THE QUANTUM INTELLIGENCER

A paper on the arXiv reports an advance in Hamiltonian learning under the name eigenphase engineering. The method operates from forward time evolution at integer multiples of a fixed interval, and attains Heisenberg-limited learning with a step size that is optimal up to logarithmic factors. For an unknown s-sparse Hamiltonian with fixed locality k and operator norm bounded by Λ, it recovers the Pauli coefficients to ℓ₂ error ε in total evolution time Õ(s/ε), matching the best known scaling while allowing a near-maximal step size of Θ̃(1/Λ). The authors likewise strengthen the lower bound on evolution time to Ω_k(s^{1−1/(2k)}/ε), leaving the algorithm within a factor Õ(s^{1/(2k)}) of optimal. The technique departs from earlier approaches. Instead of cancelling the unknown Hamiltonian or approximating its dynamics, it encodes energy expectation differences in the perturbative response of eigenphases of the actual controlled evolution, then extracts them through phase estimation and high-order extrapolation. It thus avoids the requirement that evolution steps shrink polynomially as target precision improves. A version using only single-qubit operations retains the same step size at the cost of a √n overhead in learning time for n qubits. The method also enables tolerant Hamiltonian certification and the estimation of a specified Pauli coefficient, the latter without locality or sparsity assumptions. For the record, then: the authors' own summary is that the results establish eigenphase engineering as a versatile approach to extracting Hamiltonian information with infrequent control. The abstract reports an algorithm and its bounds; it does not report an implementation on a named system, nor a schedule for one. The value claimed is mathematical, and it stands or falls on the analysis, not on a field test. —Inspector Grey Dispatch from The Prepared E0