arxiv: 2604.11155 · v1 · submitted 2026-04-13 · ❄️ cond-mat.stat-mech · cond-mat.dis-nn· math.PR

Recognition: no theorem link

Tensor-Network Population Annealing

Takumi Oshima , Yuma Ichikawa , Koji Hukushima

Authors on Pith no claims yet

Pith reviewed 2026-05-10 15:45 UTC · model grok-4.3

classification ❄️ cond-mat.stat-mech cond-mat.dis-nnmath.PR

keywords tensor networkpopulation annealingspin glassIsing modelEdwards-Andersonsamplingeffective sample sizelow temperature

0 comments

The pith

Hybrid tensor-network population annealing enables stable low-temperature sampling in spin glasses by initializing with tensor contractions and finishing with population annealing.

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

The paper introduces tensor-network population annealing (TNPA) as a way to sample equilibrium configurations of frustrated systems like the two-dimensional Edwards-Anderson Ising spin glass at low temperatures. Pure tensor-network contractions become numerically unstable once frustration grows strong, while standard population annealing started from high temperature needs a long cooling schedule to equilibrate. TNPA therefore performs tensor-network contractions only down to a temperature chosen by an effective-sample-size diagnostic that marks where the output remains reliable, then hands the population over to annealing for the final low-temperature stage. This division lets each technique operate in the regime where it is strongest. A reader would care because the method supplies concrete initial states close to equilibrium, shortening the work required to reach accurate low-temperature statistics.

Core claim

The paper claims that tensor-network population annealing, which generates an initial population of spin configurations via tensor-network contractions at an adaptively chosen temperature and then equilibrates them further with population annealing, yields reliable low-temperature samples for the two-dimensional Edwards-Anderson Ising spin glass. The effective sample size diagnostic identifies the switch temperature so that the tensor-network stage stays stable while the subsequent annealing stage requires only modest computational effort.

What carries the argument

The effective sample size diagnostic that monitors tensor-network output quality to select the initialization temperature at which to switch to population annealing.

If this is right

Low-temperature sampling of the two-dimensional Edwards-Anderson model becomes feasible with reduced total computational effort.
Tensor-network contractions can be restricted to their numerically safe temperature window without sacrificing access to deeper cooling.
The annealing schedule can be shortened because the initial population already lies near equilibrium at the switch temperature.
The same hybrid logic applies to other classical spin models where tensor networks remain accurate only above a frustration-dependent temperature.

Where Pith is reading between the lines

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

The diagnostic could be reused in other Monte Carlo schemes to decide when to switch from one sampler to another.
Extending the approach to three-dimensional or quantum spin glasses would require only that a suitable tensor-network representation exists above some temperature.
Direct comparison of final sample quality against pure population annealing on the same model would quantify the practical speedup.

Load-bearing premise

The effective sample size diagnostic picks a temperature where the tensor-network configurations are both stable and close enough to equilibrium that population annealing can finish the cooling without excessive extra cost.

What would settle it

Running the method on the two-dimensional Edwards-Anderson model and finding that the population-annealing stage still requires as many temperature steps as a pure high-temperature start, or that the final energy or overlap distributions deviate from known equilibrium values.

Figures

Figures reproduced from arXiv: 2604.11155 by Koji Hukushima, Takumi Oshima, Yuma Ichikawa.

**Figure 2.** Figure 2: FIG. 2. Top: Energy-density histogram for [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Figure 3: FIG. 3. ESS ratios for 100 random instances of the two [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

**Figure 4.** Figure 4: FIG. 4. Zig-zag ordering of spins used in the autoregressive [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗

**Figure 5.** Figure 5: FIG. 5. Temperature dependence of the energy density of [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗

**Figure 6.** Figure 6: FIG. 6. Temperature dependence of the spin-glass suscep [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗

**Figure 7.** Figure 7: FIG. 7. Top: Dependence of the spin-glass susceptibility on [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗

