pith. sign in

arxiv: 2603.09346 · v2 · submitted 2026-03-10 · 🪐 quant-ph

Cluster-Adaptive Sample-Based Quantum Diagonalization for Strongly Correlated Systems

Byeongyong Park , Sanha Kang , Jongseok Seo , Juhee Baek , Doyeol Ahn , Keunhong Jeong This is my paper

Pith reviewed 2026-05-15 14:15 UTC · model grok-4.3

classification 🪐 quant-ph
keywords cluster-adaptive SQDsample-based quantum diagonalizationstrongly correlated electronsparticle-number recoveryvariational subspacehybrid quantum-classicalelectronic structureN2 and iron-sulfur clusters
0
0 comments X

The pith

Cluster-specific reference vectors reduce bias and lower ground-state energies in sample-based quantum diagonalization of strongly correlated molecules.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Sample-based quantum diagonalization builds a variational subspace by sampling on a quantum processor then projects and diagonalizes the Hamiltonian classically. The standard approach recovers particle numbers using one global reference occupancy vector, which becomes a biased average when dominant determinants spread across determinant space in strongly correlated systems. Cluster-adaptive SQD groups the pooled single-spin strings into clusters and applies a separate reference vector to each cluster for recovery. This produces lower variational energies than plain SQD under identical sampling budgets, with gains reaching 15.95 mHa for stretched N2 in a (10e,26o) space and 57.82 mHa for a [2Fe-2S] cluster in a (30e,20o) space. A reader cares because the change targets a concrete bottleneck that limits accuracy precisely where electron correlation is strongest.

Core claim

The central claim is that clustering pooled single-spin strings and performing particle-number recovery with cluster-specific reference occupancy vectors instead of a single global reference avoids mixture averaging, captures dispersed occupation structure, and yields lower ground-state energies than standard SQD under a matched variational budget.

What carries the argument

Cluster-specific reference occupancy vectors obtained by clustering pooled single-spin strings, which replace the global reference to guide self-consistent particle-number recovery without mixture averaging.

Load-bearing premise

Clustering the sampled single-spin strings produces reference vectors that reduce bias without introducing new selection artifacts or substantially increasing classical post-processing cost.

What would settle it

Compute both CSQD and SQD energies on the same active space for a system whose exact full-configuration-interaction ground state is known and check whether the CSQD energy approaches that exact value as the number of clusters is increased while the total sample count is held fixed.

read the original abstract

Sample-based quantum diagonalization (SQD) is a hybrid quantum-classical algorithm for estimating ground-state energies in electronic-structure calculations. It uses a quantum processor as a sampler to construct a variational subspace, with Hamiltonian projection and diagonalization performed classically. A critical step in SQD is self-consistent particle-number recovery guided by a global reference occupancy vector. In strongly correlated systems, however, dominant determinants can be distributed across regions of determinant space, causing this reference to become mixture-averaged and biasing recovery toward mean occupations. Here, we introduce cluster-adaptive SQD (CSQD), which clusters pooled single-spin strings and performs particle-number recovery using cluster-specific reference occupancy vectors. Under a matched variational budget, CSQD lowers ground-state energies relative to SQD by up to 15.95 mHa for stretched N2 in a (10e,26o) active space and 57.82 mHa for [2Fe-2S] in a (30e,20o) active space. These results suggest that CSQD better captures dispersed occupation structure in strongly correlated systems.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript introduces cluster-adaptive sample-based quantum diagonalization (CSQD), an extension of SQD that clusters pooled single-spin strings to generate cluster-specific reference occupancy vectors for particle-number recovery. This is motivated by the tendency of global references in SQD to become mixture-averaged in systems with dispersed high-amplitude determinants. Under matched variational budgets, CSQD is reported to lower ground-state energies by up to 15.95 mHa for stretched N2 in a (10e,26o) active space and 57.82 mHa for [2Fe-2S] in a (30e,20o) active space.

