Pith. sign in

REVIEW 3 major objections 5 minor 29 references

Statistical models of barren plateaus and anti-concentration of Pauli observables

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper introduces statistical models of barren plateaus and proves that in all three standard sources, any two Pauli observables anti-concentrate so that their non-flat parameter regions barely overlap.

desk verdict A clean, honest paper: the anti-concentration of Pauli observables in the BP regime is a genuine new structural result, with the main caveat being that the expressivity version is proven only for 4-designs and the authors say so. read the letter →

arxiv 2505.08758 v1 pith:P6ANYATX submitted 2025-05-13 quant-ph

Nikita Nemkov This is my paper · ORCID
classification quant-ph MSC 81P68 PACS 03.67.Ac
keywords barrenplateausvariationalquantumalgorithmsPauliobservablesanti-concentrationCliffordgroupunitarydesignslandscapeswarm-start
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper introduces complementary statistical models for the three known sources of barren plateaus in variational quantum algorithms: random Pauli observables for non-locality, random stabilizer states for entanglement, and unitary designs for circuit expressivity. In each model, a typical loss function has exponentially small variance, reproducing the barren plateau phenomenon with probability exponentially close to one. The main new result is anti-concentration: for two different Pauli observables, each is concentrated in its own exponentially small parameter subspace, and these subspaces are essentially independent, so their overlap is exponentially smaller still. The individual concentration scales as $2^{-n}$ while the pairwise overlap scales as $4^{-n}$. If this picture is right, the non-flat parts of a variational quantum landscape are dominated by points where exactly one Pauli term is active, which constrains how warm-start and initialization strategies can work.

What carries the argument

The central object is the anti-concentration correlator $A_{P_1,P_2}=E_{\phi_c}[L_{P_1}(\phi_c)^2L_{P_2}(\phi_c)^2]$, the average over Clifford points of the squared losses of two Pauli terms. The engine is the Clifford-point variance formula: for a Clifford-plus-Pauli-rotations ansatz, the variance of any single-Pauli loss equals the average of its square over the finite set of parameter values where every angle is a multiple of $\pi/2$, so the circuit becomes Clifford. Averaging over random Pauli operators or random stabilizer states lets the Clifford unitaries be absorbed into the random object, reducing the overlap to elementary Pauli counting: $2^{-n}$ for one Pauli and $4^{-n}$ for the pair. For continuous expressivity, the calculation switches to Weingarten calculus on an exact unitary 4-design, whose leading identity-permutation term gives $A'_{P_1,P_2}=4^{-n}(1+O(2^{-n}))$.

What would settle it

Take a Clifford-plus-Pauli-rotation ansatz whose circuit ensemble is an exact (or exponentially close) unitary 2-design but not a 4-design, and compute $A'_{P_1,P_2}=E_{\phi}[L_{P_1}(\phi)^2L_{P_2}(\phi)^2]$ for two single-Pauli observables; if the result scales as $\Theta(2^{-n})$ rather than $\Theta(4^{-n})$, the paper's expressivity-induced anti-concentration claim is false.

Watch

Extended reading notes

Core claim

The central claim is that in the barren plateau regime, random Pauli observables are pairwise anti-concentrated with probability exponentially close to one. Concretely, for the non-locality model with independent random Pauli operators $P_1, P_2$ the Clifford-point overlap correlator satisfies $E_{P_1}E_{P_2}[A_{P_1,P_2}]=4^{-n}$, while each individual Pauli variance is $2^{-n}$. For the entanglement model, averaging over random stabilizer states yields $E_{\rho}[A_{P_1,P_2}]=O(4^{-n})$ for any two distinct non-identity Paulis. For expressivity, assuming the circuit is an exact unitary 4-design, the continuous overlap satisfies $A'_{P_1,P_2}=4^{-n}(1+O(2^{-n}))$. The authors interpret this as showing that the non-flat regions of different Pauli terms barely overlap, so a typical point where the full loss is concentrated actually has only one non-vanishing Pauli contribution.

Load-bearing premise

The general anti-concentration claim is load-bearing on the circuit ensemble being an exact unitary 4-design for continuous expressivity-induced barren plateaus, while the plateau phenomenon itself only requires a 2-design.

