REVIEW 4 major objections 5 minor 51 references
Cross-Platform Verification of Intermediate Scale Quantum Devices
T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read A fidelity between two unknown quantum states on separate devices can be measured from shared random measurements and classical communication alone.
desk verdict Solid core: Eq. (2) is new and correctly derived, but the Appendix E 'no false positives' claim is false for states with overlapping marginals, and the experiment is one platform, not two. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The workhorse is the cross-correlation estimator in Eq. (2): the weighted sum $\sum_{s_A,s'_A}(-d)^{-D[s_A,s'_A]} P^{(1)}_{U_A}(s_A) P^{(2)}_{U_A}(s'_A)$, averaged over shared local unitary 2-designs. A unitary 2-design is an ensemble whose first two moments match uniformly random unitaries, which lets the average be evaluated analytically. Under the ensemble average the two-copy operator built from this sum collapses to the swap operator, leaving exactly $\mathrm{Tr}[\rho_1\rho_2]$; the same machinery yields purities as autocorrelations. The protocol thus converts a quantum overlap into ordinary statistical correlations of measurement outcomes, which can be compared between two platforms using only classical communication.
What would settle it
Prepare the same known product state, such as $|0\cdots 0\rangle$, on both platforms and run the protocol. If the inferred $F_{\max}$ is substantially below unity and drops further when each random unitary is replaced by two consecutive random unitaries, then systematic unitary errors and decoherence during the measurement basis are contaminating the estimate, and the protocol is measuring the unitary mismatch as much as the states.
Extended reading notes
Core claim
The central claim is that the cross-platform fidelity of two unknown, possibly mixed states can be inferred directly from randomized measurements. For subsystems $A_1,A_2$ with equal size $N_A$, let $U_A = \otimes_{k=1}^{N_A} U_k$ be a product of independent local random unitaries drawn from a unitary 2-design, and let $P^{(i)}_{U_A}(s_A)$ be the probability that device $i$ observes outcome string $s_A$ after applying $U_A$. The paper proves $$\mathrm{Tr}[\rho_{1,A_1}\rho_{2,A_2}] = $d^{{N_A}}$ \sum_{s_A,s'_A} (-d)^{-D[s_A,s'_A]} \overline{ $P^{{(1)}}$_{U_A}(s_A) $P^{{(2)}}$_{U_A}(s'_A) },$$ where $D[s_A,s'_A]$ is the number of positions at which the two outcome strings differ and the overline is the ensemble average over the random unitaries. Setting the two indices equal recovers the purities from the same formula, and normalizing the overlap by the larger purity defines $F_{\max}$. The proof uses a two-design averaging identity to turn the averaged two-copy operator into the swap operation, so that the cross-correlation sum becomes the trace product of the two density matrices.
Load-bearing premise
The load-bearing premise is that both devices can implement the same declared random unitaries accurately enough that the measured cross-correlations are dominated by the difference between the two states rather than by the difference between the two implementations of the random unitaries.
Editorial extensions
If this is right
- Two quantum devices in different places and times can be checked against each other with only classical communication of random unitaries and outcomes, with no quantum link required.
- A quantum simulator can be verified against a classical simulation of its target state; the paper reports experiment-theory fidelities above 0.6 even after 5 ms of many-body dynamics on 10 qubits.
- The measurement budget scales as roughly $2^{bN_A}$ with $b\approx 0.6$-$0.8$ for $N_A$-qubit subsystems, well below the $2^{2N_A}$ scale of full state tomography, making tens of qubits accessible.
- The same data yields any fidelity that depends only on overlap and purities, including the geometric-mean fidelity $F_{\mathrm{GM}}$, which is first-order insensitive to local depolarizing noise.
- By splitting one experimental dataset into two 'experiments', the method also benchmarks the reproducibility of a single device; the authors find experiment-experiment fidelities higher than experiment-theory fidelities.
Reading between the lines
- One consequence the authors leave implicit is that the shared random-unitary set could become a common reference for a network of different quantum devices, certifying two machines as equivalent without either being trusted as the ground truth, provided both can implement the same declared unitary set.
- The sensitivity to unitary mismatch is itself a diagnostic resource: measuring the same known state with one versus two concatenated random unitaries separates state fidelity from unitary calibration error, because the estimated fidelity drops when the unitary path lengthens.