**Figure 8.** Figure 8: FIG. 8. Top: Temperature dependence of the entropy density [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗

**Figure 9.** Figure 9: FIG. 9. Temperature dependence of the energy density of the [PITH_FULL_IMAGE:figures/full_fig_p011_9.png] view at source ↗

**Figure 10.** Figure 10: FIG. 10. Temperature dependence of the family entropy dur [PITH_FULL_IMAGE:figures/full_fig_p012_10.png] view at source ↗

read the original abstract

We propose a hybrid sampling method, tensor-network population annealing (TNPA), which combines tensor-network (TN) initialization with population annealing (PA). We apply this method to the two-dimensional Edwards-Anderson Ising spin glass. The approach is motivated by the limitations of existing methods: TN-based samplers can become numerically unstable in frustrated spin systems at low temperatures, whereas conventional PA requires a long annealing schedule when started from the high-temperature limit. In TNPA, TN contractions are used only within a reliable temperature range to generate initial configurations that are close to the equilibrium distribution. The subsequent low-temperature equilibration is then carried out by PA. To stabilize the initialization process, we introduce a diagnostic based on the effective sample size that adaptively selects the initialization temperature. The proposed framework provides a practical and physically motivated route to low-temperature sampling by combining the complementary strengths of TN and PA.

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.

Desk Editor's Note private letter to a colleague

TNPA is a sensible hybrid that uses ESS to switch from tensor networks to population annealing in the 2D Edwards-Anderson model, but the diagnostic may not guarantee the initial samples are close enough to equilibrium.

read the letter

The paper puts forward tensor-network population annealing for the two-dimensional Edwards-Anderson Ising spin glass. Tensor networks generate starting configurations only down to a temperature chosen by an effective sample size diagnostic, after which population annealing takes over for the colder regime. The ESS threshold decides when the TN contractions are still stable enough to hand off. This adaptive switch is the concrete addition that is not standard in either the TN or PA literature the abstract cites. It targets the real practical limits: TN methods lose reliability at low T in frustrated systems because of truncation, while PA from the high-T limit needs a long schedule to keep diversity. Starting PA from a TN-seeded population at moderate T should shorten the remaining work. The motivation is straightforward and the diagnostic is a reasonable engineering choice for staying inside the safe contraction window. The soft spot is the assumption that ESS saturation means the TN output is close to the true equilibrium distribution rather than merely non-degenerate. In the 2D EA model, finite bond dimension still truncates frustrated loops and can produce biased low-energy configurations even when the contraction itself looks stable. ESS tracks sample collapse and numerical stability, not deviation from the correct marginals, so it is possible for the handoff to occur with systematic errors that PA then has to overcome at extra cost. The abstract contains no benchmarks, error bars, or comparisons to exact results or pure PA runs, so the size of any practical gain is not yet visible. This is aimed at people who already run Monte Carlo or tensor methods on spin glasses and want a concrete way to push to lower temperatures without prohibitive schedules. It is worth sending to peer review so the numerical tests and any validation against known 2D EA observables can be examined directly.

Referee Report

2 major / 1 minor

Summary. The manuscript proposes tensor-network population annealing (TNPA), a hybrid method for low-temperature sampling in the two-dimensional Edwards-Anderson Ising spin glass. Tensor-network contractions generate initial configurations only up to an adaptively chosen temperature selected by an effective sample size (ESS) diagnostic that monitors contraction stability; population annealing then equilibrates the system to lower temperatures.

