Quantum subspace expansion (QSE) offers promising avenues to perform spectral calculations on quantum processors but comes with a large measurement overhead. Informationally complete measurements, such as classical shadows, were recently proposed to overcome this bottleneck. Here, we report the first large-scale implementation of QSE with informationally complete measurements. In particular, we probe the quantum phase transition of a spin model with three-body interactions, for which we observe accurate ground state energy recovery and mitigation of local order parameters across system sizes of up to 80 qubits. We achieve this by reformulating QSE as a constrained optimization problem, obtaining rigorous statistical error estimates and avoiding numerical ill conditioning. With over 3×10^{4} measurement basis randomizations per circuit and the evaluation of O(10^{14}) Pauli traces, this represents one of the most significant realizations of classical shadows to date.
Quantum subspace expansion (QSE) offers promising avenues to perform spectral calculations on quantum processors but comes with a large measurement overhead. Informationally complete measurements, such as classical shadows, were recently proposed to overcome this bottleneck. Here, we report the first large-scale implementation of QSE with informationally complete measurements. In particular, we probe the quantum phase transition of a spin model with three-body interactions, for which we observe accurate ground state energy recovery and mitigation of local order parameters across system sizes of up to 80 qubits. We achieve this by reformulating QSE as a constrained optimization problem, obtaining rigorous statistical error estimates and avoiding numerical ill conditioning. With over 3×10^{4} measurement basis randomizations per circuit and the evaluation of O(10^{14}) Pauli traces, this represents one of the most significant realizations of classical shadows to date.
Accurate estimation of observables in quantum systems is a central challenge in quantum information science, yet practical implementations are fundamentally constrained by the limited number of measurement shots. In this work we explore a variation of the classical shadows protocol in which the measurements are kept local while allowing the resulting classical shadows themselves to be correlated. By constructing locally optimal shadows, we obtain unbiased estimators that are competitive with state-of-the-art methods in terms of measurement overhead, while requiring only single-qubit measurements and allowing for the estimation of any observable in pure post-processing, reducing estimation errors by orders of magnitude over standard classical shadows. We validate our approach through numerical experiments on molecular Hamiltonians with up to 40 qubits consistently observing significant reductions in estimation errors, including for estimations of multiple chemically relevant observables simultaneously.
We introduce a postprocessing technique for classical shadow measurement data that enhances the precision of ground state estimation through high-dimensional subspace expansion; the dimensionality is only limited by the amount of classical postprocessing resources rather than by quantum resources. Crucial steps of our approach are the efficient identification of useful observables from shadow data, followed by our regularized subspace expansion that is designed to be numerically stable even when using noisy data. We analytically investigate noise propagation within our method, and upper bound the statistical fluctuations due to the limited number of snapshots in classical shadows. In numerical simulations, our method can achieve a reduction in the energy estimation errors in many cases, sometimes by more than an order of magnitude. We also demonstrate that our performance improvements are robust against both coherent errors (bad initial state) and gate noise in the state-preparation circuits. Furthermore, performance is guaranteed to be at least as good—and in many cases better—than direct energy estimation without using additional quantum resources, and the approach is thus a very natural alternative for estimating ground state energies directly from classical shadow data.
The wave function prepared by a variational quantum eigensolver (VQE) can contain a substantial classically tractable component, yet its energy contribution is sampled on the quantum device. Representing the wave function as a combination of a classical reference and separately prepared quantum states allows the reference energy to be evaluated without quantum sampling. However, near-linear dependence among these components can amplify measurement errors and destabilize the resulting energy estimate. We introduce Classical Active-Space Hybrid Quantum Subspace Expansion (CASH-QSE), which retains a complete-active-space self-consistent field (CASSCF) reference classically and uses occupation structure to construct quantum states exactly orthogonal to the reference and to one another. The energy follows from an ordinary Hermitian eigenvalue problem without overlap measurements, while the same occupation constraints simplify the measured operators. Using full configuration interaction to guide component selection, we test CASH-QSE along H2O and N2 bond-stretching coordinates spanning weakly to strongly correlated regimes. The benchmarks use STO--3G for both molecules and a restricted cc-pVDZ orbital space for H2O. CASH-QSE reaches chemical accuracy while limiting the largest complete measurement circuits to a few hundred all-to-all logical controlled-NOT gates. In favorable cases, it also reduces the idealized final-energy sampling cost by several orders of magnitude relative to VQE with an adaptive derivative-assembled pseudo-Trotter ansatz (ADAPT-VQE).