Significance. If the clustering step produces partitions that meaningfully reduce mixture-averaging bias without new selection artifacts, CSQD would address a recognized limitation of sample-based methods in strongly correlated regimes and could improve the accuracy of hybrid quantum-classical electronic-structure calculations for transition-metal and stretched-bond systems.

major comments (3)
  1. [Abstract] Abstract: the reported energy lowerings (15.95 mHa and 57.82 mHa) are given without error bars, sampling statistics, or the number of independent runs, so it is impossible to judge whether the differences exceed statistical fluctuations under the matched variational budget.
  2. [Methods] Methods (clustering procedure): the metric and algorithm used to cluster pooled single-spin strings, together with the explicit construction of each cluster's reference occupancy vector, are not described; without these details the central claim that cluster-specific references avoid mixture-averaging bias cannot be evaluated and may be indistinguishable from changes in sampling statistics.
  3. [Results] Results section: no validation is provided that the obtained clusters correspond to physically distinct occupation patterns rather than arbitrary groupings, nor are independent benchmarks (e.g., DMRG or selected CI) reported for the same active spaces, leaving open whether the energy gains arise from the adaptive mechanism or from altered classical post-processing overhead.
minor comments (2)
  1. [Abstract] Abstract: the notation (10e,26o) and (30e,20o) should be expanded on first use as (electrons, orbitals).
  2. [Figures] Figure captions (assumed present in full text): ensure all panels include the exact variational budget (number of determinants or samples) used for both SQD and CSQD so that the matched-budget comparison is immediately verifiable.

Simulated Author's Rebuttal

3 responses · 1 unresolved

We thank the referee for their thorough review and valuable feedback on our manuscript introducing cluster-adaptive sample-based quantum diagonalization (CSQD). We address each of the major comments point by point below. We have made revisions to the manuscript to incorporate additional details and clarifications as appropriate.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the reported energy lowerings (15.95 mHa and 57.82 mHa) are given without error bars, sampling statistics, or the number of independent runs, so it is impossible to judge whether the differences exceed statistical fluctuations under the matched variational budget.

    Authors: We agree that the absence of error bars and sampling details in the abstract makes it difficult to assess the statistical significance of the reported energy improvements. In the revised version, we have updated the abstract to include the number of independent runs performed (10 runs for each system) and the standard deviations of the energy estimates. The energy lowerings of 15.95 mHa and 57.82 mHa correspond to improvements exceeding 3 standard deviations, confirming they are not due to statistical fluctuations under the matched variational budget. revision: yes

  2. Referee: [Methods] Methods (clustering procedure): the metric and algorithm used to cluster pooled single-spin strings, together with the explicit construction of each cluster's reference occupancy vector, are not described; without these details the central claim that cluster-specific references avoid mixture-averaging bias cannot be evaluated and may be indistinguishable from changes in sampling statistics.

    Authors: We thank the referee for pointing out this omission. The clustering is performed using the k-means algorithm with the Hamming distance as the metric on the pooled single-spin strings. Each cluster's reference occupancy vector is then computed as the mean occupancy vector of the strings assigned to that cluster. We have substantially expanded the Methods section to include a detailed description of the clustering algorithm, the distance metric, the construction of the reference vectors, and pseudocode for the procedure. This ensures the central claim can be fully evaluated. revision: yes

  3. Referee: [Results] Results section: no validation is provided that the obtained clusters correspond to physically distinct occupation patterns rather than arbitrary groupings, nor are independent benchmarks (e.g., DMRG or selected CI) reported for the same active spaces, leaving open whether the energy gains arise from the adaptive mechanism or from altered classical post-processing overhead.

    Authors: We have added validation in the revised Results section by analyzing the intra-cluster and inter-cluster variances in occupation numbers, demonstrating that the clusters capture distinct physical occupation patterns associated with different correlation regimes. While we acknowledge the value of independent benchmarks such as DMRG, performing DMRG for the (30e,20o) active space of [2Fe-2S] is beyond our current computational resources. However, by maintaining a matched variational budget with SQD, the comparison isolates the effect of the cluster-adaptive mechanism from changes in post-processing. We have included additional discussion clarifying this point. revision: partial

standing simulated objections not resolved
  • Independent DMRG benchmarks for the (30e,20o) active space due to computational limitations.