- The scaling analysis suggests that an adaptive measurement-allocation scheme, guided by bootstrap error estimates, could reduce the total budget below the quoted $2^{bN}$ bound when partial prior knowledge of the states is available.
- The same estimator could be used to track a system's memory of its own earlier state, since the fidelity of a state with its own version at a later time decays slowly in the disordered setting the paper studies.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proposes a protocol to estimate the overlap Tr[\rho_1\rho_2] and purities of two quantum states prepared on separate platforms, using only local randomized measurements with the same local random unitaries communicated classically. From these quantities it defines the mixed-state fidelity Fmax and demonstrates, using data from Ref. [28], experiment-theory fidelities of 10-qubit trapped-ion states and 'experiment-experiment' fidelities obtained by splitting one experimental data set into two halves. Appendix A derives Eq. (2) using Weingarten calculus for product local 2-designs, Appendices C and D study statistical errors and resampling, and Appendix E models the effect of unitary errors and depolarization on the estimators.
Significance. If the protocol's practical guarantees hold, this is a valuable and timely tool: it offers a concrete, scalable-in-practice route to compare unknown states across devices or against a classical target without tomographic overhead, and the paper provides an explicit machine-checkable-style derivation of the central identity, careful resource scaling numerics, and a genuine 10-qubit proof-of-principle from existing trapped-ion data. The paper also correctly identifies and experimentally probes unitary calibration errors, which is the main practical threat to cross-platform operation. The central Eq. (2) is derived, not fitted, and the statistical-error analysis with bootstrap resampling is a useful contribution in itself.
major comments (4)
- [Appendix E, Eq. (E6)] The robustness claim in the main text and Appendix E that unitary errors 'do not lead to false positives' is not established for general pairs of states. For orthogonal states with overlapping marginals, Eq. (E6) gives E_{1,2} ≈ (η_1^2 + η_2^2) Σ_k Tr[Tr_{(k)}[ρ_1] Tr_{(k)}[ρ_2]], which is positive even when Tr[ρ_1 ρ_2] = 0. With unit purities, the estimated Fmax is then ≈ η^2 while the true fidelity is zero. The numerical check in Fig. E.1(a) only considers ρ_1 = ρ_2 and therefore cannot detect this false-positive mechanism. This is load-bearing because the protocol's practical appeal is exactly that imperfections degrade, rather than inflate, the measured fidelity; as written, the claim is one-sided and the dangerous failure mode (unitary mismatch creating spurious agreement) is not excluded.
- [Section 'Fidelity estimation with trapped ions', Fig. 4(a,b)] The 'experiment-experiment' demonstration does not compare two physical platforms. The data are split into two halves E1 and E2 from the same experimental run of Ref. [28]; both halves estimate the same underlying state, so high Fmax is largely forced by construction and does not test the cross-platform scenario advertised in the title and abstract. This should be clearly labeled as a self-consistency check, and the paper should state explicitly that no two-device demonstration is provided.
- [Appendix E, Sec. 3 and Fig. E.2] Fig. E.2 shows that implementing a single random unitary twice (two concatenated unitaries) lowers the estimated fidelity of a known product state from near one to significantly below one, and the paper itself attributes this to unitary errors. In a cross-platform setting, the same random U_A is implemented by two different devices with independent calibration, so the relevant error is the mismatch between U_A^{(1)} and U_A^{(2)}, not a common error. The paper does not quantify this differential error; without a model or a two-device test, the protocol's applicability to genuinely different platforms remains unproven, and the false-positive mechanism of the first major comment becomes the relevant worst case.
- [Eq. (2) and Appendix A] The central derivation is sound: averaging the local 2-design over each qudit and using the swap trick yields Tr[ρ_{i,A_i} ρ_{j,A_j}] exactly, with no fitting parameters. I stress this as a strength rather than a weakness. The concern is only that the experimental and robustness sections claim a guarantee that the derivation itself does not provide; the derivation is for ideal unitaries, while Appendix E's error analysis is incomplete as described above.
minor comments (5)
- [Abstract and title] The title and abstract promise 'cross-platform verification', but no two physical platforms are compared anywhere in the paper. Consider rewording to 'cross-platform protocol with single-platform demonstration' or similar, to avoid overclaiming.
