pith. sign in

arxiv: 2508.17688 · v1 · submitted 2025-08-25 · 🪐 quant-ph · cond-mat.stat-mech

A Unsupervised Framework for Identifying Diverse Quantum Phase Transitions Using Classical Shadow Tomography

Chi-Ting Ho , Daw-Wei Wang This is my paper

Pith reviewed 2026-05-18 21:57 UTC · model grok-4.3

classification 🪐 quant-ph cond-mat.stat-mech
keywords quantum phase transitionsclassical shadow tomographyprincipal component analysisunsupervised learningsymmetry-breaking transitionstopological transitionsspin systemsquantum criticality
0
0 comments X

The pith

Unsupervised PCA on classical shadow data detects and classifies symmetry-breaking and topological quantum phase transitions without knowing the Hamiltonian.

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

The paper establishes a machine learning approach that combines classical shadow representations from random Pauli measurements with principal component analysis to identify different types of quantum phase transitions in spin systems. This method works by uncovering hidden statistical patterns and fluctuations in the sampled data that signal quantum criticality. A sympathetic reader would care because it offers a general, data-driven tool that does not require advance knowledge of the model's Hamiltonian or specific order parameters, making it useful for exploring unknown quantum phases. The approach is benchmarked on several one- and two-dimensional models, showing it can distinguish symmetry-breaking from topological transitions based on their unique fluctuation signatures.

Core claim

Integrating classical shadow representations with unsupervised principal component analysis enables the reliable detection and distinction of both symmetry-breaking and topological quantum phase transitions, along with their qualitative classification based on characteristic fluctuation patterns in the data, without requiring any knowledge of the Hamiltonian or explicit order parameters.

What carries the argument

Classical shadow tomography from random Pauli measurements processed by unsupervised principal component analysis to extract fluctuation patterns.

If this is right

  • This approach can probe new quantum phases in systems where order parameters are unknown.
  • It distinguishes symmetry-breaking transitions from topological ones via distinct data fluctuation patterns.
  • The method applies to both one- and two-dimensional spin-1/2 models in a model-independent manner.
  • It serves as a general tool for identifying quantum criticality across diverse Hamiltonians.

Where Pith is reading between the lines

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

  • The characteristic fluctuation patterns might correspond to underlying physical mechanisms worth mapping to specific observables.
  • This unsupervised framework could be tested on systems with mixed transition types or in three dimensions.
  • Combining the PCA output with other measurement bases might sharpen the distinction between transition classes.

Load-bearing premise

The statistical patterns and fluctuations in the classical shadow data sampled from random Pauli measurements are sufficient to capture the distinct signatures of quantum criticality for both symmetry-breaking and topological transitions without any model-specific input or knowledge of the Hamiltonian.

What would settle it

If PCA on classical shadow data from the 2D transverse-field Ising model shows no clear principal-component shift at the known critical point, or if fluctuation patterns fail to separate symmetry-breaking from topological transitions in the tested models.

Figures

Figures reproduced from arXiv: 2508.17688 by Chi-Ting Ho, Daw-Wei Wang.

Figure 1
Figure 1. Figure 1: FIG. 1. PCA results for the 1D transverse-field Ising model [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. PCA results for the 1D XZX cluster-Ising model [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. PCA results for the 1D bond-alternating XXZ chain [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. (a) Illustration of the 1D indexing scheme for a 2D [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. (a) The ternary phase diagram of 2D Kitaev hon [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. PCA covariance matrices for the 1D XZX cluster [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
read the original abstract

We provide a general machine learning methodology that integrates classical shadow representations with unsupervised principal component analysis (PCA) to explore various quantum phase transitions. By sampling spin configurations from random Pauli measurements, our approach can effectively analyze hidden statistical patterns in the data, thereby capturing the distinct signatures of quantum criticality through their fluctuations. We benchmark this approach across various spin-1/2 systems, including the 1D XZX cluster-Ising model, the 1D bond-alternating XXZ model, the 2D transverse-field Ising model, and the 2D Kitaev honeycomb model. We show that PCA not only reliably detects and distinguishes both symmetry-breaking and topological transitions, but also enables their qualitative classification based on characteristic fluctuation patterns. Our data-driven approach does not require any knowledge of the Hamiltonian or explicit order parameters, and can therefore be a general and applicable tool for probing new quantum phases.