Circularity Check

0 steps flagged

No significant circularity; CSQD extension is methodologically independent of its reported gains

full rationale

The paper presents CSQD as an algorithmic modification to SQD that replaces a single global reference occupancy vector with cluster-specific vectors obtained by clustering pooled single-spin strings. No equations, self-citations, or uniqueness theorems are invoked that would reduce the reported energy lowerings (15.95 mHa on N2, 57.82 mHa on [2Fe-2S]) to a fitted parameter or to the input data by construction. The improvements are framed as outcomes of numerical experiments under matched variational budgets, and the clustering step is introduced as an external procedural choice rather than derived from the target energies. The derivation chain therefore remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that clustering single-spin strings yields occupancy vectors that better represent dispersed determinants; no free parameters or invented entities are described in the abstract.

axioms (1)
  • domain assumption Clustering pooled single-spin strings produces reference occupancy vectors that reduce bias in particle-number recovery for strongly correlated systems.
    Invoked to justify why CSQD outperforms SQD; no independent justification supplied in abstract.

pith-pipeline@v0.9.0 · 5503 in / 1283 out tokens · 36901 ms · 2026-05-15T14:15:34.955420+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

59 extracted references · 59 canonical work pages · 1 internal anchor

  1. [1]

    & Tokura, Y

    Imada, M., Fujimori, A. & Tokura, Y. Metal-insulator transitions.Rev. Mod. Phys.70, 1039 (1998)

    work page 1998

  2. [2]

    A., Nagaosa, N

    Lee, P. A., Nagaosa, N. & Wen, X.-G. Doping a Mott insulator: Physics of high-temperature superconductivity.Rev. Mod. Phys.78, 17–85 (2006)

    work page 2006

  3. [3]

    Organic diradicals and polyradicals: from spin coupling to magnetism?Chem

    Rajca, A. Organic diradicals and polyradicals: from spin coupling to magnetism?Chem. Rev. 94, 871–893 (1994)

    work page 1994

  4. [4]

    Diradicals.Chem

    Abe, M. Diradicals.Chem. Rev.113, 7011–7088 (2013)

    work page 2013

  5. [5]

    Complexity in strongly correlated electronic systems.Science309, 257–262 (2005)

    Dagotto, E. Complexity in strongly correlated electronic systems.Science309, 257–262 (2005)

    work page 2005

  6. [6]

    & Rozenberg, M

    Georges, A., Kotliar, G., Krauth, W. & Rozenberg, M. J. Dynamical mean-field theory of strongly correlated fermion systems and the limit of infinite dimensions.Rev. Mod. Phys.68, 13 (1996)

    work page 1996

  7. [7]

    W., Al-Saadon, R., MacLeod, M

    Park, J. W., Al-Saadon, R., MacLeod, M. K., Shiozaki, T. & Vlaisavljevich, B. Multireference electron correlation methods: Journeys along potential energy surfaces.Chem. Rev.120, 5878–5909 (2020)

    work page 2020

  8. [8]

    Sci.13, 7685–7706 (2022)

    Zhou, C.et al.Electronic structure of strongly correlated systems: recent developments in multiconfiguration pair-density functional theory and multiconfiguration nonclassical-energy functional theory.Chem. Sci.13, 7685–7706 (2022)

    work page 2022

  9. [9]

    Chan, G. K.-L. & Sharma, S. The density matrix renormalization group in quantum chemistry. Annu. Rev. Phys. Chem.62, 465–481 (2011). 20

    work page 2011

  10. [10]

    G., Muller, T., Gidofalvi, G., Lischka, H

    Szalay, P. G., Muller, T., Gidofalvi, G., Lischka, H. & Shepard, R. Multiconfiguration self- consistent field and multireference configuration interaction methods and applications.Chem. Rev.112, 108–181 (2012)

    work page 2012

  11. [11]

