pith. sign in

arxiv: 2605.23300 · v1 · pith:HAKGJ6T4new · submitted 2026-05-22 · 🪐 quant-ph

Local-Observable-Guided Generative Quantum Circuits for Degenerate Ground Spaces

Yiying Chen , Lingxia Zhang , Yanzheng Zhu , Kaiyan Yang , Xiao Zeng , Zizhu Wang This is my paper

Pith reviewed 2026-05-25 04:37 UTC · model grok-4.3

classification 🪐 quant-ph
keywords degenerate ground statesgenerative quantum circuitsparameterized quantum circuitslocal observablesMajumdar-Ghosh modelAKLT modelXXZ chainenergy-diversity objective
0
0 comments X

The pith

A hybrid generative quantum circuit with local observable penalties produces an ensemble whose span reproduces the full degenerate ground space.

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

The paper tries to recover the entire degenerate ground space of a quantum many-body system instead of approximating one ground state at a time. It trains a classical generative model to sample parameters for a parameterized quantum circuit and optimizes a joint objective that minimizes energy while adding cosine-similarity penalties computed from local observable correlators. These penalties encourage the sampled states to be distinct, allowing their linear span to cover the target degenerate subspace. The approach is demonstrated on three models that realize degeneracy through different physical mechanisms, and it continues to work when energies and correlators are estimated from finite shots.

Core claim

By learning a distribution over parameters of a parameterized quantum circuit and minimizing an energy-diversity objective whose diversity term uses cosine similarities of local observable correlators, the method generates an ensemble of states whose linear span accurately reproduces the target degenerate ground space on the Majumdar-Ghosh, AKLT, and spin-1 XXZ models, sometimes recovering an approximately orthogonal basis within the ensemble.

What carries the argument

The energy-diversity objective whose diversity term consists of cosine-similarity penalties derived from local observable correlators, which distinguishes distinct ground states using only scalable local measurements.

If this is right

  • The full degenerate ground space can be recovered from a modest number of sampled states rather than by solving for each state separately.
  • Local correlators provide a measurement-efficient substitute for full tomography when enforcing diversity among candidate ground states.
  • The same objective remains effective when all expectation values are replaced by finite-shot estimates.
  • The framework applies across distinct physical origins of degeneracy, including valence-bond-solid and Haldane phases.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same local-observable penalty structure could be reused to enforce diversity in other generative tasks that involve quantum states with known symmetries.
  • If the method scales to larger lattices, it could supply initial ensembles for variational algorithms that target topological order parameters.
  • The classical generative model over circuit parameters might be replaced by other samplers without changing the local-observable guidance mechanism.

Load-bearing premise

Cosine-similarity penalties computed from local observable correlators are enough to separate all distinct ground states without missing any component of the degenerate space or introducing systematic bias.

What would settle it

On a model whose exact ground-space dimension and basis are known, generate the ensemble, compute the dimension and overlap of its linear span with the exact space, and check whether the span matches the known degeneracy to within numerical tolerance.

Figures

Figures reproduced from arXiv: 2605.23300 by Kaiyan Yang, Lingxia Zhang, Xiao Zeng, Yanzheng Zhu, Yiying Chen, Zizhu Wang.

Figure 1
Figure 1. Figure 1: FIG. 1: Generative quantum circuit architecture. The [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2: Training dynamics for the MG model. Shown are [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3: Overlap distributions of accepted states [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4: Training dynamics for the symmetry-encoded AKLT [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5: Overlap distributions of accepted states [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6: Training dynamics for the symmetry-encoded XXZ [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 9
Figure 9. Figure 9: FIG. 9: Circuit ansatz for the spin-1 model, consisting of [PITH_FULL_IMAGE:figures/full_fig_p013_9.png] view at source ↗
Figure 8
Figure 8. Figure 8: FIG. 8: Structure of the circuit ansatz for spin- [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
Figure 10
Figure 10. Figure 10: FIG. 10: Overlaps of 10 representative generated states [PITH_FULL_IMAGE:figures/full_fig_p014_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: FIG. 11: Cosine-similarity matrices for the MG model. The upper row (a)-(d) shows the exact ground-state basis [PITH_FULL_IMAGE:figures/full_fig_p015_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: FIG. 12: Overlaps of 10 representative generated states [PITH_FULL_IMAGE:figures/full_fig_p015_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: FIG. 13: Cosine-similarity matrices for the AKLT model [PITH_FULL_IMAGE:figures/full_fig_p016_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: FIG. 14: Overlaps of 10 representative generated states [PITH_FULL_IMAGE:figures/full_fig_p016_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: FIG. 15: Cosine-similarity matrices for the spin-1 XXZ [PITH_FULL_IMAGE:figures/full_fig_p017_15.png] view at source ↗
read the original abstract

