# Stationary Perturbation Theory

1. Sep 24, 2009

### tommy01

Hi together...

When reading Sakurai's Modern Quantum Mechanics i found two problems in the chapter "Approximation Methods" in section "Time-Independent Perturbation Theory: Nondegenerate Case"

First:
$$H_0 | n^{(0)}\rangle=E_n^{(0)} |n^{(0)}\rangle ~~~~~~ (1)$$
whereas the perturbed looks like
$$(E_n^{(0)} - H_0)|n \rangle = ( \lambda V - \Delta_n)|n\rangle ~~~~~~ (2)$$
with $$\Delta_n = E_n - E_n^{(0)}$$

The Equation is inverted and we arrive at
$$|n\rangle = \frac{1}{E_n^{(0)} - H_0} \Phi_n ( \lambda V - \Delta_n)|n\rangle$$ with the projection operator $$\Phi_n$$ to ensure that the inverse is well defined.

Then Sakurai says this can't be the correct form because as lambda approaches zero the perturbed state ket has to approach the unperturbed ket. Then he says even for lambda not equal to zero we can add a solution to the homogeneous equation (1) which is $$|n^{(0)}\rangle$$ and so the result is

$$|n\rangle = |n^{(0)}\rangle + \frac{1}{E_n^{(0)} - H_0} \Phi_n ( \lambda V - \Delta_n)|n\rangle$$

And there lies the problem. In my opinion (2) is also a homogeneous equation (but not the same as (1)) and you can't add a solution to a homogeneous equation to a different homogeneous equation. Or am I wrong?

Second:
The formalism then leads to an expresion for the energy shift
$$\Delta_n=\lambda V_{nn} + \lambda^2 \sum_{k\neq n} \frac{|V_{nk}|^2}{E_n^{(0)}-E_k^{(0)}} + ...$$

Then a special case of the no-level crossing theorem is stated.
Say we have 4 energy levels i, j, k and l in increasing order of magnitude.
Then Sakurai states that two levels connected by perturbation tend to repel each other. Inserting i and j in the equation above leads to a negative energy shift of i and a positive shift of j analogous for k and l.

But when we compare j and k they also repel each other which means j gets lower and k gets higher. In contradiction to the observations above.
Now i think there isn't really a contradiction because Sakurai only considered terms in second order of lambda and neglected other parts of the sum.

But then how he can deduce the statement that no two levels cross when connected by perturbation form the argumentation above?

I hope i could explain my two problems satisfying.

greetings.