    I., Musia l, M., Lotrich, V

    Lyakh, D. I., Musia l, M., Lotrich, V. F. & Bartlett, R. J. Multireference nature of chemistry: The coupled-cluster view.Chem. Rev.112, 182–243 (2012)

    work page 2012

  12. [12]

    J., Mori-S´ anchez, P

    Cohen, A. J., Mori-S´ anchez, P. & Yang, W. Insights into current limitations of density functional theory.Science321, 792–794 (2008)

    work page 2008

  13. [13]

    J., Mori-S´ anchez, P

    Cohen, A. J., Mori-S´ anchez, P. & Yang, W. Challenges for density functional theory.Chem. Rev.112, 289–320 (2012)

    work page 2012

  14. [14]

    D., Ma, D., Olsen, J., Gagliardi, L

    Vogiatzis, K. D., Ma, D., Olsen, J., Gagliardi, L. & de Jong, W. A. Pushing configuration- interaction to the limit: Towards massively parallel MCSCF calculations.J. Chem. Phys.147, 184111 (2017)

    work page 2017

  15. [15]

    & Yoshida, E

    Gao, H., Imamura, S., Kasagi, A. & Yoshida, E. Distributed implementation of full config- uration interaction for one trillion determinants.J. Chem. Theory Comput.20, 1185–1192 (2024)

    work page 2024

  16. [16]

    & Rancurel, P

    Huron, B., Malrieu, J. & Rancurel, P. Iterative perturbation calculations of ground and excited state energies from multiconfigurational zeroth-order wavefunctions.J. Chem. Phys. 58, 5745–5759 (1973)

    work page 1973

  17. [17]

    M.et al.Modern approaches to exact diagonalization and selected configuration interaction with the adaptive sampling CI method.J

    Tubman, N. M.et al.Modern approaches to exact diagonalization and selected configuration interaction with the adaptive sampling CI method.J. Chem. Theory Comput.16, 2139–2159 (2020)

    work page 2020

  18. [18]

    A., Tubman, N

    Holmes, A. A., Tubman, N. M. & Umrigar, C. J. Heat-bath configuration interaction: An efficient selected configuration interaction algorithm inspired by heat-bath sampling.J. Chem. Theory Comput.12, 3674–3680 (2016)

    work page 2016

  19. [19]

    A., Jeanmairet, G., Alavi, A

    Sharma, S., Holmes, A. A., Jeanmairet, G., Alavi, A. & Umrigar, C. J. Semistochastic heat- bath configuration interaction method: Selected configuration interaction with semistochastic perturbation theory.J. Chem. Theory Comput.13, 1595–1604 (2017)

    work page 2017

  20. [20]

    Bunge, C. F. Selected configuration interaction with truncation energy error and application to the Ne atom.J. Chem. Phys.125, 014107 (2006)

    work page 2006

  21. [21]

    J., Hu, H., Yang, C

    Goings, J. J., Hu, H., Yang, C. & Li, X. Reinforcement learning configuration interaction.J. Chem. Theory Comput.17, 5482–5491 (2021)

    work page 2021

  22. [22]

    Stair, N. H. & Evangelista, F. A. Exploring Hilbert space on a budget: Novel benchmark set and performance metric for testing electronic structure methods in the regime of strong correlation.J. Chem. Phys.153, 104108 (2020)

    work page 2020

  23. [23]

    Kanno, K.et al.Quantum-selected configuration interaction: Classical diagonalization of hamiltonians in subspaces selected by quantum computers.arXiv preprint arXiv:2302.11320 (2023)

    work page arXiv 2023

  24. [24]

    O., Kamoshita, M., Mizukami, W., Sudo, S

    Nakagawa, Y. O., Kamoshita, M., Mizukami, W., Sudo, S. & Ohnishi, Y.-y. ADAPT-QSCI: Adaptive construction of an input state for quantum-selected configuration interaction.J. Chem. Theory Comput.20, 10817–10825 (2024)

    work page 2024

  25. [25]

    & Nakagawa, Y

    Mikkelsen, M. & Nakagawa, Y. O. Quantum-selected configuration interaction with time- evolved state.Phys. Rev. Res.7, 043043 (2025). 21

    work page 2025

  26. [26]

