Advanced Quantum Mechanics with Applications

Prof. Saurabh Basu

Department of Physics
Indian Institute of Technology, Guwahati

Lecture – 21
Perturbation Theory

We are going to start a new chapter in the study of Advanced Quantum Mechanics and
its Applications. This is called as the approximate methods in quantum mechanics.

(Refer Slide Time: 00:39)

So, most of these problems in the single particle quantum mechanics, that is obtained by
the solution of Schrodinger equation. Most of them are actually unsolvable because of
the complexity of this a mathematics involved, because the second order differential
equation that we are trying to solve gets so complicated, that it is more often than not the
closed solutions are not available. .

In which case the approximate methods are used in order to arrive at a solution and most
of the time these approximation methods or the approximate methods play a very
important role in the study of different quantum systems. So, we are going to look at a
mainly 3 approximate methods. They are Perturbation theory, number 2 is Variation
theory or a Variational theory and a number 3 is a WKB approximation.
Now, the first focus will be on the perturbation theory and let us see under what
conditions perturbation theory works and it is applicable.

(Refer Slide Time: 02:52)

So, suppose there is a problem the Hamiltonian of which is represented by

H | ψn>= E n | ψn > : Not Known

So, we really do not know the solution of this problem. So, this is even I am writing it
formally the solutions are not known which means that both | ψn > and En are a
unknown.

However, the H that we just saw, the Hamiltonian of the system can be represented in
terms of a two terms the Hamiltonian is written in terms of these two terms which are

H =H 0 + λ H '

H 0 , which is called as the unperturbed Hamiltonian. And, H ' is called as the

Perturbation Hamiltonian, it is important to realize that the system actually can be split in
terms of these two Hamiltonians and that should be known. And, in addition to that there
are two more conditions, one is that that the solution to this, the unperturbed problem is
completely known which means that. So, by 0 we mean the unperturbed problem.

So,
 0 0 0
H 0 | ψn >= En | ψn >

where n is some quantum number that is relevant to the problem could be momentum,
could be spin, could be you know position etcetera; however, this problem is completely
known to us. And the next condition, which is a very strict condition that this lambda has
to be much smaller than 1, which means that the perturbation term is much smaller
compared to the original Hamiltonian H0 .

⟨H'⟩
≪1
⟨ H0⟩

And if you want a order of magnitude then λ ≃0.01 at least or even lower than that. In
fact, there is a the scale energy scale that is relevant to this discussion is that the fine
structure constant, we will come back to this later which is called, which is written with
alphas rather,

1
α 2=
137

So, a lambda has to be comparable to this the fine structure constant and in order to have
a and for the Perturbation theory to work this lambda has to be small.

So, the you know will it say that the expectation value of the Perturbation term divided
by the expectation value of the unperturbed Hamiltonian is much smaller than 1. This is
of course, assuming that we know how to calculate the expectation values and this is
what we are going to learn. Now expectation values are taken with regard to some
known states. In fact, we will see that these known states are nothing, but the
unperturbed states of the system. In fact, we may not require the corrected wave function
for most of the cases in order to have meaningful results, we could actually calculate the
expectation values with the unperturbed wave functions themselves.
(Refer Slide Time: 07:41)

So, these are the requirements and we have the two a sort of versions of a perturbation
theory which are important and one of them is called the time independent perturbation
theory. And we will just call it or rather abbreviate it as a TIPT and the second one we
will call it as, a time dependent perturbation theory and we will call it as TDPT okay.
Now even in the time so if we begin with time independent perturbation theory or the
TIPT as we said that will encounter there are again, you know degenerate and non
degenerate perturbation theory.

But before we come to that let us tell you the features of this time independent
perturbation theory. So, time independent perturbation theory, done on a system will a
cause shift in the energy level or split in the energy level. And, time dependent
perturbation theory will actually a cause transitions, among the energy levels and this the
next, the second one that is a time dependent perturbation theory will be shown to be
relevant for the discussion of a lasers or how lasing action takes place we will discuss
that. So, we will first discuss the TIPT first, but then it has 2 variant as I said this is
called as the non degenerate perturbation theory and b is a degenerate perturbation
theory.
(Refer Slide Time: 10:57)

So, these are going to be discussed and we will start with a so, a non degenerate TIPT,
we will start with that and so then, go over to the degenerate very quickly. Alright so, as
we told that the total Hamiltonian is written as

H =H 0 + λ H '

remember that this separability of the Hamiltonian into two terms is very important for
the perturbation theory to be applicable.