Editorial extensions

If this is right

  • In the non-locality and entanglement models, anti-concentration holds with probability exponentially close to one for random pairs of Pauli observables, with overlap $4^{-n}$.
  • At a typical point where the full loss deviates from its mean, a single Pauli term is responsible; regions where several Pauli terms are simultaneously non-negligible occupy an exponentially smaller fraction of the already exponentially small concentrated set.
  • Warm-start strategies that initialize near the non-flat region of one Pauli term will generically be inside the flat region of every other Pauli term, so cross-term optimization cannot be bootstrapped from a single term's signal.
  • If the gap between 2-designs and 4-designs is real, expressivity-induced barren plateaus might occur without Pauli anti-concentration, which would separate the two phenomena; the paper leaves this as an open possibility.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Implicit in the paper, not proven there: the $4^{-n}$ overlap suggests a discrete surrogate for optimization, namely searching over Clifford points where unusually many Pauli terms are simultaneously non-zero, rather than running continuous gradient descent.
  • A sharper diagnostic emerges from this work: the overlap correlator normalized by the individual concentration scale could distinguish one concentrated pocket from many independent pockets, something the loss variance alone cannot do.
  • The statistical models could be transferred to other circuit ensembles, such as matchgate or Gaussian circuits, where the Clifford-point average remains computable but Haar 4-design results do not apply; the anti-concentration rate would then test whether the phenomenon is specific to full unitary designs.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The manuscript introduces statistical models for the three standard sources of barren plateaus in variational quantum algorithms: non-locality of the observable (random Pauli operators), entanglement of the initial state (random stabilizer states), and circuit expressivity (uniform Clifford sampling at Clifford points). It then studies the overlap of the non-zero regions of two Pauli loss functions via the correlator A_{P1,P2}, and shows that in the random-Pauli and random-stabilizer models the averaged overlap is O(4^{-n}), while for a continuous Haar 4-design the analogous overlap is 4^{-n}(1+O(2^{-n})). The authors interpret this as anti-concentration: the exponentially small localized patches of different Pauli observables are essentially disjoint. They report numerical HEA simulations that are consistent with, but not conclusive for, this behavior.

Significance. This is a worthwhile contribution. The paper gives simple ensemble-average proofs rather than worst-case bounds, identifies a structural property of barren-plateau landscapes that is relevant for warm-starting and for understanding the geometry of such landscapes, and ships reproducible PennyLane code. The random-Pauli and random-stabilizer derivations are short and transparent, and the Clifford-point variance identity (Eq. (5)) is used consistently. There are no fitted parameters and the computations are explicit ensemble averages, so the risk of circular reasoning is low. If the expressivity gap discussed below is closed or the claims are properly restricted, the results would be a useful step toward understanding why barren-plateau landscapes are hard to navigate.

