arXiv:2610.03103
Orbital-Rotation Shadow Tomography Reduces the Classical Cost of Quantum-Classical Auxiliary-Field Quantum Monte Carlo
Luning Zhao, Joshua Goings, Evgeny Epifanovsky, Martin Roetteler
Abstract
Quantum-classical auxiliary-field quantum Monte Carlo (QC-AFQMC) is a promising route to using quantum computers for strongly correlated electronic structure problems, but its scalability is limited by a classical post-processing bottleneck: the measured quantum trial state must be used to evaluate overlaps, force biases, and local energies for many walkers throughout the AFQMC propagation. Existing matchgate-shadow approaches avoid the exponential cost of generic classical shadows, but their post-processing remains expensive for large systems. In this work, we show that particle-number-preserving fermionic shadows, implemented as orbital-rotation shadows, are able to drastically reduce QC-AFQMC post-processing costs. We derive efficient estimators for overlaps and their derivatives, giving $O(N^3)$ scaling for overlaps and force biases and $O(N^4)$ scaling for local energies per shadow and walker and with $O(N)$ Cholesky vector. For fixed normalized target determinants, we also prove that the overlap estimator is unbiased with variance upper bounded by $O(η)$, where $η$ is the particle number, yielding a sufficient budget of $O(η/ε^2)$ ideal independent measurements for additive overlap error $ε$. We combine these estimators with a novel virtual-correlation-energy treatment and a GPU implementation and demonstrate QC-AFQMC using orbital-rotation shadows. Performance projections for $π$-conjugated molecules up to pentacene with a (22e, 22o) active space and 356 total orbitals show that the full classical post-processing workflow can be completed in slightly more than two days on 8$\times$A100 graphics processing units (GPUs). These results significantly reduce classical post-processing costs for QC-AFQMC and motivate further improvements in preparation and measurement of the trial state on quantum hardware.
AI Summary
QC-AFQMC requires repeated overlaps, force biases, and local energies between a quantum trial state and Slater-determinant walkers. This paper studies whether particle-number-preserving orbital-rotation shadows can reduce their classical post-processing cost. The protocol adds $\eta$ ancillary modes to an $N$-mode, $\eta$-particle state, prepares a coherent superposition with an ancillary reference determinant, and measures after Haar-random orbital rotations. Specializing the fixed-particle inverse measurement channel reduces each overlap estimator to an $\eta\times\eta$ determinant; its derivatives supply force-bias and local-energy numerators. Overlaps cost $O(N\eta^2)$ per shadow and walker. At fixed filling with $O(N)$ Cholesky vectors, overlaps and force biases are cubic and local energies quartic, matching recent matchgate-shadow asymptotics. A QR-based virtual-correlation-energy formulation simplifies full-space processing. An exact Born-weighted second-moment calculation establishes unbiased overlap estimation with variance at most $4(\eta+1)(N+2)/(N+\eta+2)-1\leq4\eta+3$, and a uniform lower bound $\eta+1$. Thus $O(\eta/\epsilon^2)$ independent ideal records suffice for fixed-target additive RMS error $\epsilon$. A simulated N$_2$ demonstration uses 20,000 shadows and 1,024 walkers, agreeing with CCSD(T) within a 0.001-Ha error bar. GPU timing models project pentacene CAS(22e,22o) post-processing at about 55 hours on eight A100s using 65,536 shadows. Large-system benchmarks use HF-generated pseudo-shadows, not correlated-state demonstrations. The sub-hour projection on 320 H200s assumes perfect strong scaling. Quantum measurement time is excluded; extra ancillas and arbitrary-angle circuits increase hardware requirements. Fixed-overlap bounds do not establish final AFQMC energy accuracy or control shadow-dependent trajectories and ratios near overlap zeros.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.