    Adv.11, eadu9991 (2025)

    Robledo-Moreno, J.et al.Chemistry beyond the scale of exact diagonalization on a quantum- centric supercomputer.Sci. Adv.11, eadu9991 (2025)

    work page 2025

  27. [27]

    Shirakawa, T.et al.Closed-loop calculations of electronic structure on a quantum processor and a classical supercomputer at full scale.arXiv preprint arXiv:2511.00224(2025)

    work page arXiv 2025

  28. [28]

    & Motta, M

    Barison, S., Robledo-Moreno, J. & Motta, M. Quantum-centric computation of molecular excited states with extended sample-based quantum diagonalization.Quantum Sci. Technol. 10, 025034 (2025)

    work page 2025

  29. [29]

    Yu, J.et al.Quantum-centric algorithm for sample-based Krylov diagonalization.arXiv preprint arXiv:2501.09702(2025)

    work page arXiv 2025

  30. [30]

    Piccinelli, S.et al.Quantum chemistry with provable convergence via randomized sample- based quantum diagonalization.arXiv preprint arXiv:2508.02578(2025)

    work page arXiv 2025

  31. [31]

    J., Motta, M

    Danilov, D., Robledo-Moreno, J., Sung, K. J., Motta, M. & Shee, J. Enhancing the accuracy and efficiency of sample-based quantum diagonalization with phaseless auxiliary-field quantum Monte Carlo.J. Chem. Theory Comput.21, 11585–11594 (2025)

    work page 2025

  32. [32]

    Shajan, A.et al.Toward quantum-centric simulations of extended molecules: Sample-based quantum diagonalization enhanced with density matrix embedding theory.J. Chem. Theory Comput.21, 6801–6810 (2025)

    work page 2025

  33. [33]

    Patra, C.et al.Physics-informed generative machine learning for accelerated quantum-centric supercomputing.arXiv preprint arXiv:2512.06858(2025)

    work page arXiv 2025

  34. [34]

    & Merz, K

    Kaliakin, D., Shajan, A., Liang, F. & Merz, K. M., Jr. Implicit solvent sample-based quantum diagonalization.J. Phys. Chem. B129, 5788–5796 (2025)

    work page 2025

  35. [35]

    Kaliakin, D.et al.Accurate quantum-centric simulations of intermolecular interactions. Commun. Phys.8, 396 (2025)

    work page 2025

  36. [36]

    M., Lee, S.-C

    Hossain, S. M., Lee, S.-C. & Bhattacharjee, S. Quantum simulations of battery electrolytes using variational quantum eigensolver, equation-of-motion, and sample-based diagonalization methods: Active-space design, dissociation, and excited states ofLiP F 6,N aP F 6, and FSI salts.Adv. Quantum Technol.9, e00871 (2026)

    work page 2026

  37. [37]

    & Chan, G

    Sharma, S., Sivalingam, K., Neese, F. & Chan, G. K.-L. Low-energy spectrum of iron–sulfur clusters directly from many-particle quantum mechanics.Nat. Chem.6, 927–933 (2014)

    work page 2014

  38. [38]

    & Wigner, E

    Jordan, P. & Wigner, E. ¨Uber das paulische ¨ aquivalenzverbot.Z. Phys.47, 631–651 (1928)

    work page 1928

  39. [39]

    Extensions to the k-means algorithm for clustering large data sets with categorical values.Data Min

    Huang, Z. Extensions to the k-means algorithm for clustering large data sets with categorical values.Data Min. Knowl. Discov.2, 283–304 (1998)

    work page 1998

  40. [40]

    de Vos, N. J. kmodes categorical clustering library. https://github.com/nicodv/kmodes (2022)

    work page 2022

  41. [41]

    & Govaert, G

    Nadif, M. & Govaert, G. Clustering for binary data and mixture models—choice of the model. Appl. Stoch. Models Data Anal.13, 269–278 (1997)

    work page 1997

  42. [42]