major comments (3)
  1. [Expressivity, Eqs. (22)-(25)] The continuous anti-concentration result is proven only when U(φ) is an exact unitary 4-design. However, the expressivity-induced barren plateau that the paper builds on (Eq. (12)) already occurs for 2-designs. The manuscript acknowledges this gap ('our proof does not apply to 2-designs') and the HEA simulations are described as 'somewhat inconclusive.' Since the abstract and conclusion state the anti-concentration result for 'the barren plateau regime' without this qualification, the paper currently overclaims. The authors should either prove the anti-concentration statement for 2-designs, exhibit a 2-design-but-not-4-design counterexample, or explicitly restrict the abstract and conclusion to the statistical models introduced here and to 4-design continuous ensembles.
  2. [Entanglement, Eqs. (16)-(20)] Eq. (20) presents a single positive expression for E_ρ[A_{P1,P2}] for all non-identity, unequal P1,P2. The derivation immediately above it shows that when P1 and P2 anti-commute, the conditional average (17) is exactly zero, because any Pauli anti-commuting with Z_α has an X factor and hence vanishes in ρ0=|0⟩⟨0|. Thus Eq. (20) is only valid as an equality for commuting pairs; for anti-commuting pairs the exact value is 0. Since 0=O(4^{-n}), the anti-concentration conclusion is unchanged, but the displayed formula needs a case split.
  3. [Expressivity, Eqs. (22)-(25)] Eq. (25) is asserted for any P1,P2, but for P1=P2 the leading coefficient is not 1. In the Weingarten sum all three pairings of the four identical Pauli factors give leading contractions, yielding E[Tr(ρU†P U)^4] ≈ 3·4^{-n}; for n=1 with P=Z and Haar-random ρ, E[⟨Z⟩^4]=1/5, whereas 4^{-1}=1/4. The statement that there are 'only two non-vanishing contractions' is also not literally correct, since contractions such as Tr(P1P2P1P2)=±2^n are non-vanishing (though subleading for distinct P1,P2). The O(4^{-n}) scaling survives, but Eq. (25) should be restricted to distinct Pauli observables and the asymptotics stated accordingly.
minor comments (5)
  1. [Statistical model, Eq. (5)] There is a typo in the sentence defining |φ|: 'with |φ| being is the total number' should read 'with |φ| being the total number.'
  2. [Expressivity, Eq. (13)] The notation C O C should be C† O C (or C O C†) for consistency; the Clifford group is invariant under inversion, but the adjoint form is the correct unitary conjugation.
  3. [Anti-concentration, Eqs. (20) and (25)] The paper should explicitly state at both equations that P1 and P2 are distinct non-identity Pauli observables; this is assumed in the counting arguments but never stated clearly in the claims.
  4. [Numerics and Fig. 1] The caption for Fig. 1 says 'axes in Pauli rotation operators R_{P_{i,j}}', which is ambiguous; it should specify that each rotation axis is a randomly chosen Pauli operator.
  5. [Numerics] The numerical section reports estimates from 500 random points but does not state error bars or confidence intervals; given the exponentially decaying signal, some indication of statistical uncertainty would help the reader judge the 'somewhat inconclusive' assessment.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's anti-concentration results are explicit ensemble averages with no fitted parameters and no prediction-from-fit.

full rationale

The derivation chain is self-contained. Each claimed anti-concentration bound is computed as an explicit ensemble average: Eq. (15) for random Pauli observables factors two independent 2^{-n} averages; Eq. (20) for random stabilizer states is a direct Clifford-orbit counting calculation; Eq. (25) for 4-designs follows from a Weingarten expansion of the exact fourth moment. None of these results is defined in terms of the quantity it is said to predict, and no fitted parameter is renamed as a prediction. The only external ingredient that is load-bearing is the Clifford-point variance identity Eq. (5), cited to [17,19]; it is a parameter-free mathematical identity that does not contain the target anti-concentration statement, and it is used as a computational tool rather than as a premise presupposing the conclusion. The self-citation to [17] is for motivation and for Eq. (5), but the central group-theoretic and Weingarten calculations are new and independent of that citation. The paper itself flags the main scope limitation: after Eq. (25) it states 'While our proof does not apply to 2-designs, the anti-concentration result may still be valid,' and it describes the HEA numerical probe as 'somewhat inconclusive.' This is an honest correctness limitation on the generality of the expressivity claim, not a circular reduction. No uniqueness theorem from prior work is invoked to force the chosen model, and no known result is merely renamed.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

No free parameters appear in the derivations; the paper's output is closed-form asymptotic averages. The central assumptions are the three statistical models themselves, plus the 4-design condition for the continuous expressivity result and the Clifford-point variance lemma from prior work. The Clifford group and Weingarten calculations rely on standard facts used without proof.

