Self-averaging of perturbation Hamiltonian density in perturbed spin systems

C. Itoi

REVIEW 2 major objections 4 minor 33 references

Self-averaging of perturbation Hamiltonian density in perturbed spin systems

T0 review · 2 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 The paper proves that perturbation Hamiltonian densities are self-averaging in disordered quantum spin systems: their variance vanishes in the infinite-volume limit for almost all perturbation strengths.

desk verdict The paper's main theorem is plausible and the RSB corollary is genuinely new, but the proof of Lemma 2.3 uses an invalid limsup/integral interchange that leaves Theorem 1.1 unproven as written. read the letter →

arxiv 1908.09423 v2 pith:EMYE5AR5 submitted 2019-08-26 math-ph cond-mat.stat-mechmath.MPquant-ph

C. Itoi This is my paper · ORCID

classification math-phcond-mat.stat-mechmath.MPquant-ph MSC 82B2082B2682B44

keywords self-averagingquencheddisorderquantumspinsystemsreplicasymmetrybreakingoverlapperturbationHamiltoniandensityDuhamelproductdisordered

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

This paper proves that, in a broad class of quantum spin systems with quenched disorder, the density of any bounded perturbation is self-averaging: the combined thermal and disorder variance of its expectation vanishes as the system grows. The argument requires only that the quenched free-energy density has an infinite-volume limit and that the double commutator $[O_N,[H(S,J),O_N]]$ of the perturbation operator with the Hamiltonian vanishes in the thermodynamic limit, both checked for many short-range models. A corollary says that if the infinite-volume limit and the switched-off limit of a replica-symmetry-breaking perturbation commute, then no replica symmetry breaking occurs in the replica-symmetric Gibbs state, because the variance of every overlap order parameter must vanish. This matters because it converts a physically expected phenomenon into a theorem proved with a shorter route and fewer assumptions than earlier treatments.

What carries the argument

Two ingredients carry the argument. First, convexity of $\psi_N(\beta,\lambda,J)=N^{-1}\log Z_N(\beta,\lambda,J)$ and of its disorder average $p_N$, together with convexity of the infinite-volume limit $p$, forces the derivative $\partial_\lambda \psi_N$ to converge to $\partial_\lambda p$ for almost all $\lambda$; this is where Assumption 1 enters. Second, the Duhamel product $(O_1,\ldots,O_k)_{N,\lambda} = \int_{[0,1]^k} dt_1\cdots dt_k \langle T[O_1(t_1)\cdots O_k(t_k)]\rangle_{N,\lambda}$, whose derivative relation gives $\partial_\lambda \langle O_N\rangle_{N,\lambda}=N\beta[(O_N,O_N)_{N,\lambda}-\langle O_N\rangle_{N,\lambda}^2]$, is bounded above and below by the Gibbs expectation of $O_N^2$ through a double-commutator inequality. The correction term in that inequality is the thermal expectation of $[O_N,[H_\lambda,O_N]]$, which Assumption 2 makes vanish in the thermodynamic limit.

What would settle it

Compute the double commutator $\|[O_N,[H(S,J),O_N]]\|$ for an overlap operator $O_N$ built from powers of spin overlaps in a short-range disordered model. If this norm does not tend to zero, Assumption 2 fails and the theorem's conclusion is not guaranteed for that operator; alternatively, simulate finite systems and test whether $\mathbb{E}\langle(O_N-\mathbb{E}\langle O_N\rangle_{N,\lambda})^2\rangle_{N,\lambda}$ fails to vanish even when Assumptions 1 and 2 are satisfied.

Watch

Extended reading notes

Core claim

The central discovery is Theorem 1.1: for the perturbed Hamiltonian $H_\lambda = H_N(S,J) - N\lambda O_N$ with bounded self-adjoint $O_N$, under Assumptions 1 and 2, the limit $\lim_{N\to\infty}\mathbb{E}\langle O_N\rangle_{N,\lambda}$ exists for almost all $\lambda\in\mathbb{R}$, and the variance vanishes, $\lim_{N\to\infty}\mathbb{E}\langle (O_N-\mathbb{E}\langle O_N\rangle_{N,\lambda})^2\rangle_{N,\lambda}=0$. The theorem immediately gives $\lim_{N\to\infty}\mathbb{E}\langle O_N^2\rangle_{N,\lambda} = (\lim_{N\to\infty}\mathbb{E}\langle O_N\rangle_{N,\lambda})^2$, meaning the perturbation operator is self-averaging. In the replica-symmetric setting this implies that the variance of any replica-symmetry-breaking (RSB) order parameter vanishes once spontaneous RSB is absent, and it identifies spontaneous symmetry breaking as the only mechanism that can produce a finite order-parameter variance in the infinite-volume limit.

