Professional Documents
Culture Documents
Department of Physics
Indian Institute of Technology Gandhinagar
Gujarat, 382355, India
November 2023
1 Introduction
The study of two level systems is trivial but not that interesting. As we know, the limi-
tations of two level system is that, one cannot achieve population inversion and therefore
the lasing activity is missing. To obtain an efficient lasing activity, one has to turn to
three level systems, in which obtaining a population inversion is relatively easy.
In three level systems, we introduce a third level, which is a meta-stable level and is al-
ways in quasi-equilibrium with the system. One can then write Schrodinger equation for
the evolution of all the three levels. Since the third level is always in quasi-equilibrium
with the system(in other words, the time scale of evolution of the third level is much
smaller as compared to other two), we can always average out the dynamics of this level
and set it to zero. This particular procedure is known as the ′ Adiabatic Elimination′ ,
which reduces a three level problem to effective two level problem with which we are
quite comfortable. One can think that a three dimensional Hamiltonian being reduced
to effective two level Hamiltonian.
In addition to this, there are many more unitary operation, which one can define on
the Hamiltonian, which reduces the complexity of the problem but does not change the
effective dynamics in any sense. We define two such unitary operations as follows -
1) T urning to rotating f rame : This operation carries us to the rotating frame in which
the coefficients of the states loose their explicit time dependence and thus the problem
becomes easy to solve.
2) Driving state vector to shif ted picture : This unitary operation can be defined as
to drive the state vector to some shifted picture, but we already know that any shifted
picture where we are carried by some unitary transformation is as ’good’ as the original
picture and therefore the dynamics of the system should not change under such unitary
transformations.
In addition, the operations which carry us to rotating frame and to the shifted picture
do commute as we will demonstrate in the coming sections, and therefore the order in
which they are performed should not change the effective dynamics of the system. By
this argument, we claim that the procedure discussed above (Adiabatic elimination) can
be carried in any shifted picture without affecting the effective dynamics of the system.
1
But as we will see in the later sections, if one does not carefully follow the procedure, we
end up with some ambiguous results which shows that every shifted picture defined by
unitary transformations is different and working in each shifted picture results in different
dynamics, which does not make any sense at least physically !!
We then try to find the exact way to perform the adiabatic elimination by rigorously
solving the Schrodinger equation.
H̃ = U HU † + ih̄(∂tU )U † (2)
Now we introduce a unitary operation which drives the state vector to a shifted picture
as: Ũ = exp(−iηt) where η is a diagonal matrix. Since η is a diagonal matrix it is easy
to verify that the operators U and Ũ commute !
E
Under the rotation defined by Ũ , the state vectors are defined as: |α̃⟩ , β̃ ,meta stable
state |γ̃⟩. Again, we can write the Schrodinger equations and adiabatically eliminate the
excited state and obtain an expression as:
Ωa Ωb
γ̃ = − α̃ − − β̃ (7)
2∆(1 + η) 2∆(1 + η)
Here is the crucial step, we have performed Adiabatic elimination in the shifted picture,
but now we will again revert back to the original picture by applying the inverse unitary
operation and hope that the effective two level Hamiltonian is same as calculated in the
previous section. We can hope this because as stated earlier any shifted picture obtained
by unitary operations is as good as the original picture, and therefore calculations in these
shifted frames should not in principle affect the original dynamics. But to our surprise
we find that the two level effective Hamiltonian obtained by this procedure is:
δ Ω2a Ω∗R,η
+ Ωa Ω∗b
Hef f,η = 2 Ω4∆(1+η) 2
Ω2b
, ΩR,η = (8)
R,η
− 2δ + 4∆(1+η) 2∆(1 + η)
2
which is clearly different from eq(6). Thus we are forced to believe that the effective
dynamics of the system depends on which picture we are using to perform the adiabatic
elimination. But this conclusion does not make any sense.
To solve this problem we will look closely at the Schrodinger equation and carefully
perform the Adiabatic elimination.
3
3 Rigorous Adiabatic Elimination using Schrodinger
equation
The approach in this section is to calculate the eigenenergies and use them to Fourier
decompose the coefficients of each levels. To ease this calculations, we first introduce a
set of variables. Since the condition, |∆| >> |δ|, |Ωa |, |Ωb | is followed, we introduce λ, λk
and ϵ such that:
δ Ωk
λϵ = , λk ϵ = where 0 < ϵ << 1 (9)
2∆ 2∆
3
X
α(t) = Ak e−i∆xk t (10)
k=1
3
X
β(t) = Bk e−i∆xk t (11)
k=1
3
X
γ(t) = Ck e−i∆xk t (12)
k=1
Ωa Ωb
γrel = − α− β + O(ϵ2 ) (13)
2∆ 2∆
δ |Ωa |2 Ω∗R
2
+ 4∆ 2 Ωa Ω∗b
Hef f = , ΩR = (14)
ΩR |Ωb |2 2∆
2
− 2δ + 4∆
4
3.1 Rigorous Adiabatic Elimination in shifted picture
In this section, we drive the state vector to shifted picture using Ũ = exp(−iηt) and
keep the parameters defined in eq.(9) same. Again using the Fourier decomposition and
isolating the slow varying terms we obtain:
Ωa Ωb
γ˜rel = − α̃ − β̃ = e−iη∆t γrel (15)
2∆ 2∆
δ |Ωa |2 Ω∗R
2
+ 4∆ 2 Ωa Ω∗b
Hef f,η = , ΩR = (16)
ΩR |Ωb |2 2∆
2
− 2δ + 4∆
comparing this Hamiltonian and the one obtained in eq.(8), we see that both differ.
But the interesting thing to note is that by using this rigorous procedure we have obtained
same two dimensional effective Hamiltonian in the non- shifted picture as well as in the
shifted picture (eq.(14), eq.(16)).
Thus the ambiguity which we faced in Section(2) about the different dynamics in different
shifted picture has been cleared. This was possible when we isolated the relevant subsets
of the coefficients of all the three level. Therefore we see that all the frames are equivalent
to work in as long as we ignore the fast varying components of the states.
But here is thought to ponder upon, when we naively applied adiabatic elimination
in the non-shifted picture we obtained eq.(6). Then we applied same procedure in the
shifted picture and obtained Hamiltonian different than eq.(6). Coming to Section(3), by
following the rigorous scheme in both shifted as well as in non shifted picture we again
obtained eq.(6). Then, how did we get the right result in the non shifted picture? What
was special about the non-shifted picture that we obtained the right result, because it
could as well have been obtained in the shifted picture, was it a lucky coincidence?
Thus, elimination scheme employed in Section(2) appears as a trick, which works only
when the origin of energies lies along, the midpoint of two lower states. Therefore we
see that as soon as we change the origin of energies by moving to the shifted picture in
Section(2.1), this trick fails, and we have to employ more sophisticated techniques. Hence
one should be careful while applying these methods naively to any system.