Searching for degenerate ground spaces in quantum many-body systems is central to understanding spontaneous symmetry breaking and topological order. Although existing numerical methods can approximate individual ground states with high accuracy, recovering the full degenerate space remains a substantial challenge. Here we tackle this problem using a hybrid generative quantum circuit that combines a classical generative model with an expressive parameterized quantum circuit (PQC). The classical model learns a distribution over PQC parameters, enabling the sampling of an ensemble of ground states, while the PQC ensures compatibility with quantum hardware. To promote both low energy and state diversity, we define an energy-diversity objective composed of an energy-minimization term and cosine-similarity penalties derived from local observable correlators. These local descriptors provide a scalable, measurement-efficient means of distinguishing distinct ground states. We benchmark the framework on the Majumdar-Ghosh model, the Affleck-Kennedy-Lieb-Tasaki model, and the spin-1 XXZ chain, which realize distinct mechanisms of degeneracy. In all cases, the method produces a diverse ensemble whose linear span accurately reproduces the target ground space, in some instances, it identifies an approximately orthogonal basis within the learned ensemble. We further show that the framework remains robust under shot-based estimation and can still recover the degenerate ground space with a reduced measurement budget.

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

2 major / 1 minor

Summary. The paper proposes a hybrid generative framework combining a classical generative model with a parameterized quantum circuit (PQC) to sample ensembles of states approximating degenerate ground spaces. An energy-diversity objective is defined consisting of an energy-minimization term plus cosine-similarity penalties based on local observable correlators to encourage both low energy and diversity among samples. The approach is benchmarked on the Majumdar-Ghosh model, the AKLT model, and the spin-1 XXZ chain; the authors report that the linear span of the learned ensemble reproduces the target ground space in each case, sometimes yielding an approximately orthogonal basis, and that the method remains effective under shot-noise-limited measurements.

Significance. If validated, the framework provides a hardware-compatible route to exploring degenerate ground spaces relevant to spontaneous symmetry breaking and topological order. The emphasis on local observables for diversity is measurement-efficient, and the three benchmarks cover distinct degeneracy mechanisms. Credit is due for the explicit demonstration of robustness to finite-shot estimation. The central limitation is that success is shown only for models in which local correlators distinguish the degenerate states; broader applicability remains to be established.

major comments (2)
  1. [Abstract and benchmark section] Abstract and benchmark results: the central claim that 'the linear span accurately reproduces the target ground space' is stated without accompanying quantitative metrics (e.g., subspace overlap, average fidelity to the target projector, or comparison against exact diagonalization baselines). This absence prevents assessment of how completely or accurately the reproduction holds.
  2. [Energy-diversity objective] Energy-diversity objective (Methods): the cosine-similarity penalties rely on local observable correlators to enforce diversity. When degeneracy arises from non-local or topological features that leave all local expectations identical across ground states, the penalty term vanishes identically and the optimization can converge to a proper subspace. The three benchmark models (Majumdar-Ghosh, AKLT, spin-1 XXZ) all admit local distinction, but the manuscript provides neither a general proof that the chosen descriptors are complete nor a counter-example test on a topologically degenerate system.
minor comments (1)
  1. [Abstract] The statement that the method 'identifies an approximately orthogonal basis within the learned ensemble' is not accompanied by the specific model, the numerical criterion for orthogonality, or the relevant figure/table reference.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive feedback and positive assessment of the work's significance. We address the major comments point by point below, outlining planned revisions where appropriate.

read point-by-point responses

  1. Referee: [Abstract and benchmark section] Abstract and benchmark results: the central claim that 'the linear span accurately reproduces the target ground space' is stated without accompanying quantitative metrics (e.g., subspace overlap, average fidelity to the target projector, or comparison against exact diagonalization baselines). This absence prevents assessment of how completely or accurately the reproduction holds.