So, we just apply it on the wave function

H | ψn>= E n | ψn> (1)

and we know that if a lambda is small and can be neglected then

0
E n → En ψn → ψ0n

that is the perturbation effects can be neglected. Alright so, this is the zeroth order
problem. Let us look at the first order or rather take into account the perturbation and see
that how things can be calculated or how with the energy can be calculated, because of
the inclusion of this H ' term.
 0
And remember that your | ψn > , let us remove it | ψn > form a complete set and we
0
may expand, | ψn > in terms of | ψn > , what it means is that the full wave function
including the perturbation term can be expanded in terms of the unperturbed wave
function. So, this

0 0
| ψn>=N ( λ )[| ψn > + ∑ C nk ( λ )| ψk > ] Ansatz (2)
k≠n

So, this is the basically the ansatz, that we start with ansatz it is the assumption that okay.

For zeroth order ( λ =0)

0
N (0)=1 C nk (0)=0 | ψn >=| ψn >

this is what we have seen; we have not yet taken into account the perturbation that will
do it in the first order.

(Refer Slide Time: 15:25)

So, in general we have

2 (2)
C nk ( λ )= λ C (1)
nk + λ C nk +..... (3)

and also so this is the wave function the correction in the wave function, which are
written in terms of the amplitude of the wave function and of course, your
 2 (2)
E n= E(0)
n + λ E n + λ E n +....
(1)

So, what we mean to say is that when we take a linear power of lambda, which is trying
to do a first order perturbation theory and then there is a superscript of 1 in the bracket
and then it is a lambda square as a 2 superscript 2 in the bracket and so on.

So, if we substitute these two into this equation, let us call this as the ansatz as equation 1
and let us call these two as equation 2. So putting 2 into 1, we have,

(H 0+ λ H ' )[| ψ0n > + λ ∑ C (1) 0 2 (2) 0

nk | ψk >+ λ ∑ C nk | ψ k >+...]
k ≠n k ≠n

0 (1) 2 (2) (0) (1) (0) 2

=( E n + λ E n + λ E n +....)[ ψn + λ ∑ Cnk | ψk > + λ ∑ C(2)nk | ψ(0)
k > +...]
k≠ n k ≠n

We are writing it without we were writing the wave functions without a bracket. So, let
us just remove the bracket. Alright so, this is and then of course, you have terms now this
is put in the so, this should be actually 1, these were way of putting it. So, maybe this
will write it as 2. So, that is the ansatz and we will call this as 3. So, we will put actually
3 into 1 that is, what we are doing here. So, this is the equation that we get by putting
these two expansions into equation 1, that is the Schrodinger equation and now what we
have to do is that we will have to match terms order by order in both the sides. So, which
means that a same powers of lambda have to be matched. So, we will of course, I should
recognize that a

0 0 0
H 0 | ψn >= En | ψn > (4)

and this is nothing, but the unperturbed equation.

(Refer Slide Time: 21:21)

The first order in lambda we have a

E(1) (0) (0) (0) (1) (0) (0)

n | ψn >= ∑ (E k −E n )C nk | ψ k > +H ' | ψ n >
k ≠n

just do not you know worry about that whether we are putting a bracket on the
superscripts you can or you may not. But, just make sure that they do not mean a square
and they are really talking about the powers of the perturbation theory that we are doing.

So, in order to get the energy correction that is this one, a

(0) 3
E(1)
n =⟨ ψn | H ' | ψ n ⟩=∫ ( ψn ) H ' ψ n d r
(0) (0) (0) ∗

look the first term when you take the overlap of course, this is a number. So, it is just a
psi n 0 psi n 0 which gives us 1, because we said that psi n 0 forms a complete set of
states and when psi n 0 overlap is taken with psi K 0, since K is not equal to n because of
the orthonormality condition this goes to 0.

And the only term that survives is the last the term that is at the end in the RHS the right
hand side and if you are interested in actually calculating this and then this is equal to in
the integral form this is equal to that a psi n a 0, which are which could be functions of
you know r theta phi or it could have additional indices such as, a spin. Or other you
know quantum numbers, which are relevant to the problem it is this, and we are simply
talking about maybe just talking about a integral which is a space integral.

So, this is the first order correction. So, the expectation value of the perturbation term
needs to be taken between the unperturbed ground states and that gives you the first
order correction and. So, that is one part of the story that is we now obtain the first order
correction, what happens to the wave function that is the important thing. And the wave
function can actually be obtained; we take the inner product with a < ψ(0)
m | , where m is

not equal to n. So, with m not equal to n and to calculate a

k | H ' | ψn ⟩
⟨ ψ(0) (0)
(1)
C =
nk
E(0)
n − Ek
(0)

Ok. So, this is the first order correction to the wave function and we have already
obtained the first order correction to energy. So, these are the important things let us just
do a box of these important expressions and. So, one corresponds to the energy
correction the other corresponds to the correction in the wave function.