assumptions (6)
  • standard math Clifford-point variance formula (Eq. 5): for any CPR VQA, E_phi[L^2] equals average over Clifford points phi_c, where every parameter is a multiple of pi/2.
    Imported from [17,19]; it restricts the analysis to Clifford+Pauli-rotation circuits with independent parameters, which is the paper's framework. This is a load-bearing tool for all subsequent ensemble averages.
  • ad hoc to paper The ensemble of random Pauli operators is a faithful model of observable non-locality.
    Proposed in Section 'Non-locality' as a complement to deterministic locality-based BP criteria; no proof connects this ensemble to the usual definition of non-local observables.
  • ad hoc to paper The ensemble of random stabilizer states is a faithful model of high initial-state entanglement.
    Proposed in Section 'Entanglement'; the paper computes averages over the Clifford group instead of over generic entangled states.
  • ad hoc to paper For expressivity-induced BPs, the relevant statistical model is uniform sampling over the Clifford group at Clifford points.
    Section 'Expressivity' introduces this model as a possible statistical version of 2-design expressivity and states that no natural circuit distribution realizing it is known.
  • domain assumption The continuous anti-concentration result assumes the circuit ensemble is an exact unitary 4-design.
    Eq. (22)-(25) use Haar 4-design moments and Weingarten calculus; the paper states 'we always assume that unitary designs are exact'. Standard BP theory for expressivity requires only a 2-design.
  • standard math Standard Clifford group orbit counting facts (transitivity on non-identity Paulis, stabilizer orbit sizes, counts of commuting X^gamma).
    Used in Eqs. (16)-(20) for the entanglement anti-concentration, e.g., |C1|/|C|=(2^n-1)/(4^n-1) and the number of X-type Paulis commuting with a fixed Z^alpha is 2^{n-1}.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Statistical models of barren plateaus and anti-concentration of Pauli observables." pith.science (2026). https://pith.science/paper/P6ANYATX

@misc{pith2026250508758,
  author       = {Pith},
  title        = {Pith review of: Statistical models of barren plateaus and anti-concentration of Pauli observables},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/P6ANYATX}},
  note         = {Machine review of arXiv:2505.08758}
}
read the original abstract

We introduce statistical models for each of the three main sources of barren plateaus: non-locality of the observable, entanglement of the initial state, and circuit expressivity. For instance, non-local observables are modeled by random Pauli operators, which lead to barren plateaus with probability exponentially close to one. These models are complementary to the conventional deterministic ones, and often simpler to analyze. Using this framework, we show that in the barren plateau regime any two Pauli observables are anti-concentrated with high probability in the following sense. While each of the observables is localized in an exponentially small parameter subspace, these regions are essentially independent, so that their overlap is yet exponentially smaller than each subspace. This invites to rethink the structure of quantum landscapes with barren plateaus and approaches to their optimization, including warm-start strategies.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 3 canonical work pages

  1. [1]

    Variational quantum algorithms

    M. Cerezo, Andrew Arrasmith, Ryan Babbush, Simon C. Benjamin, Suguru Endo, Keisuke Fujii, Jarrod R. Mc- Clean, Kosuke Mitarai, Xiao Yuan, Lukasz Cincio, and Patrick J. Coles. “Variational quantum algorithms”. Nature Reviews Physics 2021 3:9 3, 625–644 (2021). arXiv:2012.09265

    arXiv 2021

  2. [2]

    Quan- tum machine learning

    Jacob Biamonte, Peter Wittek, Nicola Pancotti, Patrick Rebentrost, Nathan Wiebe, and Seth Lloyd. “Quan- tum machine learning”. Nature 549, 195–202 (2017). arXiv:1611.09347

    arXiv 2017

  3. [3]

    The effect of data encoding on the expressive power of variational quantum machine learning models

    Maria Schuld, Ryan Sweke, and Johannes Jakob Meyer. “The effect of data encoding on the expressive power of variational quantum machine learning models”. Physical Review A103 (2020). arXiv:2008.08605v2

    arXiv 2020

  4. [4]

    Training Variational Quantum Algorithms Is NP-Hard

    Lennart Bittel and Martin Kliesch. “Training Variational Quantum Algorithms Is NP-Hard”. Physical Review Let- ters 127, 120502 (2021). arXiv:2101.07267

    arXiv 2021

  5. [5]

    Critical Points in Quantum Gener- ative Models

    Eric R. Anschuetz. “Critical Points in Quantum Gener- ative Models” (2021). arXiv:2109.06957

    arXiv 2021

  6. [6]

    Quantum varia- tional algorithms are swamped with traps

    Eric R. Anschuetz and Bobak T. Kiani. “Quantum varia- tional algorithms are swamped with traps”. Nature Com- munications 13, 7760 (2022). arXiv:2205.05786

    arXiv 2022

  7. [7]

