pith. sign in

arxiv: 2604.13165 · v1 · submitted 2026-04-14 · 🪐 quant-ph

Third-Order Local Randomized Measurements for Finite-size Entanglement Certification

Giovanni Scala , Gniewomir Sarbicki This is my paper

Pith reviewed 2026-05-10 14:50 UTC · model grok-4.3

classification 🪐 quant-ph
keywords entanglement witnessrandomized measurementsreduction criterionthird-order invariantsseparable statesquantum certificationfinite-size systems
0
0 comments X

The pith

A 4x4 matrix from third-order local randomized measurements is positive semidefinite for every separable state, with negative minimum eigenvalue certifying entanglement.

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

The paper converts the reduction criterion into a form testable with local single-copy measurements by applying it to squared affine combinations of the identity, the two local marginals, and the full state. This produces a 4 by 4 matrix whose entries are second- and third-order local invariants. The authors prove that the matrix is positive semidefinite whenever the state is separable, so entanglement is witnessed by a negative smallest eigenvalue. The sign of that eigenvalue can be estimated from randomized measurements on individual copies, with sample complexity that stays constant as dimension grows. Benchmarks on isotropic states show the test detects entanglement for mixing parameters roughly twice as close to the separability boundary as second-order purity witnesses allow.

Core claim

We convert the reduction criterion into an experimentally measurable separability criterion by testing it on squared affine combinations of the identity, the local marginals, and the state itself. This yields a 4×4 matrix bar M(rho) built from experimentally accessible second- and third-order local invariants. Entanglement is certified when its minimum eigenvalue E4(rho) becomes negative. We prove that all separable states satisfy bar M(rho) succeq 0, and that the sign of E4(rho) can be inferred from single-copy randomized measurements with dimension-independent sample complexity.

What carries the argument

The 4x4 matrix bar M(rho) formed by applying the reduction criterion to squared affine combinations of the identity, local marginals, and the state rho; its minimum eigenvalue E4(rho) functions as the entanglement witness.

If this is right

  • All separable states produce a nonnegative 4x4 matrix under the construction.
  • The sign of the minimum eigenvalue can be recovered from single-copy randomized measurements whose number does not increase with dimension.
  • For isotropic states the witness detects entanglement down to mixing parameter p approximately 2/d, nearer the true separability threshold than second-order purity criteria reach.
  • When local reduced states are not maximally mixed, the affine terms involving the marginals are required for the witness to retain its power.

Where Pith is reading between the lines

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

  • The fixed sample complexity suggests the method could be used to certify entanglement in high-dimensional systems where collecting many identical copies is experimentally costly.
  • Similar affine-combination constructions might be applied to other separability criteria to obtain additional low-order witnesses.
  • In practice the third-order terms could be combined with existing second-order tests to improve detection for states close to the separability boundary.

Load-bearing premise

The reduction criterion can be converted without loss or false positives into a 4x4 matrix via squared affine combinations of the identity, local marginals, and the state itself.

What would settle it

A randomized measurement experiment on a known separable isotropic state that returns a negative minimum eigenvalue for the constructed matrix would disprove the positive-semidefiniteness claim.

read the original abstract

Randomized measurements access nonlinear functionals without full tomography, yet turning third-order local single-copy data into a strong entanglement test remains difficult. We convert the reduction criterion into an experimentally measurable separability criterion by testing it on squared affine combinations of the identity, the local marginals, and the state itself. This yields a $4\times4$ matrix $\bar{\mathfrak{M}}(\rho)$ built from experimentally accessible second- and third-order local invariants. Entanglement is certified when its minimum eigenvalue $\mathcal{E}_4(\rho)$ becomes negative. We prove that all separable states satisfy $\bar{\mathfrak{M}}(\rho)\succeq0$, and that the sign of $\mathcal{E}_4(\rho)$ can be inferred from single-copy randomized measurements with dimension-independent sample complexity. For isotropic states on $d\times d$, the second-order purity criterion detects entanglement only for $p\sim d^{-1/2}$, whereas our third-order witness reaches $p\sim 2/d$, close to the separability threshold $p\sim 1/d$. A complementary nonisotropic benchmark shows that the affine marginal directions become essential once the local states are not maximally mixed.

