# Perturbation Theory (stationary

)

Lecture 2: Non-degenerate perturbation theory and sample applications

**Lecture 3: Degenerate perturbation theory and sample applications
**

Lecture 4: Fine structure of Hydrogen atom Lecture 5: Weak-field Zeeman effect and hyperfine splitting

**Lecture 2: Stationary Perturbation Theory
**

(non-degenerate cases)

Given an exactly solvable Hamiltonian with eigenkets and eigen-energies

Can we say something about the perturbed system

Standard trick here: Power series expansion of the unknown solution in terms of the perturbation strength

Unknown solution:

λ

Known solution:

0

Assume

is a smooth function of

λ

Assume

is a smooth function of

λ:

Some remarks on the series expansion:

When λ 0,

the solution should smoothly approach the known solution at λ = 0

One can assume that the correction terms are orthogonal to the zeroth-order term. If not, we can add those components “parallel” to the zeroth-order term and then get the same expression after “requesting” the coefficient of the zeroth-order term to be unity (eigenfunction multiplied by any factor remains eigenfunction).

Normalization is okay up to the second order, so re-nomarlization might be needed depending what order of accuracy is desired.

Some plug-in’s and term-by-term comparisons:

Zeroth order term of λ: 1st order term of λ:

2nd order term of λ:

looks messy, but still okay…

**One more step in working out the approximate solution
**

Project the obtained equations onto the basis set

(you have to adopt a representation to do something specific)

1st order:

2nd order:

(these two expressions come from the last slide)

**1st order correction to the eigenket: Nondegenerate cases
**

1st order:

“Multiply” both sides by

with m = n :

Continued on the next slide

Continuing the last slide:

If the unperturbed system is nondegenerate:

Inserting completeness condition:

1st order correction to the eigenvalue: Nondegenerate cases

“Multiply” both sides by

:

Remark:

1st order correction to eigenvalue is given by the expectation value of the perturbation on zeroth-order eigenstates

**Simple derivation of Feynman-Hellmann Theorem
**

perturbed Hamiltonian: polynomial expansion: 1st order eigenvalue :

shifting this result to arbitrary λ

Feynman-Hellmann Theorem

**Example A: Perturbation to a harmonic potential
**

harmonic oscillator perturbation

Recall “Virial theorem” … Or “equal partition” of a harmonic oscilator system:

The λ parameter in our perturbative treatment is now represented by epsilon here…

**Example B: two identical bosons in an infinite square well
**

Perturbation:

Zeroth order ground state:

=

2nd order correction to the eigenvalue: Nondegenerate cases

2nd order:

“Multiply” both sides by

:

Comments on 2nd order correction to the eigenvalue

The 2nd order correction is given by the overlap between the 1st order eigenket, the perturbation, and the zeroth-order eigenket.

The 2nd order correction reflects the curvature of the eigenvalue (i.e., 2nd order derivative) as a function of the perturbation strength. Small energy level spacings lead to large curvatures of the energy curves.

**Example C: 2nd order corrections in a perturbed harmonic oscillator system
**

harmonic oscillator perturbation

=0

=

**Continuing the previous slide:
**

summation index

mass

=

Because of the special form the perturbation, all higher-order corrections are zero !

Summary

Power series expansion is used to obtain perturbative solutions of the stationary Schrodinger equation. If the zeroth order spectrum is nondegenerate, up to 2nd order results are explicitly given.

If the zeroth order spectrum is degenerate, then our 1st order eignestates are in trouble (why?) and hence we have yet to find a special treatment for degenerate cases (topic of next lecture).