    Authors: We agree that quantitative metrics are needed to rigorously support the central claim. In the revised manuscript we will add explicit measures including subspace overlap with the target ground space, average fidelity to the target projector, and direct comparisons against exact diagonalization baselines for each of the three benchmark models. These will appear in the benchmark section and be referenced in the abstract. revision: yes

  2. Referee: [Energy-diversity objective] Energy-diversity objective (Methods): the cosine-similarity penalties rely on local observable correlators to enforce diversity. When degeneracy arises from non-local or topological features that leave all local expectations identical across ground states, the penalty term vanishes identically and the optimization can converge to a proper subspace. The three benchmark models (Majumdar-Ghosh, AKLT, spin-1 XXZ) all admit local distinction, but the manuscript provides neither a general proof that the chosen descriptors are complete nor a counter-example test on a topologically degenerate system.

    Authors: We acknowledge the limitation. The framework is explicitly constructed around local observable correlators, so the diversity penalty is effective only when local features distinguish the ground states (as holds for the chosen benchmarks). We will revise the manuscript to include an explicit discussion of this scope, stating that the method applies when local observables suffice to differentiate degenerate states and noting that the penalty vanishes for purely topological cases with identical local expectations. We will clarify that no general proof of descriptor completeness is provided because completeness is model-dependent. We cannot add a counter-example test on a topologically degenerate system in the current revision. revision: partial

standing simulated objections not resolved
  • Counter-example test on a topologically degenerate system where local expectations are identical across ground states

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper defines an energy-diversity objective using external energy minimization and cosine-similarity penalties on local observable correlators (independent measurements). The central claim that the ensemble's linear span reproduces the target ground space is presented as an empirical outcome of optimization on benchmarks (Majumdar-Ghosh, AKLT, XXZ), not as a quantity defined by or fitted to the method's own outputs. No self-citations are invoked as load-bearing uniqueness theorems, no parameters are fitted then renamed as predictions, and no ansatz or renaming reduces the result to its inputs by construction. The framework is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Review based on abstract only; full details on parameters, assumptions, and evidence unavailable. Local observables are treated as distinguishing without further justification in the provided text.

axioms (1)
  • domain assumption Local observable correlators suffice to distinguish distinct states in the degenerate space
    Invoked to justify the cosine-similarity penalties as a diversity mechanism.

pith-pipeline@v0.9.0 · 5776 in / 1053 out tokens · 21061 ms · 2026-05-25T04:37:12.435794+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

58 extracted references · 58 canonical work pages · 3 internal anchors

  1. [1]

    For both system sizes, the classical generator uses encoder and decoder net- works with four hidden layers, and the latent dimension is fixed at dz =50

    Energy-diversity objective and training dynamics We benchmark the proposed generative quantum circuit on the Majumdar-Ghosh (MG) chain with OBC at system sizes N =9 and N =10, whose exact ground-state energies are E0 = −21 and E0 = −24, respectively. For both system sizes, the classical generator uses encoder and decoder net- works with four hidden layers...

    work page

  2. [2]

    Characterization of the learned ground space After satisfying the convergence criteria, we terminate training and assess the learned ensemble by sampling 1500 latent variables for each system size. A generated state is accepted if it simultaneously satisfies a relative energy constraint ∆E = |E−E 0| |E0| < 0.005 and a ground-space over- lap score pG(ψ) = ...

    work page

  3. [3]

    This benchmark is car- ried out on the MG chain at small system sizesN =5 , 6, with expectation-value simulations at the same sizes serving as a noiseless reference