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 paper converts the reduction criterion into an experimentally accessible separability test by applying it to squared affine combinations of the identity, local marginals, and the state ρ itself. This produces a 4×4 matrix bar M(ρ) whose entries are second- and third-order local invariants measurable via single-copy randomized measurements. The authors prove that bar M(ρ) is positive semidefinite for all separable states, that its minimum eigenvalue E4(ρ) certifies entanglement when negative, and that the sign of E4 can be inferred with dimension-independent sample complexity. For isotropic states the witness improves the detectable noise threshold from p∼d^{-1/2} (second-order) to p∼2/d, close to the separability bound.

Significance. If the central construction is faithful, the result supplies a concrete, finite-copy entanglement witness that leverages third-order correlations without tomography and achieves better scaling than existing second-order randomized-measurement criteria for certain families. The dimension-independent sample complexity and explicit construction from the reduction map are notable strengths that could enable near-term experiments on moderate-dimensional systems.

major comments (2)
  1. [derivation of bar M(ρ) (likely §3 or §4)] The load-bearing step is the claim that testing the reduction map on the specific squared affine span {I, ρ_A⊗I, I⊗ρ_B, ρ} yields a 4×4 matrix whose eigenvalues are non-negative precisely when the original reduction criterion holds (or at least that negativity of E4 implies entanglement). The abstract and the skeptic note indicate that the matrix entries are built from second- and third-order invariants, but without the explicit expansion of (id ⊗ R) applied to the affine combinations it is impossible to verify that no false positives are introduced or that detection power is preserved for general states. This equivalence must be shown in detail, preferably with an explicit equation relating the eigenvalues of bar M to those of (id ⊗ R)(ρ).
  2. [sample-complexity analysis] The dimension-independent sample-complexity bound for estimating the sign of E4 inherits the same assumption. If the 4×4 matrix does not faithfully reproduce the reduction criterion, the concentration inequalities used for the randomized measurements may certify a quantity that is not guaranteed to be a valid witness. The proof of the sample-complexity result should therefore be re-checked once the matrix construction is clarified.
minor comments (2)
  1. Notation for the 4×4 matrix (bar M vs. bar frak M) and the eigenvalue E4 should be made uniform throughout the text and figures.
  2. The isotropic-state benchmark would benefit from an explicit comparison table showing the exact p thresholds for the second-order purity witness, the new third-order witness, and the separability boundary.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and constructive comments on our manuscript. We address the two major comments below and will revise the manuscript to improve clarity on the matrix construction while preserving the validity of the witness and sample-complexity results.

read point-by-point responses

  1. Referee: [derivation of bar M(ρ) (likely §3 or §4)] The load-bearing step is the claim that testing the reduction map on the specific squared affine span {I, ρ_A⊗I, I⊗ρ_B, ρ} yields a 4×4 matrix whose eigenvalues are non-negative precisely when the original reduction criterion holds (or at least that negativity of E4 implies entanglement). The abstract and the skeptic note indicate that the matrix entries are built from second- and third-order invariants, but without the explicit expansion of (id ⊗ R) applied to the affine combinations it is impossible to verify that no false positives are introduced or that detection power is preserved for general states. This equivalence must be shown in detail, preferably with an explicit equation relating the eigenvalues of bar M to those of (id ⊗ R)(ρ).

    Authors: We agree that the derivation requires more explicit detail for verification. The manuscript already proves that all separable states satisfy bar M(ρ) ≽ 0 (hence E4(ρ) < 0 certifies entanglement with no false positives for the witness property), but we acknowledge that the step-by-step expansion of (id ⊗ R) on the squared affine combinations {I, ρ_A⊗I, I⊗ρ_B, ρ} and the resulting 4×4 matrix entries may not be presented with sufficient transparency. In the revision we will add the explicit expansion, showing how each entry of bar M(ρ) arises from the second- and third-order local invariants, together with the relation that bar M(ρ) is a Gram-matrix representation of the reduction map restricted to this four-dimensional affine span. This establishes that positivity of the reduction criterion on the span implies bar M(ρ) ≽ 0, while the converse need not hold (our witness is therefore generally weaker than the full reduction criterion but remains valid). We will also include the requested equation linking the eigenvalues of bar M to the action of (id ⊗ R). revision: yes

  2. Referee: [sample-complexity analysis] The dimension-independent sample-complexity bound for estimating the sign of E4 inherits the same assumption. If the 4×4 matrix does not faithfully reproduce the reduction criterion, the concentration inequalities used for the randomized measurements may certify a quantity that is not guaranteed to be a valid witness. The proof of the sample-complexity result should therefore be re-checked once the matrix construction is clarified.