    Barren plateaus in quantum neural network training landscapes

    Jarrod R. McClean, Sergio Boixo, Vadim N. Smelyanskiy, Ryan Babbush, and Hartmut Neven. “Barren plateaus in quantum neural network training landscapes”. Nature Communications 9, 1–7 (2018). arXiv:1803.11173

    arXiv 2018

  8. [8]

    Barren plateaus in variational quantum computing

    Mart´ ın Larocca, Supanut Thanasilp, Samson Wang, Kunal Sharma, Jacob Biamonte, Patrick J. Coles, Lukasz Cincio, Jarrod R. McClean, Zo¨ e Holmes, and M. Cerezo. “Barren plateaus in variational quantum computing”. Nature Reviews Physics 7, 174–189 (2025). arXiv:2405.00781

    arXiv 2025

Show all 29 references
  1. [9]

    Effect of barren plateaus on gradient-free optimization

    Andrew Arrasmith, M. Cerezo, Piotr Czarnik, Lukasz Cincio, and Patrick J. Coles. “Effect of barren plateaus on gradient-free optimization”. Quantum 5, 558 (2021). arXiv:2011.12245v2

    2021 arXiv

  2. [10]

    Quantum Con- volutional Neural Networks are (Effectively) Classically Simulable

    Pablo Bermejo, Paolo Braccia, Manuel S. Rudolph, Zo¨ e Holmes, Lukasz Cincio, and M. Cerezo. “Quantum Con- volutional Neural Networks are (Effectively) Classically Simulable” (2024). arXiv:2408.12739

    2024 arXiv

  3. [11]

    Does provable absence of barren plateaus imply classical simulability? Or, why we need to rethink variational quantum computing

    M. Cerezo, Martin Larocca, Diego Garc´ ıa-Mart´ ın, N. L. Diaz, Paolo Braccia, Enrico Fontana, Manuel S. Rudolph, Pablo Bermejo, Aroosa Ijaz, Supanut Thanasilp, Eric R. Anschuetz, and Zo¨ e Holmes. “Does provable absence of barren plateaus imply classical simulability? Or, why...

    2023 arXiv

  4. [12]

    Layerwise learning for quantum neural networks

    Andrea Skolik, Jarrod R. McClean, Masoud Mohseni, Patrick van der Smagt, and Martin Leib. “Layerwise learning for quantum neural networks”. Quantum Ma- chine Intelligence3 (2020). arXiv:2006.14904v1

    2020 arXiv

  5. [13]

    An initialization strategy for addressing barren plateaus in parametrized quantum cir- cuits

    Edward Grant, Leonard Wossnig, Mateusz Ostaszewski, and Marcello Benedetti. “An initialization strategy for addressing barren plateaus in parametrized quantum cir- cuits”. Quantum 3, 214 (2019). arXiv:1903.05076v3

    2019 arXiv

  6. [14]

    Quantum State Preparation with Optimal Circuit Depth: Implementations and Applications

    Xiao-Ming Zhang, Tongyang Li, and Xiao Yuan. “Quantum State Preparation with Optimal Circuit Depth: Implementations and Applications” (2022). arXiv:2201.11495

    2022 arXiv

  7. [15]

    Noise-induced barren plateaus in variational quantum algorithms

    Samson Wang, Enrico Fontana, M. Cerezo, Kunal Sharma, Akira Sone, Lukasz Cincio, and Patrick J. Coles. “Noise-induced barren plateaus in variational quantum algorithms”. Nature Communications 12, 6961 (2021). arXiv:2007.14384

    2021 arXiv

  8. [16]

    Synergy Between Quantum Circuits and Tensor Net- works: Short-cutting the Race to Practical Quantum Ad- vantage

    Manuel S. Rudolph, Jacob Miller, Danial Motlagh, Jing Chen, Atithi Acharya, and Alejandro Perdomo-Ortiz. “Synergy Between Quantum Circuits and Tensor Net- works: Short-cutting the Race to Practical Quantum Ad- vantage” (2022). arXiv:2208.13673

    2022 arXiv

  9. [17]

    Barren plateaus swamped with traps

    Nikita A. Nemkov, Evgeniy O. Kiktenko, and Aleksey K. Fedorov. “Barren plateaus swamped with traps”. Phys- ical Review A 111, 012441 (2025). arXiv:2405.05332