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 / 2 minor

Summary. The manuscript proposes an unsupervised framework integrating classical shadow tomography with principal component analysis (PCA) to detect and classify quantum phase transitions. By sampling from random Pauli measurements and analyzing statistical fluctuations in the shadow data, the method claims to identify signatures of both symmetry-breaking and topological transitions across four benchmarked spin-1/2 models (1D XZX cluster-Ising, 1D bond-alternating XXZ, 2D transverse-field Ising, and 2D Kitaev honeycomb) without requiring Hamiltonian knowledge or explicit order parameters. The central result is that PCA reliably detects the transitions and enables qualitative classification via characteristic fluctuation patterns in the data.

Significance. If the central claims hold with added quantitative support, the work could provide a practical, model-agnostic tool for probing quantum criticality in systems lacking known order parameters. The use of classical shadows for efficient data acquisition combined with unsupervised PCA on multiple distinct models is a clear strength, offering a data-driven alternative to traditional diagnostics. The multi-model benchmarking demonstrates broad applicability within the tested class of systems.

major comments (2)
  1. [§4 (Benchmark Results)] The claim that PCA enables reliable qualitative classification of symmetry-breaking versus topological transitions (abstract and benchmarking results) rests on visual or descriptive identification of 'characteristic fluctuation patterns' in the leading principal components. No quantitative metric—such as inter-class distance between eigenvalue spectra, silhouette score for clustering, or cross-validation separability—is reported to establish that these patterns are distinct and reproducible rather than model-specific artifacts of the chosen Pauli sampling.
  2. [§3 (Methodology)] The unsupervised and model-independent nature of the classification is asserted, yet the interpretation of fluctuation patterns as diagnostic of transition type appears to rely on post-hoc comparison with the known physics of the four benchmark Hamiltonians. This leaves open whether the same signatures would emerge for an unseen model without prior labeling of the transition classes.
minor comments (2)
  1. Figure captions and axis labels in the PCA projection plots would benefit from explicit mention of the number of shadow snapshots and the variance explained by the displayed components to aid reproducibility.
  2. [Abstract] The abstract states 'reliable detection' but does not quantify success rates or false-positive rates across the models; adding a brief summary table of detection thresholds would improve clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. We address the major comments point by point below and have revised the manuscript accordingly to improve the quantitative support and clarify the unsupervised aspects.

read point-by-point responses

  1. Referee: [§4 (Benchmark Results)] The claim that PCA enables reliable qualitative classification of symmetry-breaking versus topological transitions (abstract and benchmarking results) rests on visual or descriptive identification of 'characteristic fluctuation patterns' in the leading principal components. No quantitative metric—such as inter-class distance between eigenvalue spectra, silhouette score for clustering, or cross-validation separability—is reported to establish that these patterns are distinct and reproducible rather than model-specific artifacts of the chosen Pauli sampling.

    Authors: We thank the referee for pointing this out. In the revised manuscript, we have incorporated quantitative metrics to support the classification. Specifically, we now report the silhouette scores for the clustering of the principal component fluctuations for symmetry-breaking and topological transitions across the models. Additionally, we include the inter-class distances in the space of the leading eigenvalues to demonstrate the separability of the patterns. These additions provide a more rigorous basis for the qualitative classification claim. revision: yes

  2. Referee: [§3 (Methodology)] The unsupervised and model-independent nature of the classification is asserted, yet the interpretation of fluctuation patterns as diagnostic of transition type appears to rely on post-hoc comparison with the known physics of the four benchmark Hamiltonians. This leaves open whether the same signatures would emerge for an unseen model without prior labeling of the transition classes.