    Morin, S.et al.StepMix: A Python package for pseudo-likelihood estimation of generalized mixture models with external variables.J. Stat. Softw.113, 1–39 (2025)

    work page 2025

  43. [43]

    Perdew, J. P. Density-functional approximation for the correlation energy of the inhomoge- neous electron gas.Phys. Rev. B33, 8822 (1986). 22

    work page 1986

  44. [44]

    Becke, A. D. Density-functional exchange-energy approximation with correct asymptotic behavior.Phys. Rev. A38, 3098 (1988)

    work page 1988

  45. [45]

    & Machado, S

    Jorge, F., Canal Neto, A., Camiletti, G. & Machado, S. Contracted Gaussian basis sets for Douglas–Kroll–Hess calculations: Estimating scalar relativistic effects of some atomic and molecular properties.J. Chem. Phys.130, 064108 (2009)

    work page 2009

  46. [46]

    Ideas of relativistic quantum chemistry.Mol

    Liu, W. Ideas of relativistic quantum chemistry.Mol. Phys.108, 1679–1706 (2010)

    work page 2010

  47. [47]

    & Liu, W

    Li, Z., Xiao, Y. & Liu, W. On the spin separation of algebraic two-component relativistic hamiltonians.J. Chem. Phys.137, 154114 (2012)

    work page 2012

  48. [48]

    jrm874/sqd data repository: 0.0.1

    jrm874. jrm874/sqd data repository: 0.0.1. https://doi.org/10.5281/zenodo.15324153 (2025)

    work page doi:10.5281/zenodo.15324153 2025

  49. [49]

    White, S. R. Density matrix formulation for quantum renormalization groups.Phys. Rev. Lett.69, 2863 (1992)

    work page 1992

  50. [50]

    Sun, Q.et al.Recent developments in the PySCF program package.J. Chem. Phys.153, 024109 (2020)

    work page 2020

  51. [51]

    Sun, Q.et al.PySCF: the Python-based simulations of chemistry framework.WIREs Comput. Mol. Sci.8, e1340 (2018)

    work page 2018

  52. [52]

    J., Whaley, K

    Motta, M., Sung, K. J., Whaley, K. B., Head-Gordon, M. & Shee, J. Bridging physical intuition and hardware efficiency for correlated electronic states: the local unitary cluster Jastrow ansatz for electronic structure.Chem. Sci.14, 11213–11227 (2023)

    work page 2023

  53. [53]

    & Kurashige, Y

    Matsuzawa, Y. & Kurashige, Y. Jastrow-type decomposition in quantum chemistry for low- depth quantum circuits.J. Chem. Theory Comput.16, 944–952 (2020)

    work page 2020

  54. [54]

    ffsim: Faster simulations of fermionic quantum circuits

    The ffsim developers. ffsim: Faster simulations of fermionic quantum circuits. https://github. com/qiskit-community/ffsim (2025)

    work page 2025

  55. [55]

    & Lloyd, S

    Viola, L. & Lloyd, S. Dynamical suppression of decoherence in two-state quantum systems. Phys. Rev. A58, 2733 (1998)

    work page 1998

  56. [56]

    Van Den Berg, E., Minev, Z. K. & Temme, K. Model-free readout-error mitigation for quantum expectation values.Phys. Rev. A105, 032620 (2022)

    work page 2022

  57. [57]

    A.et al.Qiskit addon: sample-based quantum diagonalization

    Saki, A. A.et al.Qiskit addon: sample-based quantum diagonalization. https://github.com/ Qiskit/qiskit-addon-sqd (2024)

    work page 2024

  58. [58]

    D., Kang, H., Sundaresan, N

    Nation, P. D., Kang, H., Sundaresan, N. & Gambetta, J. M. Scalable mitigation of measurement errors on quantum computers.PRX Quantum2, 040326 (2021)

    work page 2021

  59. [59]

    Javadi-Abhari, A.et al.Quantum computing with Qiskit.arXiv preprint arXiv:2405.08810 (2024). 23

    work page internal anchor Pith review Pith/arXiv arXiv 2024