Pith. sign in

REVIEW 4 minor 38 references

Dynamics for spherical spin glasses: disorder dependent initial conditions

T0 review · 0 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read For spherical spin glasses, starting uniformly around a conditioned critical point yields deterministic limit equations.

desk verdict First rigorous treatment of spin-glass Langevin dynamics from disorder-dependent initial conditions; Theorem 1.1 is proved carefully and holds up, and the one real caveat (concentration for critical-point counts) is disclosed by the authors themselves. read the letter →

arxiv 1908.01126 v2 pith:CEFJOKA3 submitted 2019-08-03 math.PR

Amir Dembo , Eliran Subag This is my paper · ORCID
classification math.PR MSC 82C4482C3160H1060F1560K35
keywords Langevindynamicssphericalspinglassesmixedp-spinmodelscriticalpointsKac-Riceformulaintegro-differentialequationsagingandFDTregimethermodynamiclimit
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

Spherical spin glasses are systems of N continuous spins on a sphere whose random interactions create a rugged energy landscape. This paper studies their Langevin dynamics when the initial condition is not drawn independently of the disorder, but uniformly on the sub-sphere of fixed overlap with a critical point of the Hamiltonian, conditional on that critical point having prescribed energy and gradient data. The central claim is that in the thermodynamic limit the empirical correlation, response, overlap, and energy functions become non-random and are the unique solution of a closed system of integro-differential equations (1.17)-(1.21). This matters because it turns the random, high-dimensional dynamics into a deterministic, low-dimensional transport problem, and because the same equations predict distinct long-time regimes: standard fluctuation-dissipation behavior at high temperature and, at low temperature, either escape from or localization around the initial critical point. The extension from one conditioned critical point to an average over all critical points is made through the Kac-Rice formula and is unconditional only under a concentration hypothesis on critical-point counts that the paper states is currently proved for pure m-spin models and small mixed perturbations.

What carries the argument

The load-bearing object is the conditional Gaussian field. Conditioned on $CP(E_*,G_*,\sigma)$, the Hamiltonian splits as $H_J(x)=H_{J_o}(x)-N v(N^{-1}\langle x,\sigma\rangle)$, where $J_o$ is a centered Gaussian field with the modified covariance kernel (1.24) and $v$ is the polynomial in (1.22) built from $\nu$ and its derivatives at $q_*^2$. This deterministic drift is what produces the extra terms involving $v_*$ in the limiting integro-differential equations; with $q_o=0$ those terms disappear and the system reduces to the standard closed equations for correlation and response. The second mechanism is the Kac-Rice formula, which turns the empirical average over critical points into the conditional single-point dynamics of Theorem 1.1, with the concentration property (1.28) controlling the passage from expectations to empirical averages.

What would settle it

A concrete test of Theorem 1.1 would be to simulate the conditioned Langevin dynamics for a fixed mixed potential (say $m=3$ or $m=4$) at finite large $N$, starting uniformly on $S_\sigma(q_o)$ around a critical point with $G_*>2\sqrt{\nu''(q_*^2)}$, and compare the empirical $(C_N,\chi_N,q_N,H_N)$ on $[0,T]^2$ with the numerical solution of (1.17)-(1.21); an $N$-independent discrepancy would refute the convergence claim. For Theorem 1.5, a direct falsifier is a mixed model in which the count concentration (1.28) fails, for instance if the random number of critical points in a shrinking band around $(E_*,G_*)$ shows non-negligible fluctuations at every $N$, because the unconditional convergence statement (1.29) would then have no basis.

Watch

Extended reading notes

Core claim

The paper proves Theorem 1.1: for a spherical mixed $m$-spin Hamiltonian conditioned on $CP(E_*,G_*,\sigma)$ and initial law $\mu^{q_o}_\sigma$ on the sub-sphere $S_\sigma(q_o)$, the empirical functions $(C_N,\chi_N,q^\sigma_N,H_N)$ converge uniformly on $[0,T]^2$, almost surely and in $L^p$, to the unique bounded solution of the integro-differential system (1.17)-(1.21), with $q(0)=q_o$, $C(0,0)=1$, $R(s,t)=0$ for $t>s$, and $R(s,s)=1$. The proof shows that every limit point of the pre-compact empirical processes satisfies the integral equations (3.19)-(3.31), then proves these are equivalent to the claimed differential system, and uniqueness of bounded solutions completes the identification. The paper also derives the hard-sphere limit of the conditional equations (Proposition 1.6) and, for small $\beta$, proves the large-time limit enters the FDT regime with $R_{\rm fdt}(\tau)=-2C'_{\rm fdt}(\tau)$, where $C_{\rm fdt}$ solves the scalar equation (2.3); when $q_o=0$ the whole system reduces to the standard closed correlation-response equations. Theorem 1.5 uses the Kac-Rice formula to average over critical points in a shrinking energy window, yielding convergence in expectation unconditionally and convergence in probability under the concentration hypothesis (1.28).