Load-bearing premise

The load-bearing premise is Assumption 2, that the double commutator $\|[O_N,[H(S,J),O_N]]\|$ vanishes in the thermodynamic limit for fixed disorder; it is verified for short-range Hamiltonians with spin-density perturbations, but not proved for the replica-symmetry-breaking overlap operators considered in the corollary.

Editorial extensions

If this is right

For any bounded symmetry-breaking order operator satisfying the assumptions, the law of large numbers applies to the order parameter measured in the perturbed Gibbs state, so the measured value can be identified with its expectation.
Spontaneous symmetry breaking manifests as non-commutativity of the limits $N\to\infty$ and $\lambda\to 0$; a non-vanishing variance in the symmetric Gibbs state coexists with a vanishing variance in the perturbed state.
In replica-symmetric Gibbs states of disordered quantum spin models, absence of spontaneous replica symmetry breaking forces the variance of every RSB order operator to vanish, so non-spontaneous RSB is impossible.
The result strengthens the earlier zero-variance theorem by proving the needed vanishing of disorder fluctuations rather than assuming it, under weaker conditions.

Reading between the lines

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

Assumption 2 is the natural boundary of the method: for overlap-type operators that are not short-range spin densities, the double commutator may fail to vanish, and then the corollary's reach would stop exactly at the observed spin-glass phase.
A testable extension is to compute $\|[O_N,[H,O_N]]\|$ numerically for powers of the spin overlap in finite-size mean-field spin glasses; a nonzero limit would separate models where this theorem applies from models where the variance may survive.
The same convexity-plus-double-commutator scheme should generalize to any extensive observable with vanishing double commutator, suggesting a broader self-averaging principle for quenched disorder beyond spin systems.

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.

Desk Editor's Note

read the letter

Two things to know. The paper proves self-averaging of perturbation Hamiltonian densities under weak assumptions, and the RSB corollary is a real extension to quantum disordered systems. But the proof of Lemma 2.3 has a gap that matters: equation (58) interchanges a limsup with an integral without pointwise convergence, and that interchange is false in general. As written, Theorem 1.1 is not proven.

What is new and good: the paper upgrades the author's earlier result [19] by proving the variance bound as a lemma instead of assuming it, under weaker hypotheses (Assumptions 1 and 2). The argument is mostly self-contained and follows from convexity of the free energy plus the Duhamel product structure. The RSB application, though conditional, is a serious attempt to export Chatterjee's no-RSB argument to the quantum setting. The paper cites relevant literature properly; the self-citations are to the work being improved.

The real problem is in Lemma 2.3. After integrating the derivative identity they get ∫ f_N → 0 with f_N ≥ 0 and uniformly bounded. Then they claim Lebesgue's dominated convergence gives ∫ limsup f_N = lim ∫ f_N. That requires pointwise convergence of f_N, which hasn't been established. Bounded nonnegative sequences with ∫ f_N → 0 can have limsup equal to 1 almost everywhere (the sliding-interval example), so (58) is not a consequence of anything written. The subsequent a.e. limit (59) and the Harris inequality step depend on it. This is a load-bearing gap, not a notation slip.

There are two smaller issues. In Lemma 2.1, (33) uses σ² as if it were the per-term variance when the assumption only bounds the sum of variances; the final bound is fine if one interprets σ² as Var(J_m) and then sums. And in Section 3.2, Assumption 2 is never verified for overlap operators like (83); the corollary is conditional on a property the paper does not check, though it is plausible for short-range models.

The main theorem is probably true, and the proof strategy is sound enough that the gap may be repairable—perhaps by using convexity of p_N and boundedness of p_N' to get the a.e. limit by a different route. But the repair is not in the paper. This paper is for specialists in rigorous statistical mechanics of disordered spin systems. It deserves a serious referee, but it should go back for major revision, with instructions to focus on Lemma 2.3. I would not desk reject it, but I would not accept it in this form.

