arXiv:2609.38091
Optimal Ground-State Preparation with a Guiding State
Stacey Jeffery, Rolando D. Somma, Freek Witteveen, Ronald de Wolf
Abstract
Suppose a Hamiltonian $H$ has a unique ground state $|ψ_0\rangle$ with an eigenvalue $E_0$, and we have an estimate $\tilde{E}_0$ such that $|\tilde{E}_0-E_0|\leqδ$, and there is a gap of at least $3δ$ between $E_0$ and all other eigenvalues. Suppose we have a unitary $A$ available that can produce a "guiding state'" $A|0\rangle$ that has overlap at least $γ$ with $|ψ_0\rangle$. We show how to obtain an $η$-approximation of $|ψ_0\rangle$ with probability at least $1-\varepsilon$ using $O(\log(1/\varepsilon)/γδ+ \log(1/η)/δ)$ applications of $U=e^{iH}$ its inverse, and $O(\log(1/\varepsilon)/γ)$ applications of $A$ and its inverse. We give two different algorithms, one based on interleaving amplitude amplification and error-reduction in the style of [HMdW03], and one using the composition of transducers. This paper is the state-preparation follow-up to our two recent ground-state-energy estimation papers [JW26,SdW26]. Combined with the optimal energy-estimation of [JW26], our results show the claimed upper bound for ground-state preparation. The bound is optimal up to a constant factor in terms of the number of applications of $U$ and $U^\dagger$ if $\varepsilon=η$, as follows from [SdW26]. In one of our approaches to ground state preparation, we construct transducers for amplitude amplification and LCU that might be of independent interest.
AI Summary
The paper studies preparation of a unique ground state given controlled access to $U=e^{iH}$ and a guiding-state unitary $A$ whose output has ground-state overlap at least $\gamma$. The spectrum is normalized to $[0,\pi]$, the spectral gap is at least $\Delta$, and initially the ground energy is known within $\Delta/3$. It gives algorithms producing a state within trace-norm distance $\eta$ of the ground state, with success probability at least $1-\varepsilon$, using $O(\log(1/\varepsilon)/(\gamma\Delta)+\log(1/\eta)/\Delta)$ calls to $U$ and its inverse, and $O(\log(1/\varepsilon)/\gamma)$ calls to $A$ and its inverse. This removes a multiplicative $\log(1/\gamma)$ overhead from previous bounds. A lower bound from prior work establishes optimality of the evolution-query count when $\varepsilon=\eta$. Combining with prior energy estimation removes the supplied-estimate assumption without changing that count. Two constructions establish the bound: recursive robust amplitude amplification interleaved with increasingly accurate phase-estimation verification, and composition of transducers for spectral filtering and amplitude amplification. The latter includes an LCU construction costed by the first moment of its Fourier coefficients and an amplitude-amplification transducer with reduced space overhead. Staged verification separates the dependence on output accuracy from repeated preparation costs. The guarantees concern oracle-query complexity, not end-to-end Hamiltonian-simulation cost. The transducer construction has logarithmic ancilla overhead and polylogarithmic gate overhead; its loose constant bounds may offset the logarithmic improvement in practice. Independent concurrent work obtains similar query-optimal preparation results.
We also used LLMs for generating some figures and proofs, and extensively for proofreading.
Central claim may not hold: Without the body's spectrum restriction or an equivalent no-aliasing promise, the abstract's ground-state preparation claim is false because distinct Hamiltonians with orthogonal ground states can provide identical evolution and guiding-state oracles. The abstract leaves controlled-oracle access unspecified, although the main theorem explicitly assumes controlled calls, an additional capability not generally supplied by arbitrary black-box unitaries and their inverses.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- v2: improved the presentation