A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
For degenerate unperturbed levels, solve the matrix eigenvalue problem of H' restricted to the degenerate subspace to find correct zeroth-order states.
In regular (non-degenerate) perturbation theory, you found the first-order energy correction E¹ₙ = ⟨n|H'|n⟩ and the first-order state correction by mixing in other unperturbed states. The mixing formula contains terms like ⟨m|H'|n⟩/(E⁰ₙ − E⁰ₘ). This works beautifully when all unperturbed energies are distinct. But what happens when two or more states share the same unperturbed energy? The denominator E⁰ₙ − E⁰ₘ goes to zero, and the formula blows up. Degenerate perturbation theory is the resolution to this breakdown.
The fundamental issue is that when a subspace is degenerate, any linear combination of the degenerate states is an equally valid zero-order eigenstate. The perturbation H' will in general prefer certain combinations — it lifts the degeneracy by having different matrix elements for different basis choices. The correct strategy is to find the good states: those linear combinations of the degenerate subspace that diagonalize H' within that subspace. These good states have well-defined first-order energies and do not suffer from the zero-denominator problem when mixed with states outside the subspace.
Concretely, if you have an n-fold degenerate level with unperturbed states |ψ₁⟩, ..., |ψₙ⟩, you form the n×n degenerate subspace matrix W with elements Wᵢⱼ = ⟨ψᵢ|H'|ψⱼ⟩. Diagonalizing W gives you n eigenvalues — these are the first-order energy corrections — and n eigenvectors — these are the good zeroth-order states. Each eigenvalue E¹ describes how much H' shifts the energy of the corresponding good state. When the n eigenvalues are all different, H' has completely lifted the degeneracy to first order. If some remain equal, you have a residual degeneracy and must look to higher order.
The classic example is the hydrogen atom in an external electric field (the Stark effect). The n = 2 level is four-fold degenerate: the 2s and three 2p states all share the same unperturbed energy. The electric field perturbation H' = eEz mixes these states. Forming the 4×4 matrix and diagonalizing it reveals which combinations are shifted (by ±3eEa₀) and which are unshifted. Crucially, the good states are specific linear combinations of 2s and 2p — not the original spherical harmonics — chosen precisely so that H' is diagonal. Degenerate perturbation theory tells you both how much the levels shift and what the physically relevant quantum states become under the perturbation.