Referee Report

2 major / 4 minor

Summary. The paper studies quantum spin systems with quenched disorder and a perturbation Hamiltonian of the form H_λ = H_N(S,J) − Nλ O_N. Under Assumption 1 (existence of the infinite-volume free-energy density p(β,λ) = lim_N p_N(β,λ)) and Assumption 2 (vanishing of the double commutator lim_N ||[O_N,[H(S,J),O_N]]|| = 0), Theorem 1.1 claims that for almost all λ the expectation E⟨O_N⟩_{N,λ} converges and the combined thermal-and-disorder variance E⟨(O_N − E⟨O_N⟩_{N,λ})²⟩_{N,λ} vanishes as N → ∞. The proof has three lemmas: Lemma 2.1 gives a martingale-difference bound on the disorder variance of ψ_N; Lemma 2.2 uses convexity to show E⟨O_N⟩_{N,λ} converges to β^{-1}∂p/∂λ a.e. and that the sample-to-sample fluctuation of the Gibbs expectation vanishes; Lemma 2.3 aims to show the Gibbs thermal variance vanishes using Duhamel products and Harris' inequality. The paper then applies Theorem 1.1 to spontaneous symmetry breaking and, in Section 3.2, to replica symmetry breaking, obtaining Corollary 3.3 that absence of spontaneous RSB implies absence of RSB in Chatterjee's sense.

Significance. If the main theorem were correct, it would provide a clean and quite general proof of self-averaging of order-parameter-type perturbations under weak assumptions, and Corollary 3.3 would be a notable structural result connecting spontaneous and non-spontaneous replica symmetry breaking. The paper is self-contained, has no fitted parameters, and the variance bound in Lemma 2.1 is an elegant and correct martingale argument. The convexity argument in Lemma 2.2 is also sound in its main lines. However, the proof of Lemma 2.3 contains a false application of dominated convergence that invalidates the central theorem as written, so the significance of the contribution can only be assessed after a substantial repair.

major comments (2)

[§2, Lemma 2.3, Eq. (58)] Equation (58) asserts ∫_{λ'}^{λ''} limsup_{N→∞} f_N(λ) dλ = lim_{N→∞} ∫_{λ'}^{λ''} f_N(λ) dλ = 0, with f_N(λ) = E[(O_N,O_N)_{N,λ} − ⟨O_N⟩_{N,λ}²]. This equality is not justified. From (56) and the uniform bound (57) one can only conclude that f_N → 0 in L¹ on the interval, which gives convergence of a subsequence almost everywhere; Fatou's lemma gives liminf f_N = 0 a.e., not limsup f_N = 0 and not the pointwise convergence asserted in (59). For example, on [0,1] take f_N to be the indicator of an interval of length 1/N shifted so that every point is covered infinitely often; then ∫ f_N = 1/N → 0 but limsup f_N = 1 everywhere and no pointwise limit exists. Since (59) is the input to Harris' inequality that yields the vanishing Gibbs variance (61), the proof of Theorem 1.1 is incomplete. A repair would require a genuinely new argument, for instance using the monotonicity of E⟨O_N⟩_{N,λ} in λ together with the almost-everywhere convergence from Lemma 2.2, but no such argument is present.
[§3.2, Corollary 3.3 and the note after it] Corollary 3.3 applies Theorem 1.1 with perturbation operator O_N = R, where R is a general RSB order operator of the form (83). Theorem 1.1 requires Assumption 2, namely lim_{N→∞} ||[R,[H(S,J),R]]|| = 0. The paper verifies Assumption 2 only for spin-density perturbations such as (4); for the overlap operators in (75) and their powers in (83), no verification is supplied. The closing note only states that Assumption 1 can be proven when the set A in (83) is {1} and then says 'Corollary 3.3 is valid under Assumption 1,' which appears to conflate the two assumptions. Without a proof of (10) for the overlap class, or an explicit restriction of the corollary to operators for which Assumption 2 holds, the replica-symmetry application is not established as stated.

minor comments (4)