Significance. If the central assumptions hold and the method is validated, TNPA would offer a practical route to efficient low-T sampling in frustrated systems by exploiting TN stability at moderate temperatures and PA's strength in low-T equilibration, addressing known limitations of each technique separately. The ESS diagnostic provides a physically motivated, parameter-free way to switch between the two.

major comments (2)

[Description of the ESS diagnostic and initialization procedure] The manuscript asserts that the ESS diagnostic reliably identifies a temperature at which TN output is both stable and sufficiently close to the equilibrium distribution for the subsequent PA stage to equilibrate without excessive cost or bias. However, ESS tracks sample degeneracy and contraction stability rather than deviation from true marginals; in the 2D EA model, moderate-bond-dimension TNs can still produce biased low-energy configurations due to truncation of frustrated loops, and no explicit test, bound, or numerical check is provided to confirm distributional accuracy at the switch point.
[Application section and results] Although the abstract states that the method is applied to the two-dimensional Edwards-Anderson Ising spin glass, the manuscript contains no numerical results, error bars, acceptance rates, energy histograms, or comparisons against standard PA, pure TN sampling, or known benchmarks. Without such validation, it is impossible to assess whether TNPA actually reduces computational cost or improves accuracy relative to existing approaches.

minor comments (1)

[Method overview] A pseudocode or flowchart summarizing the full TNPA workflow, including the precise criterion for ESS saturation and the PA schedule parameters, would improve reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and constructive feedback on our manuscript proposing tensor-network population annealing (TNPA). The comments correctly identify areas where additional clarification and evidence would strengthen the presentation. We address each major point below and commit to revisions that incorporate the suggestions.

read point-by-point responses

Referee: The manuscript asserts that the ESS diagnostic reliably identifies a temperature at which TN output is both stable and sufficiently close to the equilibrium distribution for the subsequent PA stage to equilibrate without excessive cost or bias. However, ESS tracks sample degeneracy and contraction stability rather than deviation from true marginals; in the 2D EA model, moderate-bond-dimension TNs can still produce biased low-energy configurations due to truncation of frustrated loops, and no explicit test, bound, or numerical check is provided to confirm distributional accuracy at the switch point.

Authors: We agree that the ESS diagnostic primarily signals contraction stability and degeneracy rather than directly bounding the total variation distance to the true marginals. Our motivation for using ESS is that, in the 2D Edwards-Anderson model, the onset of numerical instability in TN contractions empirically coincides with the regime where truncation errors on frustrated loops become significant; thus ESS serves as a practical, parameter-free proxy for when the TN initialization remains useful. Nevertheless, the referee is correct that this correlation is not rigorously proven in the current text. In the revised manuscript we will add a dedicated subsection that (i) explicitly states the limitations of ESS as a proxy, (ii) provides small-system comparisons (exact enumeration versus TN marginals at the ESS-selected temperature), and (iii) quantifies residual bias by measuring the overlap or energy difference between TN-initialized populations and fully equilibrated PA runs started from the same temperature. revision: yes
Referee: Although the abstract states that the method is applied to the two-dimensional Edwards-Anderson Ising spin glass, the manuscript contains no numerical results, error bars, acceptance rates, energy histograms, or comparisons against standard PA, pure TN sampling, or known benchmarks. Without such validation, it is impossible to assess whether TNPA actually reduces computational cost or improves accuracy relative to existing approaches.

Authors: The present manuscript is primarily a methodological proposal that introduces the hybrid TNPA framework and the ESS-based switch criterion. We acknowledge that the absence of concrete numerical benchmarks makes it difficult for readers to evaluate practical gains. In the revised version we will add a results section that applies TNPA to the 2D Edwards-Anderson spin glass on lattices up to 32x32. This section will report (a) effective sample sizes and wall-clock times versus pure PA started from infinite temperature, (b) energy histograms and acceptance rates across the annealing schedule, (c) comparisons of ground-state energies and overlap distributions against known benchmarks, and (d) direct head-to-head cost-accuracy trade-offs with both standalone TN sampling and conventional PA. Error bars obtained from multiple independent runs will be included throughout. revision: yes