Load-bearing premise

For the unconditional Theorem 1.5, the load-bearing premise is the concentration property (1.28): with high probability the number of critical points in the selected energy and normal-derivative window is at least a fixed positive fraction of its expectation, and the paper explicitly says this is currently proved only for pure m-spin models and small mixed perturbations.

Editorial extensions

If this is right

  • For a single conditioned critical point, the empirical correlation, response, overlap, and energy functions become deterministic at $N\to\infty$ and obey the closed system (1.17)-(1.21).
  • If $q_o=0$, the overlap solution is identically zero and the dynamics reduce to the classical correlation-response equations; nonzero start overlap and nonzero $E_*,G_*$ add deterministic drift terms through $v_*$.
  • In the hard-sphere limit the conditional equations have a unique bounded solution, the reduced kernel $C(s,t)-q(s)q(t)/q_*^2$ is non-negative definite, and $q(s)$ stays in $[-q_*,q_*]$.
  • For small $\beta$ the long-time limit is the FDT solution of the scalar equation (2.3), while the paper's consistency analysis predicts localized no-aging states only when the Hessian-shift condition $G_*>2\sqrt{\nu''(q_*^2)}$ holds and equations (2.16)-(2.19) admit a solution.
  • If the concentration hypothesis holds, the empirical average over all critical points in a shrinking energy window converges in probability to the same deterministic dynamics as a single typical conditioned critical point.

Reading between the lines

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

  • If the concentration property (1.28) is proved for all mixed potentials, Theorem 1.5 would give a rigorous dynamical counterpart to the low-temperature band decomposition of the Gibbs measure: for times short compared with exponential-in-$N$ escapes, the path's observables are governed by the deterministic system around one band, and transitions between bands are rare.
  • The quantity $\alpha=\lim_{t\to\infty} q(t)/q_*$ in the FDT analysis offers a simulation-friendly way to distinguish escaped from localized metastable states without computing TAP free energies; the paper's equations predict localization only when $G_*$ exceeds the local-minimum threshold, which is directly testable in numerical Langevin runs.
  • Because $v_*$ depends explicitly on $(E_*,G_*)$, the paper suggests that conditioning on deeper energy levels with large normal derivative stabilizes the initial band; an implicit question not addressed is whether the convergence rate in Theorem 1.1 is uniform as $(E_*,G_*)$ vary over the allowed region.
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

0 major / 4 minor

Summary. This paper derives the thermodynamic limit of the empirical correlation, integrated response, overlap, and energy functions for the Langevin dynamics (1.1) on the sphere, starting from a disorder-dependent initial condition localized near a critical point of the random Hamiltonian. Theorem 1.1 treats a single critical point σ satisfying the conditioning event CP(E⋆,G⋆,σ), with an initial condition uniform on the sub-sphere of overlap q0 with σ; it proves uniform a.s. and Lp convergence of (CN,χN,qN,HN) on [0,T]^2 to the unique bounded solution of the coupled integro-differential system (1.17)–(1.21). The proof follows the framework of Ben Arous–Dembo–Guionnet: it establishes pre-compactness and self-averaging (Propositions 3.1 and 3.3), uses Gaussian conditioning identities (Lemma 3.7) to compute the modified drift v⋆, reduces the limit to integral equations (Propositions 3.5 and 4.1), identifies them with (1.17)–(1.21) (Proposition 3.6), and proves uniqueness (Proposition 3.4). Theorem 1.5 extends the dynamics statement to an empirical average over critical points in a Kac–Rice window, conditional on the concentration hypothesis (1.28) for the number of critical points; the paper correctly states that this hypothesis is currently proved only for pure m-spin models and small mixed perturbations. Section 6 proves the hard-sphere limit fL→1 (Proposition 1.6), and Section 7 analyzes the FDT regime (Proposition 2.1), including an exact solution for the spherical SK model.