    Authors: The sample-complexity bound is derived from matrix concentration inequalities applied directly to the four entries of bar M(ρ), which are estimated via the second- and third-order randomized-measurement invariants; the bound is therefore independent of the Hilbert-space dimension because the matrix size is fixed at 4×4. Once the explicit derivation requested in the first comment is added, we will re-verify that the concentration analysis remains valid and will explicitly tie the error bounds to the clarified construction of bar M(ρ). Because the witness property (bar M(ρ) ≽ 0 for separable states) is already established, the sign of the estimated E4 continues to provide a rigorous entanglement certificate with the stated dimension-independent sample complexity. revision: partial

Circularity Check

0 steps flagged

No circularity: witness matrix derived directly from reduction criterion with independent proof

full rationale

The paper constructs the 4x4 matrix bar M(rho) by applying the standard reduction criterion to squared affine combinations of I, local marginals, and rho, then separately proves that separable states satisfy bar M(rho) succeq 0. The sign of E4(rho) is inferred from this construction, and sample complexity follows from estimating the resulting second- and third-order invariants. No equation reduces the final eigenvalue test to a fitted parameter, self-referential definition, or load-bearing self-citation; the derivation remains self-contained against the external reduction map benchmark.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 1 invented entities

The central claim rests on the standard reduction criterion from quantum information and basic properties of positive semidefinite matrices and affine combinations. No free parameters are introduced. The matrix itself is a constructed object rather than an independently evidenced entity.

axioms (2)
  • domain assumption The reduction criterion is a valid separability test
    The paper converts this known criterion into the new measurable form.
  • standard math Squared affine combinations preserve the necessary positivity properties
    Used to build the 4x4 matrix and prove it is positive semidefinite for separable states.
invented entities (1)
  • 4x4 matrix bar M(rho) no independent evidence
    purpose: Entanglement witness constructed from second- and third-order local invariants
    Defined in the paper via the affine combinations; no independent falsifiable evidence outside the derivation is provided.

pith-pipeline@v0.9.0 · 5497 in / 1531 out tokens · 60295 ms · 2026-05-10T14:50:12.981306+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

56 extracted references · 56 canonical work pages

  1. [1]

    Leone, J

    L. Leone, J. Rizzo, J. Eisert, and S. Jerbi, Nature Physics 21, 1847 (2025)

    work page 2025

  2. [2]

    Horodecki, P

    R. Horodecki, P. Horodecki, M. Horodecki, and K. Horodecki, Reviews of Modern Physics81, 865 (2009)

    work page 2009

  3. [3]

    G¨ uhne and G

    O. G¨ uhne and G. T´ oth, Physics Reports474, 1 (2009)

    work page 2009

  4. [4]

    Chru´ sci´ nski and G

    D. Chru´ sci´ nski and G. Sarbicki, Journal of Physics A: Mathematical and Theoretical47, 483001 (2014)

    work page 2014

  5. [5]

    Cie´ sli´ nski, S

    P. Cie´ sli´ nski, S. Imai, J. Dziewior, O. G¨ uhne, L. Knips, W. Laskowski, J. Meinecke, T. Paterek, and T. V´ ertesi, Physics Reports1095, 1 (2024). 6

    work page 2024

  6. [6]

    S. A. Ghoreishi, G. Scala, R. Renner, L. L. Tacca, J. Bouda, S. P. Walborn, and M. Paw lowski, Physics Re- ports1149, 1 (2025)

    work page 2025

  7. [7]

    Elben, S

    A. Elben, S. T. Flammia, H.-Y. Huang, R. Kueng, J. Preskill, B. Vermersch, and P. Zoller, Nature Reviews Physics5, 9 (2022)

    work page 2022

  8. [8]

    B. M. Terhal, Theoretical computer science287, 313 (2002)

    work page 2002

  9. [9]

    Huang, R

    H.-Y. Huang, R. Kueng, and J. Preskill, Nature Physics 16, 1050 (2020)

    work page 2020

  10. [10]

    Aaronson, SIAM Journal on Computing49, STOC18 (2020)

    S. Aaronson, SIAM Journal on Computing49, STOC18 (2020)

    work page 2020

  11. [11]