- [Main text, sentence after Eq. (1)] The definition of Fmax uses 'max{Tr[ρ_1^2], Tr[ρ_2^2]}' but the text in the paragraph and elsewhere sometimes writes Tr[ρ_i^2] with indices interchanged; this is not a technical error, but the notation would be clearer if the two reduced-state labels were consistently distinguished as ρ_{1,A_i} and ρ_{2,A_j} throughout.
- [Fig. 2 caption] The captions of Fig. 2 and Fig. D.1 report scaling exponents b = 0.8 ± 0.1 and b = 0.6 ± 0.1 (and 0.8 ± 0.1, 0.5 ± 0.1 in Fig. D.1) without stating the fit range or confidence intervals; please specify the fitting procedure and the NA range used, and clarify whether the errors are statistical or systematic.
- [Appendix B, Eq. (B1)] The claim that FGM is robust to 'global dephasing of arbitrary strength λ' is stated with only the order O(D^{-1}) error; since Appendix E shows that local depolarization affects Fmax at first order, the distinction between global and local noise in this statement should be made explicit.
- [References] The paper cites relevant work on randomized measurements and direct fidelity estimation, but does not cite the companion experimental paper that introduced the data in Ref. [28] beyond the data source; consider acknowledging that the data were not collected for the present protocol and that the measurement sequences were post-hoc re-used, which is relevant to the interpretation of Fig. 4.
Circularity Check
Eq. (2) and the core protocol are independently derived; score 4 for demonstration-level circularity in the split-data experiment and the experiment-calibrated theory state.
-
self definitional
[Main text, 'Fidelity estimation with trapped ions' (p. 3-4) and Fig. 4 caption.]
"As a first step towards the cross-platform verification of two quantum devices, we now present experiment-experiment fidelities of quantum states prepared sequentially in the same experiment. To this end, we divide the data obtained in Ref. [28] into two parts, from now on called experiment E1 and experiment E2, each consisting of measurement outcomes for the same NU = 500 random unitaries and NM = 75 measurements per random unitary. Using Eq. (2), we calculate overlap and purities, and from this the fidelity Fmax(ρE1(t),ρE2(t))."
E1 and E2 are not independent device preparations: they are disjoint subsets of shots drawn from the same experimental probability distributions for the same 500 random unitaries acting on the same physical state. For a true state ρ, Eq. (2) applied to the two empirical distributions estimates Tr[ρρ]=Tr[ρ²] for the numerator while the denominator estimates max(Tr[ρ²],Tr[ρ²]), so the expected value of the estimated Fmax is 1 up to shot noise. The near-unity experiment-experiment result is therefore a property of the estimator acting on a single split dataset, not a comparison of two independently prepared quantum devices.
-
fitted input called prediction
[Main text, 'Fidelity estimation with trapped ions' (p. 3) and Fig. 3 caption.]
"To numerically simulate the experiment and obtain a corresponding theory state ρT (t), we perform exact diagonalization to simulate unitary dynamics or exactly solve a master equation to include decoherence effects. ... for (b) we additionally include decoherence effects, inherent to the state preparation (imperfect initial state preparation, spin-flips and dephasing noise) and the measurement process (depolarizing noise during the random measurement) [28]."
The 'theory state' used to compute the reported experiment-theory fidelity is not a parameter-free first-principles target: its decoherence model is taken from Ref. [28], the same experiment whose data are being scored. Because Appendix E shows that decoherence lowers the estimated fidelity, inserting those experimentally calibrated noise terms into ρT mechanically raises Fmax relative to the unitary-theory curve. The measured experiment-theory fidelity is thus partly a goodness-of-fit to a model already containing the experiment's own error parameters, rather than an independent cross-platform or cross-theory prediction.