    Robustness to finite-shot sampling We further assess the robustness of the generative quan- tum circuits in a finite-shot setting. This benchmark is car- ried out on the MG chain at small system sizesN =5 , 6, with expectation-value simulations at the same sizes serving as a noiseless reference. The finite-shot and expectation-value studies use the same c...

    work page

  4. [4]

    The variational circuit is then restricted to symmetry-preserving layers acting within the encoded space

    Energy-diversity objective and training dynamics To implement the AKLT model on variational qubit cir- cuits, we adopt the symmetry-encoding scheme introduced in Ref.[ 31], in which each spin-1 degree of freedom is em- bedded into a two-qubit symmetric subspace by a fixed lo- cal isometry. The variational circuit is then restricted to symmetry-preserving ...

    work page

  5. [5]

    Characterization of the learned ground space After training, we sample 1500 latent variables and de- code them into circuit parameters to evaluate the learned ensemble. Among these generated states, 82.5% satisfy both a relative energy error criterion ∆E = |E−(−2.5298)| |−2.5298| < 0.02 and a ground-space overlap score pG(ψ) = Pr i=1 |〈g i|ψ〉|> 0.995. Thi...

    work page

  6. [6]

    In the present setting, the N ′ =4 spin-1 system is encoded into an N =8 qubit cir- cuit, with exact ground-state energy E0 = −3

    Energy-diversity objective and training dynamics As in the AKLT case, we adopt a symmetry-encoding scheme to represent the spin-1 XXZ chain on a variational qubit circuit (see Appendix A 2). In the present setting, the N ′ =4 spin-1 system is encoded into an N =8 qubit cir- cuit, with exact ground-state energy E0 = −3. The classical generator uses a six-l...

    work page

  7. [7]

    Characterization of the learned ground space We decode and execute 1500 latent samples to evaluate the learned ensemble. Among these generated states, 92.7% satisfy the acceptance criteria, namely a relative energy er- ror ∆E = |E−(−3)| |−3| < 0.017, together with a ground-space overlap score pG(ψ) > 0.995. This high acceptance rate indicates that the gen...

    work page

  8. [8]

    Baumann, R

    K. Baumann, R. Mottl, F . Brennecke, and T . Esslinger, Explor- ing symmetry breaking at the dicke quantum phase transition, Phys. Rev. Lett.107, 140402 (2011)

    work page 2011

  9. [9]

    X. Chen, Y . Lu, and A. Vishwanath, Symmetry-protected topo- logical phases from decorated domain walls, Nat. Commun. 5, 3507 (2014)

    work page 2014

  10. [10]

    Levin and X

    M. Levin and X. Wen, Detecting topological order in a ground state wave function, Phys. Rev. Lett.96, 110405 (2006)

    work page 2006

  11. [11]

    J. W . F . Venderbos, M. Daghofer, J. van den Brink, and S. Ku- mar, Switchable quantum anomalous hall state in a strongly frustrated lattice magnet, Phys. Rev. Lett.109, 166405 (2012)

    work page 2012

  12. [12]

    Y . Su, B. Hu, S. Li, and S. Cho, Quantum fidelity for degenerate ground states in quantum phase transitions, Physical Review E88, 032110 (2013)

    work page 2013

  13. [13]

    Hamma, S

    A. Hamma, S. M. Giampaolo, and F . Illuminati, Mutual infor- mation and spontaneous symmetry breaking, Physical Review A93, 012303 (2016)

    work page 2016

  14. [14]

    Nussinov and G

    Z. Nussinov and G. Ortiz, Sufficient symmetry conditions for topological quantum order, Proceedings of the National Academy of Sciences of the United States of America106, 16944 (2009)

    work page 2009

  15. [15]

    Villainet al., Theory of frustrated systems, Journal of Physics C: Solid State Physics10, 171 (1977)

    J. Villainet al., Theory of frustrated systems, Journal of Physics C: Solid State Physics10, 171 (1977)

    work page 1977

  16. [16]

    Afflecket al., Valence-bond ground states in antiferromag- netic spin chains, Physical Review Letters59, 799 (1987)

    I. Afflecket al., Valence-bond ground states in antiferromag- netic spin chains, Physical Review Letters59, 799 (1987)

    work page 1987

  17. [17]