    2025 arXiv

  10. [18]

    A Quantum Approximate Optimization Algorithm

    Edward Farhi, Jeffrey Goldstone, and Sam Gutmann. “A Quantum Approximate Optimization Algorithm” (2014). arXiv:1411.4028

    2014 arXiv

  11. [19]

    Tight and Efficient Gradient Bounds for Parameterized Quantum Circuits

    Alistair Letcher, Stefan Woerner, and Christa Zoufal. “Tight and Efficient Gradient Bounds for Parameterized Quantum Circuits” (2023). arXiv:2309.12681

    2023 arXiv

  12. [20]

    Cost function dependent bar- ren plateaus in shallow parametrized quantum circuits

    M. Cerezo, Akira Sone, Tyler Volkoff, Lukasz Cincio, and Patrick J. Coles. “Cost function dependent bar- ren plateaus in shallow parametrized quantum circuits”. Nature Communications 2021 12:1 12, 1–12 (2021). arXiv:2001.00550

    2021 arXiv

  13. [21]

    Connecting Ansatz Expressibility to Gradient Magnitudes and Barren Plateaus

    Zo¨ e Holmes, Kunal Sharma, M. Cerezo, and Patrick J. Coles. “Connecting Ansatz Expressibility to Gradient Magnitudes and Barren Plateaus”. PRX Quantum 3, 010313 (2022). arXiv:2101.02138v2

    2022 arXiv

  14. [22]

    Symbolic inte- gration with respect to the Haar measure on the uni- tary groups

    Zbigniew Pucha la and J.A. Miszczak. “Symbolic inte- gration with respect to the Haar measure on the uni- tary groups”. Bulletin of the Polish Academy of Sciences Technical Sciences 65, 21–27 (2017). arXiv:1109.4244v2

    2017 arXiv

  15. [23]

    The Clifford group forms a unitary 3- design

    Zak Webb. “The Clifford group forms a unitary 3- design”. Quantum Information and Computation 16, 1379–1400 (2016). arXiv:1510.02769

    2016 arXiv

  16. [24]

    Multiqubit Clifford groups are unitary 3-designs

    Huangjun Zhu. “Multiqubit Clifford groups are unitary 3-designs”. Physical Review A 96 (2015). arXiv:1510.02619

    2015 arXiv

  17. [25]

    The Clifford group fails gracefully to be a unitary 4-design

    Huangjun Zhu, Richard Kueng, Markus Grassl, and David Gross. “The Clifford group fails gracefully to be a unitary 4-design” (2016). arXiv:1609.08172

    2016 arXiv

  18. [26]

    Hardware-efficient Variational Quantum 7 Eigensolver for Small Molecules and Quantum Magnets

    Abhinav Kandala, Antonio Mezzacapo, Kristan Temme, Maika Takita, Markus Brink, Jerry M. Chow, and Jay M. Gambetta. “Hardware-efficient Variational Quantum 7 Eigensolver for Small Molecules and Quantum Magnets”. Nature 549, 242–246 (2017). arXiv:1704.05018

    2017 arXiv

  19. [27]

    Estimating the randomness of quantum circuit ensem- bles up to 50 qubits

    Minzhao Liu, Junyu Liu, Yuri Alexeev, and Liang Jiang. “Estimating the randomness of quantum circuit ensem- bles up to 50 qubits”. npj Quantum Information 8, 137 (2022). arXiv:2205.09900

    2022 arXiv

  20. [28]

    Pen- nyLane: Automatic differentiation of hybrid quantum- classical computations

    Ville Bergholm, Josh Izaac, Maria Schuld, Christian Gogolin, Shahnawaz Ahmed, Vishnu Ajith, M. Sohaib Alam, Guillermo Alonso-Linaje, B. AkashNarayanan, Ali Asadi, Juan Miguel Arrazola, Utkarsh Azad, Sam Ban- ning, Carsten Blank, Thomas R Bromley, Benjamin A. Cordier, Jack Cero...

    2018 arXiv

  21. [29]

    Nemkov (2024)

    N. Nemkov (2024). url: https://github.com/idnm/ barren_traps/tree/hea

    2024

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.