Circularity Check

0 steps flagged

No circularity: algorithmic hybrid method with diagnostic-driven switch

full rationale

The paper describes an algorithmic procedure that runs tensor-network contractions only down to an adaptively chosen temperature (selected via the effective-sample-size diagnostic) and then hands off to population annealing. No equation or central claim reduces a predicted observable to a quantity fitted from the same data, no self-definitional loop appears in the temperature-selection rule, and no load-bearing uniqueness theorem is imported from the authors' prior work. The framework is therefore self-contained as a practical sampling recipe whose correctness rests on the empirical behavior of the ESS diagnostic rather than on any definitional identity.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The method relies on standard tensor-network contraction algorithms and the population-annealing resampling rule; no new free parameters are introduced beyond the usual bond dimension and population size, which are chosen by the user rather than fitted to the target observable.

axioms (1)

domain assumption Tensor-network contractions remain numerically stable and produce a distribution close to equilibrium above some temperature T_init that can be diagnosed by effective sample size.
Invoked in the description of the initialization stage.

pith-pipeline@v0.9.0 · 5450 in / 1274 out tokens · 55729 ms · 2026-05-10T15:45:08.263047+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

67 extracted references · 13 canonical work pages

[1]

1983 Transfer-matrix method 11× ∞0.0701(5)

1983
[2]

1988 Replica MC 128 2 0.071(7)

1988
[3]

1994 Exact ground-state search 36 2 0.075(2)

1994
[4]

1998 Combinatorial method 256 2 0.0709(4)

1998
[5]

2000 Flat-histogram MC 36 2 0.0709(60)

2000
[6]

2000 Exact ground-state search 20 2 0.078(5)

2000
[7]

2004 Exact partition function 50 2 0.0714(2)

2004
[8]

2012 Parallel tempering 20 2 0.0714(9) This value is consistent with previous estimates listed in Table I. Although the present estimate does not yet demonstrate a clear quantitative advantage over existing approaches, it is noteworthy that TNPA reaches system sizes comparable to those of the largest Monte Carlo- based studies and yields a value near to t...

2012
[9]

D. P. Landau and K. Binder,A Guide to Monte Carlo Simulations in Statistical Physics, 4th ed. (Cambridge University Press, 2014)

2014
[10]

M. E. J. Newman and G. T. Barkema,Monte Carlo Methods in Statistical Physics(Oxford University Press, 1999)

1999
[11]

Binder and A

K. Binder and A. P. Young, Spin glasses: Experimen- tal facts, theoretical concepts, and open questions, Rev. Mod. Phys.58, 801 (1986)

1986
[12]

Mezard, G

M. Mezard, G. Parisi, and M. Virasoro,Spin Glass Theory and Beyond(WORLD SCIENTIFIC, 1986) https://www.worldscientific.com/doi/pdf/10.1142/0271

work page doi:10.1142/0271 1986
[13]

P. G. de Gennes,Scaling Concepts in Polymer Physics (Cornell University Press, Ithaca, NY, 1979)

1979
[14]

R. H. Swendsen and J.-S. Wang, Nonuniversal critical dynamics in monte carlo simulations, Phys. Rev. Lett. 58, 86 (1987)

1987
[15]

Wolff, Collective monte carlo updating for spin sys- tems, Phys

U. Wolff, Collective monte carlo updating for spin sys- tems, Phys. Rev. Lett.62, 361 (1989)

1989
[16]

Wang and R

J.-S. Wang and R. H. Swendsen, Cluster monte carlo al- gorithms, Physica A: Statistical Mechanics and its Ap- plications167, 565 (1990)

1990
[17]

Iba, Extended ensemble monte carlo, Interna- tional Journal of Modern Physics C12, 623 (2001), https://doi.org/10.1142/S0129183101001912

Y. Iba, Extended ensemble monte carlo, Interna- tional Journal of Modern Physics C12, 623 (2001), https://doi.org/10.1142/S0129183101001912

work page doi:10.1142/s0129183101001912 2001
[18]

Or´ us, A practical introduction to tensor networks: Matrix product states and projected entangled pair states, Annals of Physics349, 117 (2014)

R. Or´ us, A practical introduction to tensor networks: Matrix product states and projected entangled pair states, Annals of Physics349, 117 (2014)

