arXiv:2610.16379
Benchmarking a randomized quantum algorithm for estimating expectation values of molecular observables
Anna Schroeder, Gian-Luca R. Anselmetti, Matthias Degroote, Florian Dommert, Michael Kaicher, Chandan Kumar, Michael Streif, Davide Vodola
Abstract
Randomized quantum algorithms for the estimation of expectation values of molecular observables offer a promising route to early fault-tolerant quantum advantage by trading circuit depth for an increased number of samples. However, this begs the question whether the resulting measurement overhead remains feasible for industrially relevant molecular systems. In this work, we benchmark the randomized sampling algorithm of Wang et al. [1] through numerical simulations on a hydrogen molecule and resource estimates for various active spaces of the cytochrome P450, an industry-relevant system in the pharmaceutical industry. Our numerical simulations reveal that the analytical error bounds significantly overestimate the actual sampling error, whereas the analytical bound on the required imaginary evolution time is comparatively tight, exceeding a direct DMRG estimate by roughly 44% for the active space where the comparison is feasible. The measurement overhead nevertheless remains prohibitively high: even for the smallest active spaces, the expectation-value estimation alone requires around 1010 samples, scaling to beyond 1015 for the largest active spaces considered. These results suggest that substantial improvements in measurement strategies or algorithmic scaling are necessary before randomized early-fault-tolerant algorithms become competitive for industrial quantum chemistry applications.
AI Summary
The paper asks whether the depth–sampling tradeoff in randomized ground-state property sampling (RGPS) permits practical molecular-observable estimation on early fault-tolerant hardware. It benchmarks an existing algorithm combining a Fourier approximation to Gaussian imaginary-time filtering, randomized Pauli compilation, and numerator/normalization estimation, using one auxiliary qubit. Classical simulations treat $\mathrm{H}_2$ in STO-3G, with ten repetitions of $10^6$ sampled circuits for several compilation depths. They show decreasing sampling error followed by a filtering-bias plateau; analytical error bounds exceed the observed errors. These simulations evaluate each circuit expectation exactly and omit Hadamard-test measurement noise. Resource estimates use DMRG-derived gaps and initial-state overlaps for cytochrome P450 active spaces spanning 8–58 orbitals. Optimizing the numerator/denominator sampling allocation gives a cost estimate scaling as $M\propto R^4(\|O\|+|\langle O\rangle|)^2e^{2\lambda_0^2\tau^2}/(\varepsilon^2\gamma^4)$. At $1$ mHa precision and failure probability $0.01$, estimated counts range from approximately $10^{10}$ to $10^{14}$–$10^{15}$ per Hamiltonian expectation, with maximal depths around $10^{11}$ non-Clifford operations for the largest space. Reducing depth below the scale $(\lambda\tau)^2$ rapidly increases sampling costs. For a 15-orbital space, the analytical filtering time exceeds a DMRG-based estimate by about 44%. The large-system estimates rely on conservative sampling bounds, heuristic gap/overlap inputs, an assumed $1$ mHa residual energy after shifting, and uncompressed Pauli decompositions. The DMRG filtering-time comparison extrapolates far beyond simulated times. Large-system force measurements and full workflow costs are not evaluated.
Central claim may not hold: The P450 counts are conservative bound-based budgets rather than necessary sampling requirements, and the hydrogen comparison additionally removes measurement noise present in the analytical bounds. The reported 44% imaginary-time comparison is numerically correct but rests on extrapolating DMRG evolution from tau^2 = 10 to approximately 1090 Ha^-2, not direct verification near the target. The resource estimates do not establish a combined 1 mHa ground-state-energy error guarantee accounting for state preparation, Fourier approximation, and statistical estimation. An overlap obtained by compressing a DMRG state does not certify an upper bound on the overlap achievable by arbitrary single-determinant initial states. The optimized sampling-budget derivation combines confidence tolerances using Gaussian error propagation without establishing the claimed high-probability bound for the ratio estimator.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.