Significance. The conditioned single-point result is the core contribution. It gives a rigorous, closed system of non-random equations for a dynamical setting in which the initial condition and the disorder are strongly correlated, which is a qualitative step beyond [10]. The derivation contains no fitted parameters: E⋆, G⋆, q⋆, and q0 are model inputs, and v⋆ is computed explicitly from the conditional Gaussian law. The paper also provides a very useful Kac–Rice bridge linking the conditioned dynamics to an average over critical points, together with the hard-constraint limit and a detailed FDT analysis. I consider the explicit disclosure of (1.28) a strength rather than a defect: the conditional status of Theorem 1.5 is clearly marked, and Theorem 1.1 is not affected. Overall, if the proofs are correct, this is a substantial and well-structured contribution suitable for publication.

minor comments (4)
  1. [Abstract and Section 1] The abstract's opening sentence, 'starting uniformly within one of the spherical bands', could be read as an unconditional statement for all mixed potentials; since the empirical-average statement of Theorem 1.5 relies on the currently unproved concentration hypothesis (1.28), I suggest adding a qualifying clause or moving the caveat more prominently into the abstract.
  2. [Section 7, proof of Proposition 2.1] The verification of hypotheses (H1)–(H2) for qo≠0 and β∈[0,β1) is delegated with 'leaving the details to the reader'; a few lines showing the exponential envelope on the set A would make the FDT proposition more self-contained.
  3. [Section 2.1, equations (2.24)–(2.31)] The derivation of the stationary solution Γ(τ) is concise; citing the relevant place in [8] at the point of (2.26) would improve readability.
  4. [Throughout] The text contains several typographical and formatting artifacts (e.g., 'W e derive' in the abstract, 'Cugliandolo-Ku rchan' in the bibliography, 'goe random matrix' near page 2); a copyedit pass is needed.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the derivation is self-contained conditional on prior independent published machinery, and the only caveat is the explicitly disclosed unproved concentration hypothesis (1.28).

full rationale

The derivation is self-contained conditional on previously published independent work. Theorem 1.1 is proved by adapting [10, Theorem 1.2], with the disorder-conditioned Gaussian calculus (Lemma 3.7), moment and concentration estimates (Propositions 3.1, 3.8, 3.10), martingale representations (Lemma 4.2), uniqueness (Proposition 3.4), and equivalence (Proposition 3.6) all proved in the paper. The conditional potential v_star is derived from Gaussian conditioning, not fitted: equations (3.32)-(3.33) compute the conditional expectation of J given CP(E,G,x_star) and yield v(r); no data-dependent parameter is matched. The limiting integro-differential system is the unique solution by Proposition 3.4, and (1.17)-(1.21) follows by Proposition 3.6 from the limit equations, so the claimed prediction does not reduce to its own inputs. Self-citations to [10] and [24] are load-bearing as technique, but those papers are independent published results whose assumptions do not include Theorem 1.1; they are not invoked as unverified authority. The only disclosed gap is the concentration hypothesis (1.28) for #C_{N,q*}(I_N,I'_N) behind Theorem 1.5; the paper explicitly states that this is 'currently proved only for pure m-spin [34] and mixed small perturbation of them [13]' and not for general mixed potentials. That is an acknowledged open hypothesis, not a circular reduction. Accordingly, no fitted input is renamed as prediction and no definition incorporates the target quantity, so the circularity score is 0.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

No fitted free parameters: model inputs are beta, nu (coefficients b_p), f, and the initial-condition parameters q*, qo, E*, G*; v*(.) is computed from Gaussian conditioning (Lemma 3.7) rather than fitted, and the phi in (1.12) is shown irrelevant (Remark 1.8). Load-bearing extra assumptions are the concentration hypothesis (1.28), unproved in general and disclosed, and the (H1)-(H2) hypotheses behind the FDT analysis, verified only for small beta. Both are stated in the text, so the ledger documents inputs rather than hidden fitting.

