arXiv:2609.37095
A quantum double-bracket algorithm for imaginary-time evolution with exponentially shorter depth
Ioannis Kolotouros, Raul Garcia-Patron
Abstract
Imaginary-time evolution (ITE) is a widely used technique for preparing the ground state of a target Hamiltonian. Building on the recently proposed framework of Double-Bracket quantum ITE (DB-QITE), we introduce a probabilistic variant (PDBQITE) that achieves an exponential reduction in circuit depth. Our algorithm requires only a single ancilla qubit and a quantum circuit that consists of one controlled real-time evolution and one mid-circuit measurement per iteration. The trade-off is an exponential decay of the total success probability with the number of iterations, for which we derive an analytical lower bound that is independent of system size. Our numerical simulations show that the per-step success probability remains close to unity, making it realistic to execute many iterations, thus providing a substantial improvement over prior constructions. We show that PDBQITE outperforms the near-term probabilistic algorithm PITE for ground-state preparation of molecular Hamiltonians, and that a hybrid QAOA-PDBQITE scheme achieves higher approximation ratios than QAOA at the same circuit depth on random regular graphs for the MaxCut problem.
AI Summary
Imaginary-time evolution (ITE) can prepare ground states, but the recursive unitary construction in double-bracket QITE (DB-QITE) requires $O(3^K)$ real-time-evolution calls for $K$ iterations. This paper asks whether those iterates can be implemented with shorter circuits by accepting postselection. The proposed probabilistic algorithm, PDBQITE, rewrites each DB-QITE update as a state-dependent linear combination of the identity and a real-time evolution. Conditional on success and exact coefficient estimates, it reproduces the corresponding DB-QITE iterate. Each first-order iteration uses one ancilla, one controlled real-time evolution, and one mid-circuit measurement, giving linear rather than exponential depth growth. An analytical calculation gives $P_{\mathrm{succ},k}\geq(1+\sqrt{s_k})^{-2}$. A higher-order commutator construction uses two ancillas and two controlled evolutions, with $P^{(2)}_{\mathrm{succ},k}\geq(1+\sqrt{5s_k}+s_k)^{-2}$. For constant timesteps, coefficients can be inferred from initial-state time-series overlaps rather than measurements of postselected intermediate states. Classical simulations cover molecular ground-state preparation, spin models, and QAOA-assisted optimization. For LiH, the higher-order variant reaches chemical accuracy in two iterations versus approximately eight for first order, with roughly tenfold higher cumulative success probability. On weighted regular MaxCut graphs up to 18 qubits, appending PDBQITE steps improves approximation ratios relative to additional QAOA layers under the stated gate-count accounting. Cumulative success probability still decays exponentially with iteration count. Performance depends on initial overlap and timestep choice; large steps can lose cooling behavior. Shot-noise propagation in coefficient estimation, hardware noise, and efficient timestep optimization remain unresolved.
Error found: The abstract's blanket description of per-step success probabilities as close to unity is inconsistent with the paper's statement that observed probabilities closely track bounds that are approximately 0.395 and 0.078 in reported operating regimes. The claimed equal-circuit-depth advantage over QAOA is supported only by gate counts, while the supplied implementation introduces shared-ancilla serialization that the depth comparison does not account for. The discussion asserts non-increasing energy at every successful step without timestep restrictions, although the appendix explicitly reports increasing first-order energy at timestep s=0.8.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 31 pages, 14 figures