And if you do it in the second order that is go back to this expression let us call it,
expression or rather equation number 4. If you go to 4 and take the second order
correction or rather second order in lambda and then equate powers in second order in
lambda, remember the second order in lambda will also come from a lambda in the left
hand side. And, the lambda in the right hand side or a lambda square on the right hand
side with no power of lambda, lambda to the power 0 on the right and a vice versa.

(Refer Slide Time: 27:03)

So, that all those combinations need to be taken you can do that and I will simply write
down the expressions that are important. So at the second order, the second order
correction in energy is simply equal to

2
(2) |⟨ ψ(0) (0)
k | H ' | ψ n ⟩|
En =∑
k ≠n E(0) (0)
n − Ek

(2) 1 ⟨ ψ0k | H ' | ψ0n ⟩ ⟨ ψ0n | H ' | ψ0m ⟩ ⟨ ψ0k | H ' | ψ0m ⟩ ⟨ ψ0m | H ' | ψ0n ⟩
C = 0
nk ∑[
E m −E 0n k≠n E 0k −E 0n
−
E 0m− E 0k
]

So, this would renormalize the wave function for the system in the at the second order.
So, remember that it has an energy denominator and. So, is the energy denominator in
0 0
the expression for the second order energy correction, which is En−E k and. So, there
is also an energy denominator in the Cnk term in the first order for the amplitude
correction of the wave function.

Now, what happens if your E0n=E 0k , then all these things will the denominator will
vanish and some denominator vanishing is not a good thing for any physics or rather
anything that is happening because that will lead to divergences. And, these divergences
are unwanted and hence we need to actually reformulate the problem when such a thing
0 0
happens that is your En=E k and needless you say that, we are actually talking about
the degenerate cases.
(Refer Slide Time: 31:07)

So, once again what is Degeneracy? Let me sort of. So, different wave functions having
same energy. So, this is the definition of degeneracy. So, you have an energy which is
say some E and this would correspond to you know say 3 wave functions having,
ψ1 , ψ2 and ψ3 all having the same energy, all having energy E. However, they are all
distinct states and they could be orthogonal to each other and. So, this is what is called as
0 0 0 0
degeneracy in case of degeneracy, one can actually have this Ek = En or Em =En ,
that is the two levels can be actually degenerate and then we have different wave
functions etcetera corresponding to them.

So, what is the way to deal with such situations in which we have degeneracy in the
perturbation theory? So, we have already said that the problem is with the energy
denominator. So, instead of so we will talk about, the degenerate time independent
perturbation theory. So, degenerate TIPT. So, instead of a unique n > , we have a
| ψ(0)
0 (0 )
finite set of | ψn > and all of which have the same energy let us call it as
i
En . So,
not only 1, but there are depending on how big i is that is i could be 2, 3, 4 or in many all
(0 )
of them have the same energy En and this is what we mean by a degeneracy.

However, each of these you know these things are. So,

0 0
⟨ ψn | ψm ⟩= δmn δij
i j
So, basically all this set is orthonormal. So, we now want to expand the wave function

| ψn>=N ( λ )[ ∑ ai | ψ(0) (1)

n > + λ ∑ C nk ∑ b i | ψ k > +...]
i
(0)
i
i k ≠n i

So, this is the ansatz now let us call it.

Okay. So, if we put this ansatz into this equation 1, which is the Schrodinger equation
then, we will get an equation of the form. So, we simply what we have done is that, we
have a replaced

0 0
| ψn> → ai | ψn > i

which is what we have there is a sum and so on okay.

(Refer Slide Time: 35:46)

So, putting this putting ansatz into 1, we have a sum over or. So, let us just write it. So,
this is a

0 0 0 0 0
H 0 ∑ C (1) i i
(1)
i
(1)
nk ∑ b i | ψ k > + H ' ∑ a i | ψn >=E n ∑ a i | ψn > + E n ∑ C nk ∑ b i | ψ k > i
k≠n i i k≠n i

be very careful with the, this indices and this I mean the basically the notation, you have
to write it once in order to you know gain confidence.
 ∑ ai ⟨ ψ0n | H ' | ψ0n ⟩=E (1)n a j
j i
i

So, it is a little more complicated than the non degenerate case that, we have just seen the
only difference being that it is a finite dimensional problem, but still we have to
recognize that it is a finite dimensional problem and should be solvable.

So your, so let you know I mean

0 0
Let, ⟨ ψn | H ' | ψn ⟩=( H ' ) ji
j i

So, that is the matrix element and so on and let us just do it do this problem for a
particular case that is say a twofold degenerate problem.