[§2, Eq. (33)] In Eq. (33) the notation σ² is used as if it were the variance of a single random variable J_m, whereas in the hypothesis of Lemma 2.1 σ² bounds the total variance sum ∑_m Var(J_m) ≤ σ²N. The correct per-term bound is 2β²C_φ² Var(J_m)/N², and the final bound (34) follows after summing over m.
[§2, proof of Lemma 2.2] The phrase 'the convex function p(λ) is continuously differentiable almost everywhere' is imprecise: a convex function is differentiable almost everywhere and its derivative is continuous almost everywhere. The limiting argument should be phrased as 'for every λ at which p′ is continuous, which is almost every λ, the last term vanishes as ε → 0.'
[§3.2, Note 3.2 and Corollary 3.3] The definition of absence of spontaneous RSB in Note 3.2 is stated only for the commutativity of the limits of E⟨R⟩_{N,λ}, but the proof of Corollary 3.3 also uses commutativity for E⟨R²⟩_{N,λ}. Please state explicitly that R² again belongs to the class of RSB perturbation operators (83), or add this as a hypothesis to the corollary.
[Throughout] The manuscript contains several typographical errors and stylistic slips, including 'v anishes' in the abstract, 'i nfinite' in Section 1, 'Chattejee' in Section 3.2, and 'law of large numberss' in Section 3.1. These should be corrected carefully.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the derivation is self-contained given Assumptions 1 and 2.

full rationale

The paper's main theorem is proven from Assumptions 1 and 2 through Lemmas 2.1, 2.2, and 2.3, and none of these steps reduces to the target conclusion. Lemma 2.1 bounds the disorder variance of the free-energy density by a self-contained martingale/Jensen argument; Lemma 2.2 derives a.e. convergence of E<O_N> and vanishing of the disorder-thermal mixed variance from convexity and Lemma 2.1; Lemma 2.3 obtains vanishing Gibbs variance from the derivative identity, Harris' inequality, and Assumption 2. The perturbation operator is not defined in terms of the variance that is being proved, and Assumption 2 is a commutator condition strictly weaker than the conclusion. The self-citations [18,19] are used only as background or as prior results for special models, and are not load-bearing: the proof explicitly does not assume the vanishing variance that [19] assumed, but proves it as Lemma 2.1. In the replica-symmetry-breaking application, Corollary 3.3 is a valid inference from Theorem 1.1 plus the definition of commutativity of limits, not a restatement of the definition. The skeptical concern about the dominated-convergence step in Lemma 2.3 (Eq. 58) is a mathematical correctness/reparability question, not a circularity, and is therefore outside the scope of this pass.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central claim rests on two explicitly stated assumptions: existence of the free energy density (Assumption 1) and vanishing of the double commutator (Assumption 2). These are domain assumptions, not derived in the paper. The proof also uses standard convexity and Harris inequality. No free parameters or invented entities.

assumptions (4)

domain assumption Assumption 1: the infinite-volume limit p(β,λ)=lim p_N(β,λ) exists for each (β,λ)
Used in Lemma 2.2 to compare ψ_N and p; cited as proven for several models, not proven here.
domain assumption Assumption 2: lim ||[O_N,[H(S,J),O_N]]||=0 for fixed J
Used in Lemma 2.3 via Harris inequality to control Gibbs variance. This is the most restrictive assumption and is not verified for general RSB operators.
standard math Convexity of log-partition function in the perturbation parameter
Used in Lemma 2.2 to bound derivatives by finite differences.
standard math Harris/Bogolyubov inequality for Duhamel two-point function
Used in Lemma 2.3 to relate Duhamel variance to Gibbs variance; quoted from [3,17].

how reviews work

0 comments

Cite this review

Pith. "Pith review of Self-averaging of perturbation Hamiltonian density in perturbed spin systems." pith.science (2026). https://pith.science/paper/EMYE5AR5

@misc{pith2026190809423,
  author       = {Pith},
  title        = {Pith review of: Self-averaging of perturbation Hamiltonian density in perturbed spin systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EMYE5AR5}},
  note         = {Machine review of arXiv:1908.09423}
}

read the original abstract

It is shown that the variance of a perturbation Hamiltonian density vanishes in the infinite-volume limit of the perturbed spin systems with quenched disorder. This is proven in a simpler way and under less assumptions than before. A corollary of this theorem indicates the impossibility of non-spontaneous replica symmetry-breaking in disordered spin systems. The commutativity between the infinite-volume limit and the switched-off limit of a replica symmetry-breaking perturbation implies that the variance of the spin overlap vanishes in the replica symmetric Gibbs state.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 33 canonical work pages

[19]

:Zero variance of perturbation Hamiltonian de nsity in perturbed spin systems

Itoi, C. :Zero variance of perturbation Hamiltonian de nsity in perturbed spin systems. J. Stat. phys. 176 556-573 (2019)

work page 2019
[1]

