 Schrieffer–Wolff transformation

The Schrieffer-Wolff transformation is a unitary conjugation that decouples selected low- and high-energy sectors to construct an effective Hamiltonian. It is a form of degenerate perturbation theory: states within the retained sector are treated together.

Let an orthogonal projection $P$ commute with $H_0$, put $Q=I-P$, and decompose the perturbation in $H=H_0+ϵV$ asThe first part acts within sectors; the second couples them. For an anti-Hermitian generator $S=ϵS_1+O({ϵ}^2)$, choose $[S_1,H_0]=-V_{od}$. It is the inter-sector coupling, rather than the whole perturbation, that this commutator removes. Projecting the consistently transformed Hamiltonian givesIn an eigenbasis of $H_0$, cross-sector entries obey $(S_1{)}_{mn}=V_{mn}/(E_m-E_n)$. A nonzero separating spectral gap controls these denominators; resonant coupled states belong in a common retained block.^[ref]

For a direct two-level check, take $H_0=\mathrm{diag}(0,\Delta )$, $\Delta >0$, and $V=t|0⟩⟨1|+t^{\ast }|1⟩⟨0|$. The formula predicts the lower energy shift $-{ϵ}^2|t{|}^2/\Delta$. Expanding the exact lower eigenvalue, $(\Delta -\sqrt{{\Delta }^2+4{ϵ}^2|t{|}^2})/2$, gives the same result. A diagonal perturbation would also contribute at first order through $PVP$.

John Robert Schrieffer and P. A. Wolff used the method to relate impurity models underlying the Kondo effect.^[ref]