    H. C. Nguyen, J. L. B¨ onsel, J. Steinberg, and O. G¨ uhne, Physical Review Letters129, 220502 (2022)

    work page 2022

  12. [12]

    S. L. Woronowicz, Reports on Mathematical Physics10, 165 (1976)

    work page 1976

  13. [13]

    Peres, Physical Review Letters77, 1413 (1996)

    A. Peres, Physical Review Letters77, 1413 (1996)

    work page 1996

  14. [14]

    Horodecki, P

    M. Horodecki, P. Horodecki, and R. Horodecki, Physics Letters A223, 1 (1996)

    work page 1996

  15. [15]

    Saggio, A

    V. Saggio, A. Dimi´ c, C. Greganti, L. A. Rozema, P. Walther, and B. Daki´ c, Nature Physics15, 935 (2019)

    work page 2019

  16. [16]

    Weilenmann, B

    M. Weilenmann, B. Dive, D. Trillo, E. A. Aguilar, and M. Navascu´ es, Physical Review Letters124, 200502 (2020)

    work page 2020

  17. [17]

    S. Imai, N. Wyderka, A. Ketterer, and O. G¨ uhne, Phys- ical Review Letters126, 150501 (2021)

    work page 2021

  18. [18]

    Aggarwal, S

    S. Aggarwal, S. Adhikari, and A. Majumdar, Physical Review A109, 012404 (2024)

    work page 2024

  19. [19]

    Sarbicki, G

    G. Sarbicki, G. Scala, and D. Chru´ sci´ nski, Phys. Rev. A 101, 012341 (2020)

    work page 2020

  20. [20]

    Sarbicki, G

    G. Sarbicki, G. Scala, and D. Chru´ sci´ nski, Journal of Physics A: Mathematical and Theoretical53, 455302 (2020)

    work page 2020

  21. [21]

    Yang and C.-F

    M.-C. Yang and C.-F. Qiao, Quantum9, 1734 (2025)

    work page 2025

  22. [22]

    Mauro D’Ariano, M

    G. Mauro D’Ariano, M. G. Paris, and M. F. Sacchi, Quantum tomography, inAdvances in Imaging and Elec- tron Physics(Elsevier, 2003) pp. 205–308

    work page 2003

  23. [23]

    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, Nature communications1, 149 (2010)

    work page 2010

  24. [24]

    Christandl and R

    M. Christandl and R. Renner, Physical Review Letters 109, 120403 (2012)

    work page 2012

  25. [25]

    J. Haah, A. W. Harrow, Z. Ji, X. Wu, and N. Yu, IEEE Transactions on Information Theory , 1 (2017)

    work page 2017

  26. [26]

    Kanter and P

    G. Kanter and P. Kumar, Nature Photonics17, 925 (2023)

    work page 2023

  27. [27]

    S. Chen, J. Li, and A. Liu, inProceedings of the 56th An- nual ACM Symposium on Theory of Computing, STOC ’24 (ACM, 2024) pp. 1331–1342

    work page 2024

  28. [28]

    Lowe and A

    A. Lowe and A. Nayak, ACM Transactions on Computa- tion Theory17, 1 (2025)

    work page 2025

  29. [29]

    Cho and D

    G. Cho and D. Kim, arXiv2509.12703 (2025)

    work page arXiv 2025

  30. [30]

    Elben, B

    A. Elben, B. Vermersch, M. Dalmonte, J. I. Cirac, and P. Zoller, Phys. Rev. Lett.120, 050406 (2018)

    work page 2018

  31. [31]

    Vermersch, A

    B. Vermersch, A. Elben, M. Dalmonte, J. I. Cirac, and P. Zoller, Phys. Rev. A97, 023604 (2018)

    work page 2018

  32. [32]

    Rico and F

    A. Rico and F. Huber, Physical Review Letters132, 070202 (2024)

    work page 2024

  33. [33]

    S. J. van Enk and C. W. J. Beenakker, Phys. Rev. Lett. 108, 110503 (2012)

    work page 2012

  34. [34]

    S. Imai, G. T´ oth, and O. G¨ uhne, Phys. Rev. Lett.133, 060203 (2024)

    work page 2024

  35. [35]

    Wyderka, A

    N. Wyderka, A. Ketterer, S. Imai, J. L. B¨ onsel, D. E. Jones, B. T. Kirby, X.-D. Yu, and O. G¨ uhne, Phys. Rev. Lett.131, 090201 (2023)

    work page 2023

  36. [36]