2014
[19]

Okunishi , author T

K. Okunishi, T. Nishino, and H. Ueda, Devel- opments in the tensor network — from statisti- cal mechanics to quantum entanglement, Journal of the Physical Society of Japan91, 062001 (2022), https://doi.org/10.7566/JPSJ.91.062001

work page doi:10.7566/jpsj.91.062001 2022
[20]

Levin and C

M. Levin and C. P. Nave, Tensor renormalization group approach to two-dimensional classical lattice models, Phys. Rev. Lett.99, 120601 (2007)

2007
[21]

Z. Y. Xie, H. C. Jiang, Q. N. Chen, Z. Y. Weng, and T. Xiang, Second renormalization of tensor-network 13 states, Phys. Rev. Lett.103, 160601 (2009)

2009
[22]

Z. Y. Xie, J. Chen, M. P. Qin, J. W. Zhu, L. P. Yang, and T. Xiang, Coarse-graining renormalization by higher- order singular value decomposition, Phys. Rev. B86, 045139 (2012)

2012
[23]

Evenbly and G

G. Evenbly and G. Vidal, Tensor network renormaliza- tion, Phys. Rev. Lett.115, 180405 (2015)

2015
[24]

Nishino and K

T. Nishino and K. Okunishi, Corner transfer ma- trix renormalization group method, Journal of the Physical Society of Japan65, 891 (1996), https://doi.org/10.1143/JPSJ.65.891

work page doi:10.1143/jpsj.65.891 1996
[25]

Vidal, Efficient simulation of one-dimensional quan- tum many-body systems, Phys

G. Vidal, Efficient simulation of one-dimensional quan- tum many-body systems, Phys. Rev. Lett.93, 040502 (2004)

2004
[26]

Or´ us and G

R. Or´ us and G. Vidal, Infinite time-evolving block deci- mation algorithm beyond unitary evolution, Phys. Rev. B78, 155117 (2008)

2008
[27]

A. J. Ferris, Unbiased monte carlo for the age of tensor networks (2015), arXiv:1507.00767 [cond-mat.stat-mech]

work page arXiv 2015
[28]

Huggins, C

W. Huggins, C. D. Freeman, M. Stoudenmire, N. M. Tubman, and K. B. Whaley, Monte carlo tensor network renormalization (2017), arXiv:1710.03757 [cond-mat.str- el]

work page arXiv 2017
[29]

Todo, Markov chain monte carlo in tensor network representation (2024), arXiv:2412.02974 [cond-mat.stat- mech]

S. Todo, Markov chain monte carlo in tensor network representation (2024), arXiv:2412.02974 [cond-mat.stat- mech]

work page arXiv 2024
[30]

Fr´ ıas-P´ erez, M

M. Fr´ ıas-P´ erez, M. Mari¨ en, D. P´ erez Garc´ ıa, M. C. Ba˜ nuls, and S. Iblisdir, Collective monte carlo updates through tensor network renormalization, SciPost Phys. 14, 123 (2023)

2023
[31]

T. Chen, E. Guo, W. Zhang, P. Zhang, and Y. Deng, Tensor network monte carlo simulations for the two- dimensional random-bond ising model, Phys. Rev. B 111, 094201 (2025)

2025
[32]

T. Chen, J. Liu, Y. Deng, and P. Zhang, Tensor net- work markov chain monte carlo: Efficient sampling of three-dimensional spin glasses and beyond (2025), arXiv:2509.23945 [cond-mat.stat-mech]

work page arXiv 2025
[33]

Oshima, Y

T. Oshima, Y. Arai, and K. Hukushima, Phase transitions in a modified ising spin glass model: A tensor-network-based sampling approach (2026), arXiv:2603.18486 [cond-mat.dis-nn]

work page arXiv 2026
[34]

Metropolis, A

N. Metropolis, A. W. Rosenbluth, M. N. Rosenbluth, A. H. Teller, and E. Teller, Equation of state calculations by fast computing machines, The Journal of Chemical Physics21, 1087 (1953)