    Authors: We clarify that the detection of phase transitions is entirely unsupervised and model-independent, as it relies solely on the statistical fluctuations in the shadow data without any Hamiltonian input. For the classification into symmetry-breaking or topological types, while we use the benchmark models to illustrate the distinct patterns, the patterns themselves (such as the specific structure in the principal components) are derived from the data. In the revision, we have added a section discussing how these patterns can be applied to new models by matching to the characteristic signatures observed in the benchmarks, without requiring prior knowledge of the new model's physics. We acknowledge that a complete validation on novel unseen models would benefit from additional numerical experiments, which we plan to pursue in future work. revision: partial

Circularity Check

0 steps flagged

No significant circularity; data-driven unsupervised PCA analysis is self-contained

full rationale

The paper presents a machine-learning methodology that applies PCA directly to classical shadow snapshots sampled from random Pauli measurements on spin systems. The detection of phase transitions and their qualitative classification into symmetry-breaking versus topological types emerges from post-hoc inspection of the leading principal components and fluctuation patterns in the data for four specific benchmark Hamiltonians. No equations or steps reduce a claimed prediction to a fitted parameter by construction, no self-definitional loops exist, and the central claims do not rest on load-bearing self-citations that presuppose the observed patterns. The method is explicitly model-independent and unsupervised, with benchmarks serving as empirical validation rather than circular input-output equivalence. The derivation chain is therefore independent of the target results.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that classical shadow statistics contain sufficient fluctuation information to distinguish phase transitions; this is a domain assumption rather than a derived result.

axioms (1)
  • domain assumption Classical shadow representations from random Pauli measurements encode the necessary statistical fluctuations to reveal quantum criticality signatures.
    Invoked in the methodology description to justify using shadows as input to PCA.

pith-pipeline@v0.9.0 · 5682 in / 1172 out tokens · 26520 ms · 2026-05-18T21:57:05.579667+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

45 extracted references · 45 canonical work pages

  1. [1]

    Savary and L

    L. Savary and L. Balents, Quantum spin liquids: a re- view, Rep. Prog. Phys 80, 016502 (2016)

    work page 2016

  2. [2]

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

    work page 2017

  3. [3]

    B. A. Bernevig, Topological Insulators and Topological Superconductors (Princeton University Press, Princeton, 2013)

    work page 2013

  4. [4]

    Vermersch, A

    B. Vermersch, A. Elben, L. M. Sieberer, N. Y. Yao, and P. Zoller, Probing scrambling using statistical correla- tions between randomized measurements, Phys. Rev. X 9, 021061 (2019)

    work page 2019

  5. [5]

    T. B. et al., Probing r´ enyi entanglement entropy via ran- domized measurements, Science 6, eaaz3666 (2020)

    work page 2020

  6. [6]

    A. E. et al., Many-body topological invariants from ran- domized measurements in synthetic quantum matter, Sci. Adv. 6, eaaz3666 (2020)

    work page 2020

  7. [7]

    Aaronson, Shadow tomography of quantum states, in Proceedings of the 50th annual ACM SIGACT sympo- sium on theory of computing (2018) pp

    S. Aaronson, Shadow tomography of quantum states, in Proceedings of the 50th annual ACM SIGACT sympo- sium on theory of computing (2018) pp. 325–338

    work page 2018

  8. [8]

    Huang, R

    H.-Y. Huang, R. Kueng, and J. Preskill, Predicting many properties of a quantum system from very few measure- ments, Nat. Phys. 16, 1050–1057 (2020)

    work page 2020

  9. [9]

    Carrasquilla and R

    J. Carrasquilla and R. G. Melko, Machine learning phases of matter, Nat. Phys. 13, 431 (2017)

    work page 2017

  10. [10]

    Zhang, J

    W. Zhang, J. Liu, and T.-C. Wei, Machine learning of phase transitions in the percolation and xy models, Phys. Rev. E 99, 032142 (2019)

    work page 2019

  11. [11]

    Giannetti, B

    C. Giannetti, B. Lucini, and D. Vadacchino, Machine learning as a universal tool for quantitative investigations of phase transitions, Nucl. Phys. B 944, 114639 (2019)

    work page 2019

  12. [12]

    Shiina, H

    K. Shiina, H. Mori, Y. Okabe, and H. K. Lee, Machine- learning studies on spin models, Sci Rep 10, 2177 (2020)

    work page 2020

  13. [13]

