arXiv:2610.03528
Breaking the chain: geometry-native state preparation with ASPIRE
Fredrik Hasselgren, Matthew L. Sims-Goh
Abstract
Loading data into quantum computers is a key bottleneck that threatens end-to-end quantum advantage. Quantum-circuit compilers for structured states like matrix product states (MPS) are a critical enabler to manage this in quantum chemistry, finance, and many-body quantum simulation. Approximate preparation of MPS typically relies on layered sequential staircases of nearest-neighbour gates. We present Adaptive State Preparation by Iterative Removal of Entanglement (ASPIRE), an algorithm that approximately compiles state-preparation circuits by selecting gates based on the state's underlying entanglement geometry and the available connectivity of the target quantum processor. Our protocol scores candidate qubit pairs on the removability of their entanglement and identifies layers of parallelised long-ranged gates that remove the most entanglement. We demonstrate our protocol empirically across a range of settings, illustrating its success for preparing Hamiltonian ground states and multivariate amplitude-encoded functions, and accounting for qubit connectivity, hardware noise, and both NISQ and early fault-tolerant settings. Throughout, ASPIRE typically achieves higher fidelities and shallower circuits than existing methods for states with long-ranged entanglement, even when classical optimisation is accounted for. Our investigation positions ASPIRE as a competitive, resource-frugal state-preparation protocol, with particular promise for applications tolerant to imperfect fidelity such as preparation of guide states for quantum phase estimation.
AI Summary
Approximate compilation of matrix product states (MPS) commonly uses sequential nearest-neighbour disentangling circuits, even when correlations or hardware connectivity are not one-dimensional. ASPIRE instead chooses two-qubit gate locations adaptively from the residual state’s correlations, subject to connectivity and gate-budget constraints. For each admissible pair, a $4\times4$ eigendecomposition supplies the unitary maximally reducing its mutual information and the corresponding removable-correlation score. Maximum-weight matching selects disjoint gates for parallel layers; reversing the disentangling circuit gives state preparation. The paper establishes exact additive descent of the total single-site entropy $T=\sum_a S(\rho_a)$ and bounds preparation error using single-site marginals. Pair scoring costs $O(N^2\chi^3)$ per layer in the MPS implementation. Optional Procrustes sweeps refine gates without changing their locations. Numerical tests cover two-dimensional spin ground states, random MPS, amplitude-encoded functions, and noisy heavy-hex circuits. For a $100$-qubit transverse-field Ising target, unoptimised ASPIRE reaches amplitude-overlap infidelity $\varepsilon=1-|\langle\psi|\psi_C\rangle|=0.0161$, versus $0.124$ for the staircase baseline at equal estimated $T$ count. Advantages persist in several optimisation comparisons with matched classical budgets, but random-MPS controls show little advantage. Bivariate Gaussian tests report gate savings of roughly $1.8$–$4.7$ at $\varepsilon=10^{-3}$, depending on size and ordering. The method remains limited by classical state-representation costs, truncation, routing, and greedy selection. Some comparisons deliberately use orderings unfavourable to the baseline; noise studies mainly use global depolarisation. Pair scores can vanish despite residual multipartite correlations, including for non-stabiliser states; composite moves or an MPD fallback are proposed. No general convergence or fidelity-advantage guarantee is established.
Error found: The 30-qubit heavy-hex advantage uses an imposed asymptotic MPD routing multiplier of 1.4, not an actual routing compilation or an attained routing cost for the simulated patch. The noise benchmark labels counts as native two-qubit gates while charging and applying noise once per generic SU(4) gate, inconsistent with the formal model's three-CNOT cost per such gate. The QPE figure switches from the benchmarks' amplitude-infidelity convention to squared-overlap infidelity, making its approximately 1.7e-3 threshold equivalent to approximately 8.7e-4 in benchmark units.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 34 pages, 11 figures