time. We will focus on the
case in which there is a harmonic time dependence which is ``turned on''
at time t = 0:
H1(t) = V(r) ( ei wt + e-i wt ) Q(t), |
| (6) |
where Q(t) denotes the Heaviside step function.
If the system is initially (t < 0) in the zero order state F0n, the
effects of the perturbation to first order in V is given by
Fn(r,t) » F0n(r)e-i E0n t/(h/2p) + |
å
m
|
c(1)m(t)F0m(r)e-i E0m t/(h/2p) , |
| (7) |
where
c(1)m(t) = - |
Vmn (h/2p)
|
|
é ê
ë
|
ei(wmn+ w) t - 1 wmn+ w
|
+ |
ei(wmn- w) t - 1 wmn- w
|
ù ú
û
|
. |
| (8) |
In this expression, wmn º [(E0m - E0n)/( (h/2p) )].
For large times t, it can be shown that the squared modulus of the
exitation coefficient c(1)m(t) determines the transition rate:
Rn® m = |
|c(1)m(t)|2 t
|
» |
2 p (h/2p)2
|
|Vmn|2 ( d(wmn+ w) + d(wmn- w)), |
| (9) |
or
Rn® m » |
2 p (h/2p)
|
|Vmn|2 ( d(E0m - E0n + (h/2p)w) + d(E0m-E0n- (h/2p)w)), |
| (10) |
File translated from TEX by TTH, version 2.20.
On 2 Feb 2002, 16:22.