full rationale
The central mathematical claim, Eq. (2), is self-contained: Appendix A evaluates the two-copy twirl under independent local 2-designs and explicitly reduces the weighted cross-correlation to the swap operator, giving Tr[ρ_i ρ_j]. No parameter is fitted in that derivation, and the cited Weingarten identity is a standard, externally checkable fact rather than a load-bearing self-citation. The fidelity Fmax is then simply the ratio of the estimated overlap to estimated purities, so the protocol itself is not circular. The circularity is confined to the experimental demonstration. The experiment-experiment comparison splits one dataset from Ref. [28] into E1 and E2 with identical random unitaries, so the high Fmax is the expected value of the estimator for identical states and does not demonstrate agreement between independent devices. The theory-experiment comparison is also weakened as an independent test because the theory state in Fig. 3(b) includes decoherence parameters taken from the same experiment [28]. These are demonstration-level circularities and do not invalidate the protocol derivation. The Appendix E claim that imperfections 'do not lead to false positives' is contradicted by Eq. (E6) for orthogonal states with overlapping marginals, but that is a correctness/robustness flaw, not a circularity, so it is not counted in the score.
Assumptions & free parameters
free parameters (2)
- b (scaling exponent, product states) =
0.8 ± 0.1
- b (scaling exponent, Haar random states) =
0.6 ± 0.1
assumptions (5)
- domain assumption Local random unitaries Uk are sampled independently from a unitary 2-design on Cd.
- domain assumption The same set of random unitaries UA is applied on both platforms.
- standard math Outcome probabilities are obtained from projective measurements and finite samples; ensemble averages are approximated by finite sums.
- domain assumption Fmax as defined in Eq. (1) is a valid mixed-state fidelity (satisfies Jozsa axioms).
- ad hoc to paper The noise model parameters in the theory states (Appendix E) match the experimental imperfections of Ref. [28].
Cite this review
Pith. "Pith review of Cross-Platform Verification of Intermediate Scale Quantum Devices." pith.science (2026). https://pith.science/paper/AMJXFFQV
@misc{pith2026190901282,
author = {Pith},
title = {Pith review of: Cross-Platform Verification of Intermediate Scale Quantum Devices},
year = {2026},
howpublished = {\url{https://pith.science/paper/AMJXFFQV}},
note = {Machine review of arXiv:1909.01282}
}
abstract
We describe a protocol for cross-platform verification of quantum simulators and quantum computers. We show how to measure directly the overlap $\textrm{Tr}\left[\rho_1 \rho_2\right]$ and the purities $\textrm{Tr}\left[\rho^2_{1,2}\right]$, and thus a fidelity of two possibly mixed quantum states $\rho_1$ and $\rho_2$ prepared in separate experimental platforms. We require only local measurements in randomized product bases, which are communicated classically. As a proof-of-principle, we present the measurement of experiment-theory fidelities for entangled $10$-qubit quantum states in a trapped ion quantum simulator.
Figures
Reference graph
Works this paper leans on
-
[17]
Elben, B
A. Elben, B. Vermersch, C. F. Roos, and P. Zoller, Phys. Rev. A99, 052323 (2019)
2019
-
[28]
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)
2019
-
[1]
Preskill, Quantum2, 79 (2018)
J. Preskill, Quantum2, 79 (2018)
2018
- [2]
- [3]
-
[4]
J. Emerson, M. Silva, O. Moussa, C. Ryan, M. Laforest, J. Baugh, D. G. Cory, and R. Laflamme, Science317, 1893 (2007)
