This paper shows that when physicists adjust the positions of orbitals in a lattice model to make certain measures of “quantum geometry” as small as possible, the best placement must respect the same spatial symmetries the model itself allows. The result applies to two commonly used measures: the variance (how much it wobbles) of the Berry curvature and the integral of the trace of the quantum metric tensor (a single number summarizing a band’s geometric spread).

Why this matters: those quantum‑geometric measures are used to predict and compare physical behavior of electron bands, for example in efforts to create exotic states like fractional Chern insulators. But for a common type of model called a tight‑binding model, those measures can change if you move the orbitals around in real space. That ambiguity is called the orbital embedding problem. The authors focus on the minimal possible value of a geometric measure across all embeddings, which is a geometry‑independent quantity, and ask which embeddings achieve that minimum.

What the researchers did: working from a precise definition of tight‑binding models and the quantum geometric quantities, the authors prove that any embedding that minimizes either the Berry‑curvature variance or the integrated quantum metric must itself be symmetric under every spatial symmetry compatible with the model. They make this claim for ordinary spatial symmetries and then extend it to more complicated cases where a spatial operation must be combined with a change of phase (a gauge transformation) or with time reversal. They also point out concrete consequences: for the Hofstadter model the usual choice of placing orbitals on a regular grid inside the magnetic unit cell indeed minimizes these measures, and for “ideal bands” (bands that meet a geometric lower bound) the orbital positions must obey the model’s symmetries.