arXiv:2610.33834
Initial-State Preparation for Strongly Correlated Molecules from Matrix Product States: Benchmarking and Environment-Aware Divide-and-Conquer Compilation
Haiqi Liu, Longfei Chang, Zhendong Li, Wei-Hai Fang
Abstract
Compiling classically optimized matrix product states (MPSs) into quantum circuits is a promising route to initial-state preparation for strongly correlated molecules, but the accuracy, gate cost, and classical scalability of existing MPS compilation methods have not been systematically assessed. Here, we benchmark representative analytical, variational, and hybrid MPS compilation methods for strongly correlated systems including hydrogen chains and iron-sulfur clusters. We compare state infidelity, CNOT gate count, classical runtime, and parallel scalability across different system sizes and bond dimensions. We also introduce a divide-and-conquer environment-aware compilation algorithm (DCE), which optimizes each local isometry approximation according to its contribution to the prepared-state fidelity. The results reveal complementary operating regimes: direct variational compilation methods are gate-efficient for some smaller molecular systems with $O(10)$ qubits and $O(10)$ bond dimensions or moderate accuracy targets with infidelity of order $10^{-1}$, whereas divide-and-conquer methods scale more favorably for larger systems and relatively higher bond dimensions of $O(10\text{-}10^2)$, particularly when target infidelities below $10^{-1}$ are required. Within the latter family, environment-aware optimization improves accuracy over unitary-distance-based compilation while retaining task-level parallelism. For the FeMo-cofactor (FeMoco), we find the DCE algorithm together with the MPS obtained in the entanglement-minimized orbitals (EMOs) provides a resource-efficient route to initial-state preparation. Our results provide practical guidance for selecting and developing MPS compilation methods for strongly correlated molecular simulations.
AI Summary
The paper asks how to compile classically optimized molecular matrix product states (MPSs) into approximate quantum circuits while balancing fidelity, gate count, and classical cost. It benchmarks analytical matrix-product disentanglers (MPD), two variational refinements, and divide-and-conquer unitary-distance compilation (DCD), and introduces environment-aware compilation (DCE). DCE independently optimizes each local isometry approximation against its contribution to state fidelity, keeping all other blocks exact during optimization; these independent tasks can run in parallel. Numerical benchmarks cover hydrogen chains up to 100 qubits, iron–sulfur clusters including 152-qubit FeMoco, and random real MPSs. Direct variational methods can use fewer CNOTs for smaller systems: for 20-qubit hydrogen with bond dimension $32$, they reach infidelity below $10^{-2}$ with fewer than 30 CNOTs per qubit. Their performance deteriorates for larger systems within the tested circuit budgets. DCE reaches approximately $10^{-2}$ infidelity or better across the hydrogen-chain size benchmarks and improves on DCD’s fidelity–CNOT tradeoff for the molecular states. DCE and DCD perform similarly on random MPSs. A triangle-inequality argument establishes $\arcsin\sqrt{\mathcal L}\leq\sum_n\arcsin\sqrt{\mathcal L_n}$, giving the conservative sufficient local-error scaling $\mathcal L_n\lesssim\epsilon/N^2$ for global infidelity $\epsilon$. The tests concern fidelity to the input MPS, not directly to the exact molecular ground state. Calculations use real-valued states, finite depth budgets, and modest bond dimensions; full-circuit optimization is restricted to ten layers because of contraction costs. Reported ideal parallel timings assume sufficient resources. Symmetry preservation, hardware connectivity, and error-aware allocation remain future extensions.
Error found: The FeMoco initial-state-preparation conclusion lacks reported ground-state overlaps or compression errors for its low-bond-dimension MPS targets, so the demonstrated compilation fidelity does not establish end-to-end QPE usefulness.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.