arXiv:2610.31070
Norm-Aware Product Formulas for Hamiltonian Simulations
Carlos A. Piñeiro Moreno, Pascal Krause, Annika Bande
Abstract
With the target of efficiently solving the quantum-mechanical Schrödinger equation on future quantum computers, first- and second-order product formulas for the time-evolution operator are proposed, designed to exploit hierarchies in the operator norms of the individual Hamiltonian components. Inspired by the THRIFT product formula, these decomposition schemes achieve a suppression of the leading error term by a factor of alpha<1 with respect to the standard Lie--Trotter--Suzuki splitting of the same order, where alpha quantifies the norm ratio between the dominant and subdominant parts of the Hamiltonian. Unlike THRIFT, the proposed formulas retain the general product-formula structure, making them applicable to Hamiltonians whose terms act on an arbitrary number of qubits. The advantages of this approach are demonstrated through numerical experiments on engineered two-dimensional spin-1/2 XY model systems and isolated molecules within the double-factorization framework.
AI Summary
The paper studies product-formula simulation of Hamiltonians $H=H_0+\alpha\sum_{k=1}^L H_k$, with a dominant, efficiently exponentiable $H_0$ and small $|\alpha|$. It seeks THRIFT-like suppression of leading errors while using only exponentials of individual Hamiltonian components, rather than grouped terms $H_0+\alpha H_k$. The first-order “Partial–Strang” formula sandwiches a first-order splitting of the perturbation between two half-steps of $H_0$. Triangle inequalities and commutator bounds give local error $O(\alpha^2t^2)+O(|\alpha|t^3)$, compared with $O(|\alpha|t^2)$ for ordinary first-order splitting. A second-order construction combines an existing five-factor, two-operator splitting with symmetric splittings of the perturbation. Choosing $\lambda=1/2-\sqrt{3}/6\approx0.211325$ cancels the cubic error linear in $\alpha$, leaving a leading cubic contribution proportional to $\alpha^2$. It uses $4L+1$ exponentials, versus $2L+1$ for second-order Suzuki splitting, and is compared with an existing related optimized formula. Exact classical simulations benchmark a nine-spin XY lattice and five STO-3G molecular Hamiltonians in double-factorized form. Molecular tests report first-order state-error reductions of $4.5$–$8$ relative to Lie–Trotter, using 100 steps to total time one atomic unit. The stated scaling advantages concern short steps and fixed decompositions. Partial–Strang loses quadratic $\alpha$ scaling when its $|\alpha|t^3$ residual dominates. Molecular gains depend on commutators and fragment counts, not norm ratios alone. Benchmarks use small systems, Frobenius or state-distance errors, and no quantum hardware; the circuit discussion does not quantify end-to-end resource savings.
Error found: The first-order formula has an additional O(|alpha| t^3) error, so its quadratic-alpha behavior requires sufficiently small steps relative to alpha, as the appendix acknowledges. The second-order formula generally retains an O(|alpha| t^5) error, so its quadratic-alpha behavior is conditional even though its leading t^3 error coefficient has the claimed suppression. Equation 'second_order_alpha' incorrectly equates a full perturbation step with a half-step product, although the final second-order construction uses the half steps correctly. The molecular benchmark's Frobenius-norm ratios do not directly establish the operator-norm hierarchy assumed in the theoretical analysis.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 17 pages, 6 figures, submitted to Academia Quantum