    E. P. L. van Nieuwenburg, Y.-H. Liu, and S. D. Huber, Learning phase transitions by confusion, Nat. Phys. 13, 435 (2017)

    work page 2017

  14. [14]

    Ch’ng, J

    K. Ch’ng, J. Carrasquilla, R. G. Melko, and E. Khatami, Machine learning phases of strongly correlated fermions, Phys. Rev. X 7, 031038 (2017)

    work page 2017

  15. [15]

    B. S. Rem, N. K¨ aming, M. Tarnowski, L. Asteria, N. Fl¨ aschner, C. Becker, K. Sengstock, and C. Weiten- berg, Identifying quantum phase transitions using arti- ficial neural networks on experimental data, Nat. Phys. 15, 917 (2019)

    work page 2019

  16. [16]

    X.-Y. Dong, F. Pollmann, and X.-F. Zhang, Machine learning of quantum phase transitions, Phys. Rev. B 99, 121104(R) (2019)

    work page 2019

  17. [17]

    Wang, Discovering phase transitions with unsuper- vised learning, Phys

    L. Wang, Discovering phase transitions with unsuper- vised learning, Phys. Rev. B 94, 195105 (2016)

    work page 2016

  18. [18]

    S. J. Wetzel, Unsupervised learning of phase transitions: From principal component analysis to variational autoen- coders, Phys. Rev. E 96, 022140 (2017)

    work page 2017

  19. [19]

    Ch’ng, N

    K. Ch’ng, N. Vazquez, and E. Khatami, Unsupervised machine learning account of magnetic transitions in the hubbard model, Phys. Rev. E 97, 013306 (2018)

    work page 2018

  20. [20]

    Liu and E

    Y.-H. Liu and E. P. L. van Nieuwenburg, Discrimina- tive cooperative networks for detecting phase transitions, Phys. Rev. Lett. 120, 176401 (2018). 9

    work page 2018

  21. [21]

    Sch¨ afer and N

    F. Sch¨ afer and N. L¨ orch, Vector field divergence of pre- dictive model output as indication of phase transitions, Phys. Rev. E 99, 062107 (2019)

    work page 2019

  22. [22]

    Canabarro, F

    A. Canabarro, F. F. Fanchini, A. L. Malvezzi, R. Pereira, and R. Chaves, Unveiling phase transitions with machine learning, Phys. Rev. B 100, 045129 (2019)

    work page 2019

  23. [23]

    J. F. Rodriguez-Nieva and M. S. Scheurer, Identifying topological order through unsupervised machine learn- ing, Nat. Phys. 15, 790–795 (2019)

    work page 2019

  24. [24]

    Walker, K.-M

    N. Walker, K.-M. Tam, and M. Jarrell, Deep learning on the 2-dimensional ising model to extract the crossover region with a variational autoencoder, Sci Rep 10, 13047 (2020)

    work page 2020

  25. [25]

    M. S. Scheurer and R.-J. Slager, Unsupervised machine learning and band topology, Phys. Rev. Lett.124, 226401 (2020)

    work page 2020

  26. [26]

    J. Wang, W. Zhang, T. Hua, and T.-C. Wei, Unsuper- vised learning of topological phase transitions using the calinski-harabaz index, Phys. Rev. Res.3, 013074 (2021)

    work page 2021

  27. [27]

    Mendes-Santos, A

    T. Mendes-Santos, A. Angelone, A. Rodriguez, R. Fazio, and M. Dalmonte, Intrinsic dimension of path integrals: Data-mining quantum criticality and emergent simplic- ity, PRX Quantum 2, 030332 (2021)

    work page 2021

  28. [28]

    K¨ aming, A

    N. K¨ aming, A. Dawid, K. Kottmann, M. Lewenstein, K. Sengstock, A. Dauphin, and C. Weitenberg, Unsuper- vised machine learning of topological phase transitions from experimental data, Mach. Learn.: Sci. Technol. 2, 035037 (2021)

    work page 2021

  29. [29]