assumptions (7)
  • domain assumption The Gaussian potential HJ is a centered Gaussian field with covariance N nu(<x,y>/N), nu(r) = sum_{p=2}^m b_p^2 r^p, realized via (1.3)-(1.4).
    Model definition inherited from the spin-glass literature [10, 23, 24]; all results are internal to this model family, and Remark 1.8 notes the m=infinity extension is expected but not treated.
  • domain assumption The confining potential f satisfies (1.6) and (1.10); the hard-sphere limit uses fL(r) = L(r-1)^2 + phi/(4k) r^{2k} of (1.12).
    Assumptions ensure existence and uniqueness of strong solutions (Proposition 3.8) and the L to infinity limit (Proposition 1.6); standard for this program.
  • domain assumption The conditioning event CP(E*,G*,sigma) of (1.16) is interpreted as a Gaussian conditional law, conditioning on a measure-zero affine subspace.
    Footnote 1 states this interpretation explicitly; it is standard for Gaussian fields, and the Kac-Rice transfer (Theorem 1.5) gives it dynamical meaning, but the measure-zero conditioning is itself an idealization.
  • ad hoc to paper Concentration of the critical point count, assumption (1.28): #C_{N,q*}(I_N,I'_N) exceeds a fixed positive fraction of its expectation with high probability.
    Bridge from the conditioned single-point dynamics (Theorem 1.1) to the empirical average over critical points in (1.29) of Theorem 1.5. The paper states it is currently proved only for pure m-spin [34] and small mixed perturbations [13]; for general mixed models it is an open hypothesis.
  • ad hoc to paper Hypotheses (H1)-(H2) of Proposition 7.1: uniform t-integrability of {R(t+.,t)}, positive mean decay (7.5), contraction of the map Psi on a set A, and non-empty subset S with the FDT property (2.1).
    These hypotheses drive the FDT claims of Proposition 2.1. They are verified for small beta (with qo=0 falling back to [24, Section 4]); at general beta the paper only says it expects (2.1) to hold, so the general case is conditional.
  • standard math Kac-Rice formula [1, Theorem 12.1.1], upper-bound version (5.1).
    Standard counting tool; the paper proves the needed upper bound directly (Proposition 5.1) and uses uniform integrability of the Hessian determinant (Lemma 5.3).
  • domain assumption Physics ansatz for aging contributions: single aging regime with R_aging(lambda) = A C'_aging(lambda), starting at C_aging(1)=D_infinity and ending at C_aging(0)=alpha^2 (equation (2.10)).
    Used only in the heuristic part of Section 2 for beta >= beta_c, not in any theorem; explicitly labeled as the 'physics ansatz', with the caveat that if self-consistency fails one moves to multiple aging regimes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Dynamics for spherical spin glasses: disorder dependent initial conditions." pith.science (2026). https://pith.science/paper/CEFJOKA3

@misc{pith2026190801126,
  author       = {Pith},
  title        = {Pith review of: Dynamics for spherical spin glasses: disorder dependent initial conditions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CEFJOKA3}},
  note         = {Machine review of arXiv:1908.01126}
}
abstract

We derive the thermodynamic limit of the empirical correlation and response functions in the Langevin dynamics for spherical mixed $p$-spin disordered mean-field models, starting uniformly within one of the spherical bands on which the Gibbs measure concentrates at low temperature for the pure $p$-spin models and mixed perturbations of them. We further relate the large time asymptotics of the resulting coupled non-linear integro-differential equations, to the geometric structure of the Gibbs measures (at low temperature), and derive their FDT solution (at high temperature).

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 37 canonical work pages

  1. [10]

    DEMBO, A

    BEN AROUS, G. ; DEMBO, A. ; GUIONNET, A. ; Cugliandolo-Ku rchan equations for dynamics of Spin-Glasses. Probab. Theory Relat. Fields , 136:619-660 (2006)

    work page 2006

  2. [1]

    ADLER, R. J. ; TAYLOR, J. E.; Random fields and geometry . Springer Monographs in Mathematics. Springer, New York, 2007

    work page 2007

  3. [2]

    W.; The integral of a symmetric unimodal fun ction over a symmetric convex set and some probability inequalities

    ANDERSON, T. W.; The integral of a symmetric unimodal fun ction over a symmetric convex set and some probability inequalities. Proc. Amer. Math. Soc. 6:170–176 (1955)

    work page 1955

  4. [3]

    et altri; Sur les in´ egalit´ es de Sobolev logarithmiques

    AN ´E, C. et altri; Sur les in´ egalit´ es de Sobolev logarithmiques. Panoramas et Syntheses , 10, Soci´ et´ e Math´ ematique de France (2000)

    work page 2000

  5. [4]