Accurate treatment of excited states remains challenging for classical electronic-structure methods, motivating quantum algorithms that may provide more favorable resource scaling. We systematically assess the practical performance of Quantum Subspace Expansion (QSE) for low-lying electronic states across 28 organic molecules from the Thiel benchmark set. Statevector simulations reveal that QSE representability depends strongly on the target state, active space, and molecular electronic structure, with multiconfigurational character emerging as an important factor. In several cases, the one-body QSE subspace cannot adequately represent the target states, while explicit double-excitation operators improve representability at the cost of a larger operator space. Analysis of the QSE overlap matrix shows that although eigenmodes with small eigenvalues drive numerical instability, they cannot be discarded indiscriminately without compromising spectral accuracy. Under finite-shot sampling, energy errors exhibit an empirical scaling of approximately n2 orb/ √ Nshots, which, combined with the growth in measurement groups, yields a total measurement cost of approximately O(n11 orb) for constant accuracy under our direct-estimation protocol. These results expose a central tension between representability, numerical stability, and measurement cost. Encouragingly, we find that the projected matrices exhibit recurring structural patterns, while the finite-shot perturbations in the Hamiltonian and overlap matrices show nontrivial covariance, suggesting opportunities for more efficient structure-aware measurement and reconstruction strategies. Overall, our results provide an assessment of QSE across chemically-relevant organic molecules, identifying present limitations while highlighting potential routes toward more practical future implementations.
Improving the performance of quantum algorithms is a fundamental task to achieve quantum advantage. In many cases, extracting information from quantum systems poses an important challenge for practical implementations of real-world quantum computers, given the high resource cost of performing state tomography. In this scenario, randomized measurements have emerged as a promising tool. In particular, the classical shadow protocol allows one to retrieve expected values of low-weight Pauli observables by performing only local measurements. In this paper, we present an implementation of the recently introduced feedback-based algorithm for quantum optimization using the Ket quantum programming platform to solve the MaxCut optimization problem. We employ classical shadows for the feedback routine of parameter estimation and compare this approach with the direct estimation of observables. Our results show that depending on the graph geometry of the MaxCut problem, the number of measurements required to estimate expected values of observables with classical shadows can be up to 16 times lower than with direct-observable estimation. Furthermore, by analyzing complete graphs, we numerically confirm logarithmic growth in the required number of measurements relative to the number of observables, reinforcing that classical shadows can be a useful tool for estimating low-locality Pauli observables in quantum algorithms.
Accurate treatment of excited states remains challenging for classical electronic-structure methods, motivating quantum algorithms that may provide more favorable resource scaling. We systematically assess the practical performance of Quantum Subspace Expansion (QSE) for low-lying electronic states across 28 organic molecules from the Thiel benchmark set. Statevector simulations reveal that QSE representability depends strongly on the target state, active space, and molecular electronic structure, with multiconfigurational character emerging as an important factor. In several cases, the one-body QSE subspace cannot adequately represent the target states, while explicit double-excitation operators improve representability at the cost of a larger operator space. Analysis of the QSE overlap matrix shows that although eigenmodes with small eigenvalues drive numerical instability, they cannot be discarded indiscriminately without compromising spectral accuracy. Under finite-shot sampling, energy errors exhibit an empirical scaling of approximately n_{\mathrm{orb}}^2/\sqrt{N_{\mathrm{shots}}} , which, combined with the growth in measurement groups, yields a total measurement cost of approximately \mathcal{O}(n_{\mathrm{orb}}^{11}) for constant accuracy under our direct-estimation protocol. These results expose a central tension between representability, numerical stability, and measurement cost. Encouragingly, we find that the projected matrices exhibit recurring structural patterns, while the finite-shot perturbations in the Hamiltonian and overlap matrices show nontrivial covariance, suggesting opportunities for more efficient structure-aware measurement and reconstruction strategies. Overall, our results provide an assessment of QSE across chemically-relevant organic molecules, identifying present limitations while highlighting potential routes toward more practical future implementations.