: On the stability of quenched state in mean-ﬁeld spin glass models

Aizenman, M., Contucci, P. : On the stability of quenched state in mean-ﬁeld spin glass models. J. Stat. Phys. 92, 765-783(1997)

work page 1997
[2]

W.:Operator algebras and qua ntum statistical mechanics

Bratteli, O., Robinson, D. W.:Operator algebras and qua ntum statistical mechanics. Springer (1987)

work page 1987
[3]

Brankov, J. G., N.S. Tonchev, T. S.: Generalized inequal ities for the Bogoliubov-Duhamel in- ner product with applications in the Approximating Hamilto nian Method. Cond. Matt. Phys. 14,13003-1-13003-17(2011)

work page 2011
[4]

: Absence of replica symmetry-breaking i n the random ﬁeld Ising model

Chatterjee, S. : Absence of replica symmetry-breaking i n the random ﬁeld Ising model. Commun. Math .Phys. 337, 93-102(2015)

work page 2015
[5]

The Ghirlanda-Guerra identities without averaging

Chatterjee,S.: The Ghirlanda-Guerra identities witho ut averaging. preprint, arXiv:0911.4520 (2009)

work page Pith review arXiv 2009
[6]

: Spin-glass stochastic sta bility: A rigorous proof

Contucci, P., Giardin` a, C. : Spin-glass stochastic sta bility: A rigorous proof. Annales Henri Poincare, 6, 915-923, (2005)

work page 2005
[7]

: The Ghirlanda-Guerra iden tities

Contucci, P., Giardin` a, C. : The Ghirlanda-Guerra iden tities. J. Stat. Phys. 126, 917-931,(2007)

work page 2007

Show all 33 references

[8]

Contucci, P., Lebowitz, J. L. : Correlation inequalitie s for quantum spin systems with quenched centered disorder. J. Math. Phys. 51, 023302-1 -6 (2010)

2010
[9]

: Thermodynamics and universality for mean ﬁeld quantum spin glasses

Crawford, N. : Thermodynamics and universality for mean ﬁeld quantum spin glasses. Commun. Math. Phys. 274, 821-839(2007)

2007
[10]

J., Lieb, E

Dyson, F. J., Lieb, E. H., Simon, B.:Phase transition in quantum spin systems with isotropic and nonisotropic interactions J. Stat. Phys. 18335-383 (1978)

1978
[11]

F., Anderson, P

Edwards,S. F., Anderson, P. W. : Theory of spin glasses J . Phys. F: Metal Phys. 5, 965-974(1975)

1975
[12]

: General properties of overl ap probability distributions in disordered spin systems

Ghirlanda, S., Guerra, F. : General properties of overl ap probability distributions in disordered spin systems. Towards Parisi ultrametricity. J. Phys. A 31, 9149-9155(1998) 12

1998
[13]

Goldschmidt, C., Ueltschi, D., Windridge, P:Quantum H eisenberg models and their probabilistic representations Entropy and the quantum II, Contemp. Math. 562 177-224, (2011)

2011
[14]

Griﬃths, R. B. : Spontaneous magnetization in idealize d ferromagnets. Phys.Rev. 152 , 240-246, (1964)

1964
[15]

Griﬃths, R. B. : A proof that the free energy of spin syste m is extensive. J. Math. Phys. 5 , 1215-1222, (1964)

1964
[16]

Guerra, F.:The phenomenon of spontaneous replica symm etry-breaking in complex statistical mechanics systems. J. Phys: Conf. Series 442 012013(2013)

2013
[17]

:Bounds for certain thermodynamic averag esJ

Harris, A.B. :Bounds for certain thermodynamic averag esJ. Math. Phys. 8 1044-1045.(1967)

1967
[18]

:General properties of overlap operators in di sordered quantum spin systems

Itoi, C. :General properties of overlap operators in di sordered quantum spin systems. J. Stat. phys. 163 1339-1349 (2016)

2016
[20]

M., Sinai, Ya

Khanin, K. M., Sinai, Ya. G. :Existence of free energy fo r models with Long-range random Hamiltonians. J. Stat. Phys. 20 573-584 (1979)

1979
[21]

Symmetry breaking in Heisenberg an tiferromagnets

Koma, T., Tasaki, H. Symmetry breaking in Heisenberg an tiferromagnets. Commun. Math. Phys. 158, 198-214 (1993)

