Anderson delocalization in dimension d3d\ge3

OPENLandmarkConjectureProposed 1958 · Canonical special case

Canonical statement

Let d3d\ge3 and
(Hλ,ωψ)(x)=yx=1ψ(y)+λωxψ(x)on 2(Zd), (H_{\lambda,\omega}\psi)(x)= -\sum_{|y-x|=1}\psi(y)+\lambda\omega_x\psi(x) \quad\text{on }\ell^2(\mathbb Z^d),
where the ωx\omega_x are iid with a bounded compactly supported density that is positive near 00. For all sufficiently small λ>0\lambda>0, there is a nonempty open interval II in the interior of [2d,2d][-2d,2d] on which Hλ,ωH_{\lambda,\omega} has almost surely nonempty purely absolutely continuous spectrum.
View source LaTeX
Let \(d\ge3\) and \[ (H_{\lambda,\omega}\psi)(x)= -\sum_{|y-x|=1}\psi(y)+\lambda\omega_x\psi(x) \quad\text{on }\ell^2(\mathbb Z^d), \] where the \(\omega_x\) are iid with a bounded compactly supported density that is positive near \(0\). For all sufficiently small \(\lambda>0\), there is a nonempty open interval \(I\) in the interior of \([-2d,2d]\) on which \(H_{\lambda,\omega}\) has almost surely nonempty purely absolutely continuous spectrum.

Anderson's 1958 analysis of diffusion in random lattices predicted that disorder can localize quantum particles [Anderson1958AbsenceDiffusion]; in three or more dimensions, a localization–delocalization transition is expected, with extended states surviving at weak disorder. In mathematical terms, for the discrete Schrödinger operator on 2(Zd)\ell^2(\mathbb Z^d) with iid random potential of strength λ\lambda, the conjecture asserts that for d3d\ge3 and small λ\lambda there is an energy interval in the interior of the spectrum on which the spectrum is almost surely purely absolutely continuous — the spectral signature of extended states.

The localized side of the picture is on firm ground: localization is proved at large disorder and near spectral edges, notably by the fractional-moment method [AizenmanMolchanov1993Localization], and the theory is laid out in surveys [Kirsch2008RandomSchrodinger] [Stolz2011AndersonLocalization]. Delocalization, by contrast, is established only on trees and in mean-field or random-matrix analogues of the model.

For the standard iid lattice model on Zd\mathbb Z^d itself, no interval of absolutely continuous spectrum has been exhibited. The conjecture is open, and a resolution appears to require genuinely new tools for proving the existence of extended states in finite dimensions.

The boxed statement is the canonical open formulation — not a stronger variant or a related research program. The status reflects the catalog's last review; do your own literature search before investing serious effort.