    BEN AROUS, G

    AUFFINGER, A. ; BEN AROUS, G. ; Complexity of random smoot h functions on the high-dimensional sphere. Ann. Probab. 41(6):4214–4247 (2013)

    work page 2013

  6. [5]

    BEN AROUS, G

    AUFFINGER, A. ; BEN AROUS, G. ; ˇCERN ´Y, J. ; Random matrices and complexity of spin glasses. Comm. Pure Appl. Math. 66(2):165–201 (2013)

    work page 2013

  7. [6]

    BURIONI, R

    BARRAT, A. ; BURIONI, R. ; M ´EZARD, M. Dynamics within metastable states in a mean-field s pin glass. J. Phys. A: Math. Gen. , 29:L81–L87 (1996)

    work page 1996

  8. [7]

    FRANZ, S

    BARRAT, A. ; FRANZ, S. Basins of attraction of metastable states of the spherical p-spin model. J. Phys. A: Math. Gen. , 31:L119–L127 (1998)

    work page 1998

Show all 38 references
  1. [8]

    DEMBO, A

    BEN AROUS, G. ; DEMBO, A. ; GUIONNET, A. Aging of spherical spin glasses. Probab. Theory Relat. Fields , 120:1–67 (2001). 38 AMIR DEMBO AND ELIRAN SUBAG

    2001

  2. [9]

    Aging and spin-glass dynamics

    BEN AROUS, G. ; Aging and spin-glass dynamics. Proceedings of the International Congress of Mathematicia ns, Vol. III , 3–14, Higher Ed. Press, Beijing, 2002 (2002)

    2002

  3. [11]

    GHEISSARI, R

    BEN AROUS, G. ; GHEISSARI, R. ; JAGANNATH, A. ; Bounding fl ows for spherical spin glass dynamics. arXiv:1808.00929, (2018)

    2018 arXiv

  4. [12]

    JAGANNATH, A

    BEN AROUS, G. ; JAGANNATH, A. ; Spectral gap estimates in mean field spin glasses. Commun. Math. Phys. , 361:1-52 (2018)

    2018

  5. [13]

    SUBAG, E

    BEN AROUS, G. ; SUBAG, E. ; ZEITOUNI, O. ; Geometry and tem perature chaos in mixed spherical spin glasses at low temperature – the perturbative regime. Commun. Pure Appl. Math. , to appear

  6. [14]

    BIROLI, G.; Theoretical perspective on t he glass transition and amorphous materials

    BERTHIER, L. ; BIROLI, G.; Theoretical perspective on t he glass transition and amorphous materials. Reviews of Modern Physics, 83:587- (2011)

    2011

  7. [15]

    CUGLIANDOLO, L.F

    BOUCHAUD, J.P. ; CUGLIANDOLO, L.F. ; KURCHAN, J. ; MEZAR D, M. ; Out of equilibrium dynamics in spin-glasses and other glassy systems. Spin glass dynamics and Random Fields , A. P Young Editor (1997)

    1997

  8. [16]

    GIANNOPOULOS, A

    BRAZITIKOS, S. ; GIANNOPOULOS, A. ; V ALETTAS, P. ; VRITS IOU, B.-H. ; Geometry of Isotropic Convex Bodies . Mathematical surveys and monographs, 196 (2014)

    2014

  9. [17]

    The Aizenman-Sims-Starr scheme and Paris i formula for mixed p-spin spherical models

    CHEN, W.-K. ; The Aizenman-Sims-Starr scheme and Paris i formula for mixed p-spin spherical models. Electron. J. Probab. 18, No. 94, 14 (2013)

    2013

  10. [18]

    PANCHENKO, D

    CHEN, W.-K. ; PANCHENKO, D. ; SUBAG, E. ; The generalized TAP free energy. arXiv:1812.05066, (2018)

    2018 arXiv

  11. [19]

    PANCHENKO, D

    CHEN, W.-K. ; PANCHENKO, D. ; SUBAG, E. ; The generalized TAP free energy II. arXiv:1903.01030, (2019)

    2019 arXiv

  12. [20]

    Zeitschrift fur Physik B Condensed Matter , 92:257271 (1993)

    CRISANTI, A.; HORNER, H.; SOMMERS, H-J; The spherical p -spin interaction spin-glass model. Zeitschrift fur Physik B Condensed Matter , 92:257271 (1993)