work page 2007
-
[5]
Knill, D
E. Knill, D. Leibfried, R. Reichle, J. Britton, R. B. Blakestad, J. D. Jost, C. Langer, R. Ozeri, S. Seidelin, and D. J. Wineland, Phys. Rev. A77, 012307 (2008)
2008
-
[6]
D. Lu, H. Li, D.-A. Trottier, J. Li, A. Brodutch, A. P. Krismanich, A. Ghavami, G. I. Dmitrienko, G. Long, J. Baugh, and R. Laflamme, Phys. Rev. Lett. 114, 140505 (2015)
work page 2015
Show all 51 references
-
[7]
Erhard, J
A. Erhard, J. J. Wallman, L. Postler, M. Meth, R. Stricker, E. A. Martinez, P. Schindler, T. Monz, J. Emerson, and R. Blatt, arXiv:1902.08543 (2019)
2019 arXiv
-
[8]
M. P. da Silva, O. Landon-Cardinal, and D. Poulin, Phys. Rev. Lett.107, 210404 (2011)
2011
-
[9]
S. T. Flammia and Y.-K. Liu, Phys. Rev. Lett. 106, 230501 (2011)
2011
-
[10]
B. P. Lanyon, C. Maier, M. Holzäpfel, T. Baumgratz, C. Hempel, P. Jurcevic, I. Dhand, A. S. Buyskikh, A. J. Daley, M. Cramer, M. B. Plenio, R. Blatt, and C. F. Roos, Nat. Phys.13, 1158 (2017)
2017
-
[11]
Liang, Y.-H
Y.-C. Liang, Y.-H. Yeh, P. E. M. F. Mendonça, R. Y. Teh, M. D. Reid, and P. D. Drummond, Rep. Prog. Phys. 82, 076001 (2019)
2019
-
[12]
Jozsa, J
R. Jozsa, J. Mod. Opt.41, 2315 (1994)
1994
-
[13]
N. M. Linke, D. Maslov, M. Roetteler, S. Debnath, C. Figgatt, K. A. Landsman, K. Wright, and C. Monroe, Proc. Nat. Acad. Sci.114, 3305 (2017)
2017
- [14]
-
[15]
S. J. van Enk and C. W. J. Beenakker, Phys. Rev. Lett. 108, 110503 (2012)
2012
-
[16]
Elben, B
A. Elben, B. Vermersch, M. Dalmonte, J. I. Cirac, and P. Zoller, Phys. Rev. Lett.120, 050406 (2018)
2018
-
[18]
Häffneret al., Nature438, 643 (2005)
H. Häffneret al., Nature438, 643 (2005)
2005
-
[19]
Gross, Y
D. Gross, Y. K. Liu, S. T. Flammia, S. Becker, and J. Eisert, Phys. Rev. Lett.105, 150401 (2010)
2010
-
[20]
Riofrío, D
C. Riofrío, D. Gross, S. Flammia, T. Monz, D. Nigg, R. Blatt, and J. Eisert, Nat. Commun.8, 15305 (2017)
2017
-
[21]
Torlai, G
G. Torlai, G. Mazzola, J. Carrasquilla, M. Troyer, R. Melko, and G. Carleo, Nat. Phys.14, 447 (2018)
2018
-
[22]
A. C. Keith, C. H. Baldwin, S. Glancy, and E. Knill, Phys. Rev. A98, 042318 (2018)
2018
-
[23]
Carleo, I
G. Carleo, I. Cirac, K. Cranmer, L. Daudet, M. Schuld, N. Tishby, L. Vogt-Maranto, and L. Zdeborová, arXiv:1903.10563 (2019)
2019 arXiv
-
[24]
Existence of a quantum link would allow quantum state transfer and provide, in prin- ciple, an efficient quantum protocol for Tr[ρ1ρ2] [25–27]
Our protocol assumes only classical communication be- tween platforms 1 and 2. Existence of a quantum link would allow quantum state transfer and provide, in prin- ciple, an efficient quantum protocol for Tr[ρ1ρ2] [25–27]
-
[25]
A. J. Daley, H. Pichler, J. Schachenmayer, and P. Zoller, Phys. Rev. Lett.109, 020505 (2012)
2012
-
[26]
Islam, R
R. Islam, R. Ma, P. M. Preiss, M. E. Tai, A. Lukin, M. Rispoli, and M. Greiner, Nature528, 77 (2015)
2015
-
[27]
N. M. Linke, S. Johri, C. Figgatt, K. A. Landsman, A. Y. Matsuura, and C. Monroe, Phys. Rev. A 98, 052334 (2018)
2018
-
[29]
Gross, K
D. Gross, K. Audenaert, and J. Eisert, J. Math. Phys. 48, 052104 (2007). 6
2007
-
[30]
Dankert, R
C. Dankert, R. Cleve, J. Emerson, and E. Livine, Phys. Rev. A80, 012304 (2009)
2009
-
[31]
See Supplemental Material
-
[32]
To allow for estimation of the statistical uncertainty of estimated fidelity a minimal number ofNU ≫ 1 is re- quired
-
[33]
Efron and G
B. Efron and G. Gong, Am. Stat.37, 36 (1983)
1983
-
[34]
Serbyn and D
M. Serbyn and D. A. Abanin, Phys. Rev. B96, 014202 (2017)
2017
-
[35]
R. Fan, P. Zhang, H. Shen, and H. Zhai, Sci. Bull.62, 707 (2017)
2017
