A critical review plus small exact-FCI experiments concludes that sample-based quantum diagonalization has not beaten classical selected CI and maps where, if anywhere, a quantum or generative advantage could survive.
Noise and Configuration Recovery Impact on Quantum Selected Configuration Interaction
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
Quantum-selected configuration interaction (QSCI) is a promising hybrid quantum-classical approach in which a quantum device generates configurations for subsequent classical diagonalization. Here, we analyze the performance of QSCI combined with the local unitary cluster Jastrow (LUCJ) ansatz, focusing on the interplay between ansatz expressivity, sampling, noise, and configuration recovery. Using the dissociation of N2 in a large active space as a benchmark, we show that noiseless LUCJ sampling produces compact and biased configurational spaces, limiting the accuracy of the resulting CI energies, particularly in strongly correlated regimes. By introducing a simple noise model, we demonstrate that sampling noise can enhance Hilbert-space exploration by generating additional configurations beyond those supported by the ideal ansatz. When combined with configuration recovery, this leads to systematically improved energies. Moreover, recovery alone (starting from randomly generated configurations) can efficiently construct accurate CI spaces, highlighting its central role in QSCI.
fields
quant-ph 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Machine learning for sample-based quantum diagonalization: generative configuration recovery and the classical-simulability frontier
A critical review plus small exact-FCI experiments concludes that sample-based quantum diagonalization has not beaten classical selected CI and maps where, if anywhere, a quantum or generative advantage could survive.