    Y . Zhou, K. Kanoda, and T .-K. Ng, Quantum spin liquid states, Rev. Mod. Phys.89, 025003 (2017). 12

    work page 2017

  18. [18]

    Wen, Choreographed entanglement dances: Topological states of quantum matter, Science363, eaal3099 (2019), https://www.science.org/doi/pdf/10.1126/science.aal3099

    X.-G. Wen, Choreographed entanglement dances: Topological states of quantum matter, Science363, eaal3099 (2019), https://www.science.org/doi/pdf/10.1126/science.aal3099

    work page doi:10.1126/science.aal3099 2019

  19. [19]

    A. M. Läuchli, Exact diagonalization: A smart tool to study new states of quantum matter, IPAM QS2009 Workshop, Jan- uary 27 2009, UCLA (2009)

    work page 2009

  20. [20]

    Jung and J

    J. Jung and J. D. Noh, Guide to exact diagonalization study of quantum thermalization, Journal of the Korean Physical Society76, 670 (2020)

    work page 2020

  21. [21]

    J. I. Cirac, D. Pérez-García, N. Schuch, and F . Verstraete, Ma- trix product states and projected entangled pair states: Con- cepts, symmetries, theorems, Rev. Mod. Phys.93, 045003 (2021)

    work page 2021

  22. [22]

    Schollwöck, The density-matrix renormalization group in the age of matrix product states, Annals of Physics326, 96 (2011), january 2011 Special Issue

    U. Schollwöck, The density-matrix renormalization group in the age of matrix product states, Annals of Physics326, 96 (2011), january 2011 Special Issue

    work page 2011

  23. [23]

    Y.-C. He, D. N. Sheng, and Y. Chen, Obtaining topological de- generate ground states by the density matrix renormalization group, Physical Review B89, 075110 (2014)

    work page 2014

  24. [24]

    Vaezi and A

    M.-S. Vaezi and A. Vaezi, Entanglement distance between quantum states and its implications for a density-matrix renor- malization group study of degenerate ground states, Physical Review B96, 165129 (2017)

    work page 2017

  25. [25]

    R. N. C. Pfeifer, Symmetry-protected local minima in infinite dmrg, Physical Review B92, 205127 (2015)

    work page 2015

  26. [26]

    Peruzzoet al., A variational eigenvalue solver on a photonic quantum processor, Nature Communications5, 4213 (2014)

    A. Peruzzoet al., A variational eigenvalue solver on a photonic quantum processor, Nature Communications5, 4213 (2014)

    work page 2014

  27. [27]

    Nakanishiet al., Subspace-search variational quantum eigensolver for excited states, Physical Review Research1, 033062 (2019)

    K. Nakanishiet al., Subspace-search variational quantum eigensolver for excited states, Physical Review Research1, 033062 (2019)

    work page 2019

  28. [28]

    J. R. McCleanet al., The phase estimation algorithm: Reassess- ing the standard quantum approach, Quantum Information & Computation16, 1 (2016)

    work page 2016

  29. [29]

    PennyLane: Automatic differentiation of hybrid quantum-classical computations

    V . Bergholmet al., Pennylane: Automatic differentiation of hybrid quantum-classical computations, arXiv preprint arXiv:1811.04968 (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  30. [30]

    Quantum computing with Qiskit

    A. Javadi-Abhariet al., Quantum computing with qiskit (2024), arXiv:2405.08810[quant-ph]

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  31. [31]

    Xuet al., Mindspore quantum: A user-friendly, high- performance, and ai-compatible quantum computing frame- work (2024), arXiv:2406.17248[quant-ph]

    X. Xuet al., Mindspore quantum: A user-friendly, high- performance, and ai-compatible quantum computing frame- work (2024), arXiv:2406.17248[quant-ph]

    work page arXiv 2024

  32. [32]

    Hanin, Which neural net architectures give rise to ex- ploding and vanishing gradients?, inAdvances in Neural In- formation Processing Systems 31 (NeurIPS 2018), edited by S

