Non-Degenerate Perturbation Theory#
Given a easily solvable solution to the Schrodinger equation denoted as,
\(H^0\) : simple solvable Hamiltonian
\(\psi_n^0\) : general solution to the simple Hamiltonian assuming there are \(n\) solutions.
\(E_n^0\) : general energy eigenvalue to the simple Hamiltonian assuming there are \(n\) energy eigenvalues.
We would like to perturb the Hamiltonian very slightly transforming the Hamiltonian into,
\(H'\) : perturbation that contributes to the Hamiltonian
\(\lambda\) : strength of the perturbation, it’s arbitrary and assume to be very small.
Perturbation theory states that the solution to the perturbed hamiltonian is in the form of a geometric series,
\(\psi_n^j\) : the \(j\)-th order correction to the eigenfunction
\(E_n^j\) : the \(j\)-th order correction to the eigenvalue
\(\lambda_n^k\) : The \(k\)-th power
While it is a confusing notation, \(j\)-th index does not mean power only \(k\)-th does.
We may now rewrite the TISE for the perturbed hamiltonian. This simplifies to,
First Order Perturbation#
For very small perturbations where \(\lambda \ll 1\) we may approximate first order by considering terms in the Equation \eqref{eq: perturbed-tise} with \(\lambda^1\),
The first order correction to the eigenvalue and eigenfunction are the following,
Equation \eqref{eq: first-order-eigenfunction} is valid only if the system is non-degenerate hence the name of the theory.
Correction to Eigenvalue#
We wish to find \(E_n^1\) by applying \(\langle \psi_n^0 |\),
We find that the first order correction of the eigenvalue \(E_n^1\) is the expectation value of \(H'\) for the unperturbed eigenfunction \(\psi_n^0\),
Correction to Eigenfunction#
We wish to find the first order correction to the wavefunction \(\psi_n^1\). To do so let’s rewrite Equation \eqref{eq: first-order} as an inhomogeneous differential equation,
Recall that the \(\psi_n^1\) is a wavefunction that can definitely be built by a linear combination of the complete set of the unperturbed wavefunctions \(\psi_n^0\). We will use the subscript \(m\) to preventing you from assuming \(n=m\),
The solution to finding \(\psi_n^1\) is to find what each weight \(c_m^{(n)}\) corresponds to,
Let’s plug this back into Equation \eqref{eq: first-order-2},
Notice that the term in the sum where \(n=m\) goes to zero since \((H^0 - E_n^0)\psi_n^0 = 0\) so we can ignore the term \(m=n\). Also notice that we do know the eigenvalue for every application of the hamiltonian on \(\psi_m^0\) to be \(E_m^0\).,
You’re able to isolate \(c_m^{(n)}\) if we were to apply \(\langle \psi_l^0 |\) where \(l\neq n\), and notice that \(\langle \psi_l^0 | \psi_m^0\rangle\) only exist if \(m=l\)
Let’s switch the index \(l\) back to \(m\) since we won’t be needing anymore confusion,
Thus the first order correction to the eigenfunction \(\psi_n^1\) is ,
Second Order Perturbation#
Similar to first order, we will only consider terms with \(\lambda^2\) in the Equation \eqref{eq: perturbed-tise},