arXiv:2611.00751
Quantum algorithm for weakly correlated molecular electronic ground states with sub-quintic scaling in the system size
Oskar Leimkuhler, Marvin Kronenberger, Davide Castaldo, Timothy Stroschein, Srushti Patil, Rebecca Tomann, Arta Safari, Marek Miller, Gemma C. Solomon, Matthias Christandl, Markus Reiher
Abstract
Weakly correlated electronic structures are predominant among molecules, and are routinely solved with heuristic classical methods such as the ``gold standard'' CCSD(T) coupled cluster model, incurring a formal time complexity $O(n^7)$ in the number of molecular orbitals and with no guarantee of energy error $\varepsilon$ at or below chemical accuracy. In this work, a quantum algorithm is presented for ground state energy estimation of weakly correlated molecules using Trotterized time evolution, with an end-to-end quantum operations cost that scales as $\tilde{O}(n^{4.8}/\varepsilon^2)$ in the limit of high-order product formulae when run as a low-depth, shot-noise limited algorithm, or $\tilde{O}(n^{4.8} + n^{3.8}/\varepsilon)$ when scaling the simulation time $T\propto 1/\varepsilon$. The algorithm combines adiabatic state preparation using a smoothed Gevrey schedule with eigenstate filtering via quantum prolate diagonalization (QPD). Both steps of the algorithm rely on a size-independent spectral gap assumption, which is validated by numerical studies on alanine chains with up to 64 monomers. We benchmark the performance of our eigenstate filtering algorithm on a set of 28 molecules to compare our approach against prior phase estimation algorithms, finding that QPD achieves state-of-the-art cost tradeoffs for $\varepsilon$-accurate results. Physical quantum resource estimates are obtained for running Trotter step circuits with up to 40 spin-orbitals, providing a clear path for scaling up quantum simulation of chemistry on early fault-tolerant hardware.
AI Summary
The paper studies ground-state energy estimation for weakly correlated molecules, including state preparation rather than assuming a favorable initial overlap. It combines adiabatic preparation along a Hartree–Fock-to-interacting interpolation using a smooth Gevrey schedule with quantum prolate diagonalization (QPD) of measured autocorrelation functions. The analysis assumes efficient Hartree–Fock convergence, a size-independent spectral gap throughout the interpolation, and a Hamiltonian spectral norm scaling extensively with system size. Under these assumptions, adiabatic preparation takes simulation time $\tilde O(n/\Delta^2)$. A single-frequency QPD estimator supports a depth–sampling tradeoff: evolution time $O(1/\Delta)$ with approximately $O(\varepsilon^{-2})$ shots, or $O(1/\varepsilon)$ evolution with polylogarithmically many circuit runs. Combining product-formula error bounds with double-factorized simulation gives high-order-limit gate costs $\tilde O(n^{4.8}/\varepsilon^2)$ or $\tilde O(n^{4.8}+n^{3.8}/\varepsilon)$. These exponents additionally rely on empirical factorization-rank and Hamiltonian one-norm scalings, including $\lambda=O(n^{1.8})$. Numerical evidence includes Hartree–Fock orbital gaps for alanine chains through 64 residues, interpolation-gap calculations for one residue, and phase-estimation benchmarks on 28 systems with 4–16 spin-orbitals. QPD provides a runtime–depth tradeoff; QMEGS requires less total evolution time at chemical accuracy. Fault-tolerant estimates put one 40-spin-orbital Trotter step at roughly 71–1509 seconds, depending on architecture. The gap evidence does not establish asymptotic many-body gaps. Benchmarks omit adiabatic preparation and coherent errors, Trotter bounds are conservative, and physical estimates cover individual steps rather than complete energy calculations. A matched accuracy-dependent classical complexity comparison remains unresolved.
The authors used LLMs in conversational mode to survey the relevant literature and to help devise the propositions, theorems and lemmas
Central claim may not hold: The headline n^4.8 scaling additionally relies on empirical factorization ranks and a tentative Hamiltonian one-norm exponent that are not established asymptotically for growing molecules. The 64-residue calculations establish Hartree–Fock orbital gaps, not the size-independent interacting spectral gap throughout the adiabatic pathway required by the algorithm. Taking recursive Suzuki product-formula order to infinity without accounting for its exponential implementation overhead does not establish the stated polylogarithm-suppressed n^4.8 complexity. The QPD analysis asserts a false exponential lower bound on fixed-bandwidth Slepian eigenvalues, whose large-index tails instead decay super-exponentially. The QPD proof uses a false local Bernstein inequality for band-limited functions, leaving its supplied conditioning and error-guarantee argument unsupported.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 61 pages, 19 figures