    Tirelli, D

    A. Tirelli, D. O. Carvalho, L. A. Oliveira, J. P. de Lima, N. C. Costa, and R. R. dos Santos, Unsupervised ma- chine learning approaches to the q-state potts model, Eur. Phys. J. B 95, 013306 (2022)

    work page 2022

  30. [30]

    Ng and M.-F

    K.-K. Ng and M.-F. Yang, Unsupervised learning of phase transitions via modified anomaly detection with autoencoders, Phys. Rev. B 108, 214428 (2023)

    work page 2023

  31. [31]

    Sadoune, G

    N. Sadoune, G. Giudici, K. Liu, and L. Pollet, Unsuper- vised interpretable learning of phases from many-qubit systems, Phys. Rev. Res. 5, 013082 (2023)

    work page 2023

  32. [32]

    Mendes-Santos, X

    T. Mendes-Santos, X. Turkeshi, M. Dalmonte, and A. Ro- driguez, Unsupervised learning universal critical behav- ior via the intrinsic dimension, Phys. Rev. X 11, 011040 (2021)

    work page 2021

  33. [33]

    Ho and D.-W

    C.-T. Ho and D.-W. Wang, Robust identification of topo- logical phase transition by self-supervised machine learn- ing approach, New J. Phys. 23, 083021 (2021)

    work page 2021

  34. [34]

    Ho and D.-W

    C.-T. Ho and D.-W. Wang, Self-supervised ensemble learning: A universal method for phase transition classi- fication of many-body systems, Phys. Rev. Res.5, 043090 (2023)

    work page 2023

  35. [35]

    K. P. F.R.S., Liii. on lines and planes of closest fit to systems of points in space, The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science 2, 559 (1901)

    work page 1901

  36. [36]

    Tantivasadakarn, R

    N. Tantivasadakarn, R. Thorngren, A. Vishwanath, and R. Verresen, Pivot hamiltonians as generators of symme- try and entanglement, Sci. Rep. 14, 012 (2023)

    work page 2023

  37. [37]

    W. Choi, M. Knap, and F. Pollmann, Finite- temperature entanglement negativity of fermionic symmetry-protected topological phases and quantum critical points in one dimension, Phys. Rev. B 109, 115132 (2024)

    work page 2024

  38. [38]

    Yu-Chin, D

    T. Yu-Chin, D. Li, C. Ming-Chiang, L. Amico, and L.-C. Kwek, Entanglement convertibility by sweeping through the quantum phases of the alternating bonds xxz chain, Sci. Rep. 14, 012 (2023)

    work page 2023

  39. [39]

    Kitaev, Anyons in an exactly solved model and be- yond, Ann

    A. Kitaev, Anyons in an exactly solved model and be- yond, Ann. Phys 321, 2 (2006)

    work page 2006

  40. [40]

    Sachdev, Quantum phase transitions, Physics world 12, 33 (1999)

    S. Sachdev, Quantum phase transitions, Physics world 12, 33 (1999)

    work page 1999

  41. [41]

    Giamarchi, Quantum physics in one dimension , Vol

    T. Giamarchi, Quantum physics in one dimension , Vol. 121 (Clarendon press, 2003)

    work page 2003

  42. [42]

    W. P. Su, J. R. Schrieffer, and A. J. Heeger, Solitons in polyacetylene, Phys. Rev. Lett. 42, 1698 (1979)

    work page 1979

  43. [43]

    Friedman, Ising model with a transverse field in two dimensions: Phase diagram and critical properties from a real-space renormalization group, Phys

    Z. Friedman, Ising model with a transverse field in two dimensions: Phase diagram and critical properties from a real-space renormalization group, Phys. Rev. B 17, 1429 (1978)

    work page 1978

  44. [44]

    Fishman, S

    M. Fishman, S. R. White, and E. M. Stoudenmire, The itensor software library for tensor network calculations, SciPost Phys. Codebases , 4 (2022)

    work page 2022

  45. [45]

    Fishman, S

    M. Fishman, S. R. White, and E. M. Stoudenmire, Code- base release 0.3 for itensor, SciPost Phys. Codebases , 4 (2022)

    work page 2022