(Refer Slide Time: 39:29)

So, alright so what is it? So, it is equal to

(H 0 −E 0n )| ψ(1) 0
n >=(E n −H ')| ψn >
(1)

| ψ0n>=a(1) (1)
1 | u 1 > +a2 |u2 >
(0 )
u 1 u 2 are two degenerate states having energy En

(H 0 −E 0n )| ψ(1)
n >=(E n −H ')(a 1 |u 1 >+a2 |u 2 > )
(1) (1) (1)
 ⟨u1 |(H 0 −E 0n)| ψ(1) 1
n ⟩=⟨u1 |(E n −H ' )|(a1 |u1 > +a 2 | u2 >)⟩
(1) (1)

0=E(1) (1) (1) (1)

n a1 −H ' 11 a 1 −H ' 12 a2 ⟨u1 | H ' | u2 ⟩= H ' 12

So, this one will be um.

(Refer Slide Time: 43:41)

So, we will have a two equations I will just write them in the next page. So, we have two
equations which are

a1 (E (1)
n −H ' 11)−a2 H ' 12=0

−a1 H ' 21 +a2 ( E(1)

n −H ' 22 )=0

E(1)
n −H ' 11 −H ' 12
det [ ]=0
−H ' 21 E (1)
n −H ' 22

Now, you see it is like this that we are trying to calculate for this degenerate perturbation
theory the first order energy correction, which involves the matrix elements of these
H ' term that is a perturbation term between the non the unperturbed states. So, these
are u1 and u2 are the degenerate unperturbed states and they are so, this is basically,
(1 ) (11) (21)
so I will get two values of a En ones may be, En and E n that is these are the
two values, which would give that how these two levels that is the energy levels will
actually respond to this perturbation.

So um, in fact, the energy level which was earlier degenerate before the application of
perturbation theory may actually respond differently to this perturbation and this is what
most of the time it happens. So, that would be obtained by solving this secular equation,
which is as I told that it is called let us call them as you would not know
H ' 1 H ' 12 and H ' 11 and H ' 21 and H ' 22 unless, you are given a specific problem,
which is what we are going to see in a particular problem called as the stark effect.

So, let us at least write down the general form for this.

1 2 2
E(1)
n 1
=
2
[(H ' 11 +H ' 22 )+ (H '√22 −H ' 11 ) +4 |H ' |
12 ]

1 2
E(1)
n = [(H ' 11 +H ' 22)− (H ' 22− H ' 11 )2 +4|H ' 21| ]
√
2
2

we just check that it is just um in most of the cases, we are dealing with the Hermitian
matrices. So, it really does not make any difference. So, these are the corrections in the
0
energy En . So, you see that because of this plus and minus sign here and here the two
levels, which were earlier degenerate before the application of perturbation theory has
now reacted differently. So, they will either shift or they will split on either side of the
0
En and we are interested in finding out that and just that the two the first order
correction in the wave function can be obtained by knowing this ratio which is equal to
nothing, but 2 H 21 .

We just all talking about solving a 2 by 2 secular equation; so, this is equal

a(1)
2 2 H ' 21
= 2
a1 (1)
√ 2
(H ' 11 −H ' 22)+ (H ' 22−H ' 11) + 4|H ' 21|

And of course, this gives you the ratio, but there is a because of the normalization there
is another condition that exists, which is the
 a(1)2 (1)2
1 +a2 =1

now this is the full solution to this problem.

So, we have calculated the first order energy correction due to the perturbation term. So,
these are the two of values of the correction. So, the two wave functions or rather which
were degenerate before the perturbation was switched on such as, u1 and u2 they react
or they respond to the perturbation in this particular fashion. And, the coefficients the
first order change in coefficients of the wave function that corresponds to this kind of a
correction which are subject to this normalization condition.

Now, the situation is that that we have done a perturbation theory, degenerate
perturbation theory. Now, suppose this is not a twofold degenerate perturbation theory it
is a fourfold degenerate, which is what we are going to see in stark effect then we have to
solve a 4 by 4 secular equation. Just the complexity of the problem goes up a little, but
you will see that because of the symmetry of the problem many of the terms, which are
cancelled because of you know, because of the symmetry.

And the integrals if you before you actually calculate all the matrix elements such as
H 12 H 21 etcetera; you can look at the symmetry of the problem. And, this quantum
mechanics actually very importantly teaches us how to exploit the symmetry of the
problem and reduce the complexity and you know in a way that is simplify a given
problem. So, we will apply this degenerate perturbation theory to Stark effect, which is
as I told you it is a fourth it is a you know degeneracy the level of degeneracy or the
degree of degeneracy is 4 instead of 2, that we have just shown here.