    1993

  13. [21]

    CRISANTI, A.; SOMMERS, H-J; Thouless-Anderson-Palme r approach to the spherical p-spin spin glass model. J. Phys. I France, 5(7):805-813 (1995)

    1995

  14. [22]

    Course 7: Dynamics of glassy systems

    CUGLIANDOLO, L.F. Course 7: Dynamics of glassy systems . In Slow Relaxations and nonequilibrium dynamics in condensed matter, 367521. Springer (2003)

    2003

  15. [23]

    KURCHAN, J

    CUGLIANDOLO, L.F. ; KURCHAN, J. ; Analytical solution o f the off-equilibrium Dynamics of a Long-Range Spin-Glass Model.Phys. Rev. Lett. 71, 173 (1993)

    1993

  16. [24]

    GUIONNET, A

    DEMBO, A. ; GUIONNET, A. ; MAZZA, C.; Limiting dynamics f or spherical models of spin glasses at high temperature. J. Stat. Phys. 126: 781-816 (2007)

    2007

  17. [25]

    arXiv:1903.01421, (2019)

    FOLENA, G.; FRANZ, S.; RICCI-TERSENGHI, F.; Memories f rom the ergodic phase: the awkward dynamics of spherical mixed p-spin models. arXiv:1903.01421, (2019)

    2019 arXiv

  18. [26]

    JAGANNATH, A

    GHEISSARI, R. ; JAGANNATH, A. ; On the spectral gap of sph erical spin glass dynamics. Ann. Henri Poincar´ e (B), to appear

  19. [27]

    MAZZA, C.; Long time behaviour of non-com mutative processes solution of a linear differential equati on

    GUIONNET, A. ; MAZZA, C.; Long time behaviour of non-com mutative processes solution of a linear differential equati on. Prob. Theory. Rel. Fields 131:493-518 (2005)

    2005

  20. [28]

    Proceedings of the Ascona meeting 2004

    GUIONNET, A.; Dynamics for spherical models of sping gl ass and Aging. Proceedings of the Ascona meeting 2004

    2004

  21. [29]

    PARISI, G

    KURCHAN, J. ; PARISI, G. ; VIRASORO, M. A. ; Barriers and m etastable states as saddle points in the replica approach. J. Phys. I France , 3:1819-1838 (1993)

    1993

  22. [30]

    PARISI, G

    M ´EZARD, M. ; PARISI, G. ; SOURLAS, N. ; TOULOUSE, G. ; VIRASORO, M. A. ; Nature of the spin-glass phase. Phys. Rev. Lett., 52(13):1156–1159 (1984)

    1984

  23. [31]

    PARISI, G

    M ´EZARD, M. ; PARISI, G. ; VIRASORO, M. A. ; Solution of ‘Solvabl e model of a spin glass’. Spin Glass Theory and Beyond, W orld Scientific (1987)

    1987

  24. [32]

    PANCHENKO, D.; The Parisi ultrametricity conjecture. Ann. of Math. (2), 177(1):383–393 (2013)

    2013

  25. [33]

    arXiv:1804.10576

    SUBAG, E.; Free energy landscapes in spherical spin gla sses. arXiv:1804.10576. (2018)

    2018 arXiv

  26. [34]

    SUBAG, E.; The complexity of spherical p-spin models—A second moment approach. Ann. Probab. 45(5):3385-3450 (2017)

    2017

  27. [35]

    SUBAG, E.; The geometry of the Gibbs measure of pure sphe rical spin glasses. Invent. Math. , 210(1):135-209 (2017)

    2017

  28. [36]

    TALAGRAND, M.; Free energy of the spherical mean field mo del. Prob. Th. Rel. Fields 134:339-382 (2006)

    2006

  29. [37]

    TALAGRAND, M.; Construction of pure states in mean field models for spin glasses. Prob. Th. Rel. Fields 3-4(148):601-643 (2010)

    2010

  30. [38]

    THOULESS, D. J. ; ANDERSON, P. W. ; PALMER, R. G. ; Solutio n of ‘Solvable model of a spin glass’. Philosophical Magazine, 35(3):593-601 (1977). Department of Statistics and Department of Mathematics, St anford University, Stanford, CA 94305. E-mail address : adembo@stanford.e...

    1977

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.