🌱 Seedling — sparse and possibly wrong.
Lattice models with a macroscopically degenerate flat band — the α–T₃ / dice family, and Kekulé-modulated honeycombs that gap or flatten the Dirac spectrum. The frontier question I care about is what happens when you try to realize these in a real material and treat them with DFT: the flat band is the worst case for self-interaction error, because vanishing kinetic energy lets SIE-driven (de)localization dominate the physics.
Concretely: CRYSTAL23 two-component (2c) DFT and a QE + Wannier90 route to extract the effective model, with flat-band SIE flagged as the systematic to control. Open and speculative — hence L3.
Prerequisites
Tight-binding band theory · hybrids and self-interaction error
Builds toward: a documented flat-band SIE benchmark