Imagine you're trying to tune a piano, but you can only reach the keys nearest to you. If the piano is well-behaved — each key influences mostly its neighbors — you can tune the whole instrument by working section by section. But some pianos have a nasty property: a key in the middle changes its behavior depending on keys at the far ends. For those pianos, local tuning is provably impossible. This paper is about which quantum pianos admit local tuning and which don't. The actual claim: for free-fermion Hamiltonians, the authors derive a complete criterion — necessary and sufficient in 1D, sufficient in all dimensions — for when the ground state can be prepared using exponentially local linear jump operators in a quadratic Lindbladian. The core technical move is a Fourier correspondence: spatial locality of jump operators maps to analyticity in momentum space, converting a physical question into a clean complex-analysis problem. This lets them read the answer off the Hamiltonian's band structure. The no-go theorem is the sharpest result. If the Fermi surface has sign-changing crossings — places where bands cross the Fermi level and the sign of the gap flips — then no exponentially local linear jump operators exist that prepare the ground state. In 1D, this is the only obstruction. In higher dimensions, additional obstructions appear, and the authors illustrate these with two quadratic-band-touching models whose ground states have algebraically (not exponentially) decaying correlations. On the positive side, short-range ground-state correlations are shown to be sufficient for local dissipative preparation, and this result holds without requiring translation invariance. This is a useful practical guarantee: if your target state has exponentially decaying correlations, you're in the clear regardless of the system's symmetry structure. The ladder here is interesting. Prior work on dissipative state preparation (Verstraete, Wolf, Cirac 2009; Kraus et al. 2008) established the general framework of Lindbladian engineering, but locality constraints were treated case-by-case. More recent work by Chen, Kastoryano, and Gilyén (2023) on rapid mixing gives runtime bounds but doesn't address the existence question for local generators. This paper fills that specific gap for free fermions with a criterion that's checkable from the Hamiltonian alone. The architecture is elegant in its narrowness. Free-fermion Hamiltonians are quadratic in creation/annihilation operators; the Lindbladian is also quadratic with linear jump operators. This is not a variational method or a circuit — it's an analytical framework. The compute cost is essentially zero; the result is a mathematical criterion, not a numerical one. The tradeoff is that the free-fermion restriction is severe. Interacting systems — the ones you actually want to simulate — remain out of reach. The integrity is strong for a theory paper: the claims are mathematical theorems with proofs, not numerical experiments. The no-go result is a theorem, not a conjecture. The illustrative examples (quadratic-band-touching models in 2D) are concrete and falsifiable in principle. No benchmarks to cherry-pick, no hyperparameters to tune. The obvious question is whether the criteria extend to interacting systems, and the authors are honest that this remains open.