1953
[35]

W. K. Hastings, Monte carlo sampling methods using markov chains and their applications, Biometrika57, 97 (1970)

1970
[36]

Wang, S.-M

C. Wang, S.-M. Qin, and H.-J. Zhou, Topologically in- variant tensor renormalization group method for the edwards-anderson spin glasses model, Phys. Rev. B90, 174201 (2014)

2014
[37]

Zhu and H

Z. Zhu and H. G. Katzgraber, Do tensor renormalization group methods work for frustrated spin systems? (2019), arXiv:1903.07721 [cond-mat.dis-nn]

work page arXiv 2019
[38]

Iba, Population Monte Carlo algorithms., Transac- tions of the Japanese Society for Artificial Intelligence 16, 279 (2001), arXiv:cond-mat/0008226 [cond-mat.stat- mech]

Y. Iba, Population Monte Carlo algorithms., Transac- tions of the Japanese Society for Artificial Intelligence 16, 279 (2001), arXiv:cond-mat/0008226 [cond-mat.stat- mech]

work page arXiv 2001
[39]

Hukushima and Y

K. Hukushima and Y. Iba, Population annealing and its application to a spin glass, AIP Conference Proceedings 690, 200 (2003)

2003
[40]

W. Wang, J. Machta, and H. G. Katzgraber, Population annealing: Theory and application in spin glasses, Phys. Rev. E92, 063307 (2015)

2015
[41]

Barzegar, C

A. Barzegar, C. Pattison, W. Wang, and H. G. Katz- graber, Optimization of population annealing monte carlo for large-scale spin-glass simulations, Phys. Rev. E 98, 053308 (2018)

2018
[42]

D. Wu, L. Wang, and P. Zhang, Solving statistical me- chanics using variational autoregressive networks, Phys. Rev. Lett.122, 080602 (2019)

2019
[43]

McNaughton, M

B. McNaughton, M. V. Miloˇ sevi´ c, A. Perali, and S. Pilati, Boosting monte carlo simulations of spin glasses using au- toregressive neural networks, Phys. Rev. E101, 053312 (2020)

2020
[44]

L. M. Del Bono, F. Ricci-Tersenghi, and F. Zamponi, Performance of machine-learning-assisted monte carlo in sampling from simple statistical physics models, Phys. Rev. E112, 045307 (2025)

2025
[45]

L. M. Del Bono, F. Ricci-Tersenghi, and F. Zamponi, Nearest-neighbors neural network architecture for effi- cient sampling of statistical physics models, Machine Learning: Science and Technology6, 025029 (2025)

2025
[46]

A. J. Bray and M. A. Moore, Lower critical dimension of ising spin glasses: a numerical study, Journal of Physics C: Solid State Physics17, L463 (1984)

1984
[47]

R. N. Bhatt and A. P. Young, Numerical studies of ising spin glasses in two, three, and four dimensions, Phys. Rev. B37, 5606 (1988)

1988
[48]

Houdayer, A cluster monte carlo algorithm for 2- dimensional spin glasses, Eur

J. Houdayer, A cluster monte carlo algorithm for 2- dimensional spin glasses, Eur. Phys. J. B22, 479 (2001)

2001
[49]

Z. Zhu, A. J. Ochoa, and H. G. Katzgraber, Efficient cluster algorithm for spin glasses in any space dimension, Phys. Rev. Lett.115, 077201 (2015)

2015
[50]

Hukushima and K

K. Hukushima and K. Nemoto, Exchange monte carlo method and application to spin glass simulations, Jour- nal of the Physical Society of Japan65, 1604 (1996), https://doi.org/10.1143/JPSJ.65.1604

work page doi:10.1143/jpsj.65.1604 1996
[51]

B. A. Berg and T. Neuhaus, Multicanonical ensemble: A new approach to simulate first-order phase transitions, Phys. Rev. Lett.68, 9 (1992)