    L. Lami, M. Berta, and B. Regula, Nature Physics22, 439 (2026)

    work page 2026

  37. [37]

    Brydges, A

    T. Brydges, A. Elben, P. Jurcevic, B. Vermersch, C. Maier, B. P. Lanyon, P. Zoller, R. Blatt, and C. F. Roos, Science364, 260 (2019)

    work page 2019

  38. [38]

    Zhang, G

    T. Zhang, G. Smith, J. A. Smolin, L. Liu, X.-J. Peng, Q. Zhao, D. Girolami, X. Ma, X. Yuan, and H. Lu, npj Quantum Information10(2024)

    work page 2024

  39. [39]

    Horodecki and P

    M. Horodecki and P. Horodecki, Physical Review A59, 4206 (1999)

    work page 1999

  40. [40]

    M. A. Jivulescu, N. Lupa, and I. Nechita, Quantum Info. Comput.15, 1165–1184 (2015)

    work page 2015

  41. [41]

    Collins and P

    B. Collins and P. ´Sniady, Communications in Mathemat- ical Physics264, 773 (2006)

    work page 2006

  42. [42]

    Elben, B

    A. Elben, B. Vermersch, C. F. Roos, and P. Zoller, Phys. Rev. A99, 052323 (2019)

    work page 2019

  43. [43]

    Knips, Quantum Views4, 47 (2020)

    L. Knips, Quantum Views4, 47 (2020)

    work page 2020

  44. [44]

    Scala, A

    G. Scala, A. Bera, and G. Sarbicki, Quantum Science and Technology (2025)

    work page 2025

  45. [45]

    Webb, Quantum Information and Computation16, 1379 (2016)

    Z. Webb, Quantum Information and Computation16, 1379 (2016)

    work page 2016

  46. [46]

    Zhu, Phys

    H. Zhu, Phys. Rev. A96, 062336 (2017)

    work page 2017

  47. [47]

    Ketterer, N

    A. Ketterer, N. Wyderka, and O. G¨ uhne, Quantum4, 325 (2020)

    work page 2020

  48. [48]

    Y. Zhou, P. Zeng, and Z. Liu, Phys. Rev. Lett.125, 200502 (2020)

    work page 2020

  49. [49]

    Vermersch, A

    B. Vermersch, A. Rath, B. Sundar, C. Branciard, J. Preskill, and A. Elben, PRX Quantum5, 010352 (2024)

    work page 2024

  50. [50]

    Huber, Journal of Mathematical Physics62, 022203 (2021)

    F. Huber, Journal of Mathematical Physics62, 022203 (2021)

    work page 2021

  51. [51]

    G.-C. Li, L. Chen, X.-S. Hong, S.-Q. Zhang, H. Xu, Y. Liu, Y. Zhou, G. Chen, C.-F. Li, and G.-C. Guo, npj Quantum Information12(2026)

    work page 2026

  52. [52]

    X.-D. Yu, S. Imai, and O. G¨ uhne, Phys. Rev. Lett.127, 060504 (2021)

    work page 2021

  53. [53]

    Neven, J

    A. Neven, J. Carrasco, V. Vitale, C. Kokail, A. Elben, M. Dalmonte, P. Calabrese, P. Zoller, B. Vermersch, R. Kueng, and B. Kraus, npj Quantum Information7 (2021)

    work page 2021

  54. [54]

    J. C. Hoke, M. Ippoliti, E. Rosenberg, and e. all, Nature 622, 481 (2023)

    work page 2023

  55. [55]

    Ketterer, N

    A. Ketterer, N. Wyderka, and O. G¨ uhne, Phys. Rev. Lett. 122, 120505 (2019)

    work page 2019

  56. [56]

    O. Lib, S. Liu, R. Shekel, Q. He, M. Huber, Y. Bromberg, and G. Vitagliano, Physical Review Letters134, 210202 (2025). Appendix A: Maximally entangled benchmark: derivation of Eq.(29)in the main text This section derives the closed-form minimum eigen- valueλ min( ¯M(|Φd⟩⟨Φd|)). Letd A =d B =dandρ= |Φd⟩⟨Φd|with|Φ d⟩= 1√ d Pd−1 j=0 |jj⟩. Thenρis pure and ma...

    work page 2025