    B. Hanin, Which neural net architectures give rise to ex- ploding and vanishing gradients?, inAdvances in Neural In- formation Processing Systems 31 (NeurIPS 2018), edited by S. Bengio, H. M. Wallach, H. Larochelle, K. Grauman, N. Cesa- Bianchi, and R. Garnett (Curran Associates, Inc., 2018) pp. 580–589

    work page 2018

  33. [33]

    Grant, L

    E. Grant, L. Wossnig, M. Ostaszewski, and M. Benedetti, An initialization strategy for addressing barren plateaus in parametrized quantum circuits, Quantum3, 214 (2019)

    work page 2019

  34. [34]

    Zhang, X

    L. Zhang, X. Lin, P . Wang, K. Yang, X. Zeng, Z. Wei, and Z. Wang, Variational optimization for quantum problems us- ing deep generative networks, Communications Physics8, 334 (2025)

    work page 2025

  35. [35]

    Swingle, Reconstructing quantum states from local data, Physical Review A89, 034112 (2014)

    B. Swingle, Reconstructing quantum states from local data, Physical Review A89, 034112 (2014)

    work page 2014

  36. [36]

    Huang, R

    H.-Y . Huang, R. Kueng, and J. Preskill, Predicting many prop- erties of a quantum system from very few measurements, Nature Physics16, 1050 (2020)

    work page 2020

  37. [37]

    Cramer, M

    M. Cramer, M. B. Plenio, S. T . Flammia, R. Somma, D. Gross, S. D. Bartlett, O. Landon-Cardinal, D. Poulin, and Y.-K. Liu, Efficient quantum state tomography , Nature Communications 1, 149 (2010)

    work page 2010

  38. [38]

    K. Yang, Y. Zhu, X. Zeng, Z. Zou, M.-H. Yung, and Z. Wang, Cost of locally approximating high-dimensional ground states of contextual quantum models, Communications Physics8, 1 (2025)

    work page 2025

  39. [39]

    J. M. Kübler, A. Arrasmith, L. Cincio, and P . J. Coles, An adap- tive optimizer for measurement-frugal variational algorithms, Quantum4, 263 (2020)

    work page 2020

  40. [40]

    L. Zhu, S. Liang, C. Yang, and X. Li, Optimizing shot assign- ment in variational quantum eigensolver measurement, Jour- nal of Chemical Theory and Computation20, 2390 (2024)

    work page 2024

  41. [41]

    and Yueshu Gu, A. Lowe, P . A. Dub, P . J. Coles, and A. Arra- smith, Adaptive shot allocation for fast convergence in varia- tional quantum algorithms, ArXivabs/2108.10434(2021)

    work page arXiv 2021

  42. [42]

    L. V . Foster and T . A. Davis, Algorithm 933: Reliable calcu- lation of numerical rank, null space bases, pseudoinverse solutions, and basic solutions using suitesparseqr, ACM trans- actions on mathematical software40, 1 (2013)

    work page 2013

  43. [43]

    Kolodrubetz, V

    M. Kolodrubetz, V . Gritsev, and A. Polkovnikov, Classifying and measuring geometry of a quantum ground state manifold, Physical Review B88, 064304 (2013)

    work page 2013

  44. [44]

    Sherif, Simultaneous approximation of multiple degener- ate states using a single neural network quantum state, Ma- chine Learning: Science and Technology7, 015029 (2026)

    W . Sherif, Simultaneous approximation of multiple degener- ate states using a single neural network quantum state, Ma- chine Learning: Science and Technology7, 015029 (2026)

    work page 2026

  45. [45]

    Klambauer, T

    G. Klambauer, T . Unterthiner, A. Mayr, and S. Hochreiter, Self-Normalizing Neural Networks, inAdvances in Neural Information Processing Systems, Vol. 30 (Curran Associates, Inc., 2017)

    work page 2017

  46. [46]

    J. Chen, Z. Ji, D. Kribs, Z. Wei, and B. Zeng, Ground-state spaces of frustration-free hamiltonians, Journal of mathemat- ical physics53(2012)

    work page 2012

  47. [47]