1992
[52]

Wang, Is the broad histogram random walk dynam- ics correct?, The European Physical Journal B - Con- densed Matter and Complex Systems8, 287 (1999)

J.-S. Wang, Is the broad histogram random walk dynam- ics correct?, The European Physical Journal B - Con- densed Matter and Complex Systems8, 287 (1999)

1999
[53]

Wang and R

J.-S. Wang and R. H. Swendsen, Low-temperature prop- erties of the±jising spin glass in two dimensions, Phys. Rev. B38, 4840 (1988)

1988
[54]

Lukic, A

J. Lukic, A. Galluccio, E. Marinari, O. C. Martin, and G. Rinaldi, Critical thermodynamics of the two- dimensional±jising spin glass, Phys. Rev. Lett.92, 117202 (2004)

2004
[55]

H. F. Cheung and W. L. McMillan, Equilibrium proper- ties of the two-dimensional random (+or-j) ising model, Journal of Physics C: Solid State Physics16, 7027 (1983)

1983
[56]

Saul and M

L. Saul and M. Kardar, The 2d±j ising spin glass: exact partition functions in polynomial time, Nuclear Physics, Section B432, 641 – 667 (1994), cited by: 69

1994
[57]

J. A. Blackman, J. R. Gon¸ calves, and J. Poulter, Prop- erties of the two-dimensional random-bond±jising spin glass, Phys. Rev. E58, 1502 (1998)

1998
[58]

Z. F. Zhan, L. W. Lee, and J.-S. Wang, A new approach to the study of the ground-state properties of 2d ising spin glass, Physica A: Statistical Mechanics and its Ap- plications285, 239 (2000)

2000
[59]

A. K. Hartmann, Ground-state clusters of two-, three-, and four-dimensional±jising spin glasses, Phys. Rev. E 14 63, 016106 (2000)

2000
[60]

Perez-Morelo, A

D. Perez-Morelo, A. Ramirez-Pastor, and F. Rom´ a, Ground-state energy and entropy of the two-dimensional edwards–anderson spin-glass model with different bond distributions, Physica A: Statistical Mechanics and its Applications391, 937 (2012)

2012
[61]

A. A. Gangat and J. Gray, Hyperoptimized approxi- mate contraction of tensor networks for rugged-energy- landscape spin glasses on periodic square and cubic lat- tices, Phys. Rev. E110, 065306 (2024)

2024
[62]

D. Wu, R. Rossi, and G. Carleo, Unbiased monte carlo cluster updates with autoregressive neural networks, Phys. Rev. Res.3, L042024 (2021)

2021
[63]

E. M. Inack, S. Morawetz, and R. G. Melko, Neural an- nealing and visualization of autoregressive neural net- works in the newman–moore model, Condensed Matter 7, 10.3390/condmat7020038 (2022)

work page doi:10.3390/condmat7020038 2022
[64]

Ciarella, J

S. Ciarella, J. Trinquier, M. Weigt, and F. Zamponi, Machine-learning-assisted monte carlo fails at sampling computationally hard problems, Machine Learning: Sci- ence and Technology4, 010501 (2023)

2023
[65]

Tagliacozzo, T

L. Tagliacozzo, T. R. de Oliveira, S. Iblisdir, and J. I. La- torre, Scaling of entanglement support for matrix product states, Phys. Rev. B78, 024410 (2008)

2008
[66]

Gessert, W

D. Gessert, W. Janke, and M. Weigel, Resampling schemes in population annealing: Numerical and theo- retical results, Phys. Rev. E108, 065309 (2023)

2023
[67]

Doucet and A

A. Doucet and A. M. Johansen, A tutorial on particle filtering and smoothing : fiteen years later, inThe Ox- ford handbook of nonlinear filtering, Oxford handbooks in mathematics, edited by D. Crisan and B. Rozovskii (Oxford University Press, Oxford ; N.Y., 2011) pp. 656– 705

2011