-
[36]
X. Chen, T. Zhou, D. A. Huse, and E. Fradkin, Ann. Phys. 529, 1600332 (2017)
2017
-
[37]
Huang, Y.-L
Y. Huang, Y.-L. Zhang, and X. Chen, Ann. Phys.529, 1600318 (2017)
2017
-
[38]
Vermersch, A
B. Vermersch, A. Elben, L. M. Sieberer, N. Y. Yao, and P. Zoller, Phys. Rev. X9, 021061 (2019)
2019
-
[39]
Haake, Quantum Signatures of Chaos , Vol
F. Haake, Quantum Signatures of Chaos , Vol. 54 (Springer, 2010)
2010
-
[40]
Watrous,The Theory of Quantum Information(Cam- bridge University Press, 2018)
J. Watrous,The Theory of Quantum Information(Cam- bridge University Press, 2018)
2018
-
[41]
D. A. Roberts and B. Yoshida, Journal of High Energy Physics 2017, 121 (2017)
2017
-
[42]
Nakata, C
Y. Nakata, C. Hirche, M. Koashi, and A. Winter, Phys. Rev. X7, 021006 (2017)
2017
-
[43]
Vermersch, A
B. Vermersch, A. Elben, M. Dalmonte, J. I. Cirac, and P. Zoller, Phys. Rev. A97, 023604 (2018)
2018
-
[44]
Hotelling, J
H. Hotelling, J. Royal Stat. Soc15, 193 (1953)
1953
-
[45]
7 Supplemental Material Appendix A: Proof of Equation 2 In this section, we prove Eq
This is an ensemble of Hermitian random matrices with complex entries which real and imaginary parts are inde- pendently distributed according to the standard normal distribution [39]. 7 Supplemental Material Appendix A: Proof of Equation 2 In this section, we prove Eq. (2) of...
-
[46]
Alternatively, one can use global random unitaries UA from a unitary2-design defined on the entire Hilbert spaceH = ( Cd)NA with dimension DA = dNA
Fidelities from global random unitaries In the main text, we present a protocol to estimate the overlap of two density matrices Tr[ρ1ρ2] using local random unitaries of the formUA = ⨂ k∈AUk with Uk from a unitary2-design defined on the local Hilbert space Cd. Alternatively, one...
-
[47]
(2) of the MT]
Local random unitaries For local random unitaries, the probabilitiesP (i) U (sA) (i = 1, 2) are not independent for differentsA and the fidelities are functions of the probabilitiesP (i) U (sA) for all basis states sA [see Eq. (2) of the MT]. We thus rely on numerical simulation...
-
[48]
To achieve this, we use Bootstrap resampling [33] across the random unitaries to estimate the standard error
Resampling techniques and allocation of the measurement budget In an experiment, one would like to infer the statistical uncertainty of the measured fidelities directly from the measured data (and not from performing the experiment many times with the same parametersNM andNU). ...
-
[49]
Modelling of errors Our approach to estimate cross-platforms fidelities is based on realizing thesame local random unitariesUA on two different platforms. Restricting for clarity to the 12 case of qubits (d = 2), we study the effects of systematic errors due to: (i) Unitary error...
-
[50]
For simplicity of notation, we drop the subscriptA in this subsection
Error estimates In the following, we evaluate the estimatorsEi,j. For simplicity of notation, we drop the subscriptA in this subsection. Due to the independence ofV (i) and U, we can use Eq. (2) MT, and find E1,2 = Tr [ V (1) ˜ρ1V (1)†V (2) ˜ρ2V (2)† ] (E3) and Ei,i = Tr[˜ρi ˜ρ...
-
[51]
2 4 6 8 10 N A 0.8 1.0 Fmax (ρT ρρ E ) Concatenated Single 2 4 6 8 10 N A 0.8 1.0 FGM (ρT ρρ E ) FIG
Testing experimentally the influence of imperfections In this subsection, we analyze the influence of imper- fections in our protocol experimentally. 2 4 6 8 10 N A 0.8 1.0 Fmax (ρT ρρ E ) Concatenated Single 2 4 6 8 10 N A 0.8 1.0 FGM (ρT ρρ E ) FIG. E.2. Testing experimentally...
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.