    Li and R

    Z. Li and R. S. Mong, Detecting topological order from mod- ular transformations of ground states on the torus, Physical Review B106, 235115 (2022)

    work page 2022

  48. [48]

    Banchi and G

    L. Banchi and G. E. Crooks, Measuring Analytic Gradients of General Quantum Evolution with the Stochastic Parameter Shift Rule, Quantum5, 386 (2021), arXiv:2005.10299[quant- ph]

    work page arXiv 2021

  49. [49]

    W . J. Caspers, K. M. Emmett, and W . Magnus, The Majumdar- Ghosh chain. Twofold ground state and elementary excita- tions, Journal of Physics A: Mathematical and General17, 2687 (1984)

    work page 1984

  50. [50]

    R. W . Chhajlany , P . Tomczak, A. Wojcik, and J. Richter, Entan- glement in the Majumdar-Ghosh model, Physical Review A 75, 032340 (2007), arXiv:cond-mat/0610379

    work page internal anchor Pith review Pith/arXiv arXiv 2007

  51. [51]

    Affleck, T

    I. Affleck, T . Kennedy, E. H. Lieb,et al., Rigorous results on valence-bond ground states in antiferromagnets, Physical Review Letters59, 799 (1987)

    work page 1987

  52. [52]

    Zhanget al., Quantum spin-1/2 xxz model at the first-order transition point δ = −1, Physical Review B101, 134406 (2020)

    Y . Zhanget al., Quantum spin-1/2 xxz model at the first-order transition point δ = −1, Physical Review B101, 134406 (2020)

    work page 2020

  53. [53]

    Di and H.-R

    Y.-M. Di and H.-R. Wei, Synthesis of multivalued quantum logic circuits by elementary gates, Phys. Rev. A87, 012325 (2013). Appendix A: Ansatz Design for Degenerate Ground Spaces In this appendix, we present the parameterized quantum- circuit ansätze used to represent degenerate ground spaces in the models considered in the main text. Since the spin- 1 2 ...

    work page 2013

  54. [54]

    As shown in Fig

    Circuit ansatz for spin- 1 2 model To represent the degenerate ground space of the Majumdar-Ghosh (MG) model, we employ a variational ansatz built from sequential layers of nearest-neighbor two- qubit gates. As shown in Fig. 8, each circuit layer consists of a chain of universal two-qubit gate blocks, and each block is composed of 6 layers of single-qubit...

    work page

  55. [55]

    Circuit ansatz for spin-1 models For the spin-1 models, we employ a symmetry-based en- coding scheme in which the Hilbert space of N ′ spin-1 de- grees of freedom is embedded into the Hilbert space of N =2 N ′ qubits. Each spin-1 site is represented in the sym- metric subspace of a qubit pair, so that the encoded qubit circuit preserves the local structur...

    work page

  56. [56]

    MG model For the Majumdar-Ghosh (MG) chain with open boundary conditions, which is defined as HMG = PN−2 i=1 (Si ·Si+1 +S i+1 · Si+2 +Si ·Si+2), (Si are spin-1 2 operators), we consider system sizes N =9 and N =10, whose ground-space degeneracies arek=4 andk=5, respectively . Fig. 10 shows the overlaps of 10 representative generated states with the exact ...

    work page

  57. [57]

    Accordingly , the most informative observables are localized near the boundaries

    AKLT model For the symmetry-encoded AKLT chain at sys- tem size N =10, which is defined as HAKLT =PN−1 i=1 Si ·S i+1 + 1 3 (Si ·S i+1)2 (Si are spin-1 operators on site i), the fourfold ground-state degeneracy originates from the fractionalized edge spin- 1 2 degrees of freedom. Accordingly , the most informative observables are localized near the boundar...

    work page

  58. [58]

    Spin-1 XXZ model We consider the spin-1 XXZ chain at the first-order transition point ∆ = −1, which is defined as HXXZ =PN−1 i=1 S x i S x i+1 +S y i S y i+1 +∆S z i Sz i+1 (Sα i are spin-1 operators at site i and α = x,y,z ), where the ground space forms a ferromagnetic multiplet. In this appendix, we examine how representative generated states are distr...

    work page