1993
[22]

:Pressure and variational principle fo r random Ising model

Ledrappier, F. :Pressure and variational principle fo r random Ising model. Commun. Math. Phys. 56, 297-302 (1977)

1977
[23]

:Non-diﬀerentiability of the eﬀective poten tial and the replica symmetry-breaking in the random energy model J

Mukaida, H. :Non-diﬀerentiability of the eﬀective poten tial and the replica symmetry-breaking in the random energy model J. Phys. A 49 45002,1-15 (2016)

2016
[24]

: The Ghirlanda-Guerra identities for mi xed p-spin glass model

Panchenko, D. : The Ghirlanda-Guerra identities for mi xed p-spin glass model. Compt. Read. Math. 348, 189-192(2010)

2010
[25]

:A sequence of approximate solutions to the S -K model for spin glasses

Parisi, G. :A sequence of approximate solutions to the S -K model for spin glasses. J. Phys. A 13, L-115 (1980)

1980
[26]

A¿, Figotin, A

Pastur, L. A¿, Figotin, A. L. :Theory of disordered spin systems. Theor. Math. Phys. 35, 403-414 (1978)

1978
[27]

: Nelson’s symmetry and all that in Yukawa and (φ4)3 theories

Seiler, E., Simon, B. : Nelson’s symmetry and all that in Yukawa and (φ4)3 theories. Ann. Phys. 97, 470-518, (1976)

1976
[28]

Sherrington, S., Kirkpatrick, S : Solvable model of spi n glass. Phys. Rev. Lett. 35, 1792-1796, (1975)

1975
[29]

: The Parisi formula

Talagrand, M. : The Parisi formula. Ann. Math. 163, 221-263 (2006)

2006
[30]

: Mean ﬁeld models for spin glasses

Talagrand, M. : Mean ﬁeld models for spin glasses. Sprin ger, Berlin (2011)

2011
[31]

van Enter, A. C. D., Griﬃths, R. B.:The order parameter i n a spin glass. Commun. Math. Phys. 90, 319-327 ( 1983)

1983
[32]

Vuillermot, P. A. :Thermodynamics of quenched random s pin systems, and application to the problem of phase transitions in magnetic (spin) glasses. J. Phys. A 10, 1319-1333 (1977)

1977
[33]

Zegarlinski, B.:Interactions and pressure functiona ls for disordered lattice systems. Commun. Math. Phys. 139 305-339 (1991) 13

1991

Pith tools

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

[1] [19]

:Zero variance of perturbation Hamiltonian de nsity in perturbed spin systems

Itoi, C. :Zero variance of perturbation Hamiltonian de nsity in perturbed spin systems. J. Stat. phys. 176 556-573 (2019)

work page 2019

[2] [1]

: On the stability of quenched state in mean-ﬁeld spin glass models

Aizenman, M., Contucci, P. : On the stability of quenched state in mean-ﬁeld spin glass models. J. Stat. Phys. 92, 765-783(1997)

work page 1997

[3] [2]

W.:Operator algebras and qua ntum statistical mechanics

Bratteli, O., Robinson, D. W.:Operator algebras and qua ntum statistical mechanics. Springer (1987)

work page 1987

[4] [3]

Brankov, J. G., N.S. Tonchev, T. S.: Generalized inequal ities for the Bogoliubov-Duhamel in- ner product with applications in the Approximating Hamilto nian Method. Cond. Matt. Phys. 14,13003-1-13003-17(2011)

work page 2011

[5] [4]

: Absence of replica symmetry-breaking i n the random ﬁeld Ising model

Chatterjee, S. : Absence of replica symmetry-breaking i n the random ﬁeld Ising model. Commun. Math .Phys. 337, 93-102(2015)

work page 2015

[6] [5]

The Ghirlanda-Guerra identities without averaging

Chatterjee,S.: The Ghirlanda-Guerra identities witho ut averaging. preprint, arXiv:0911.4520 (2009)

work page Pith review arXiv 2009

[7] [6]

: Spin-glass stochastic sta bility: A rigorous proof

Contucci, P., Giardin` a, C. : Spin-glass stochastic sta bility: A rigorous proof. Annales Henri Poincare, 6, 915-923, (2005)

work page 2005

[8] [7]

: The Ghirlanda-Guerra iden tities

Contucci, P., Giardin` a, C. : The Ghirlanda-Guerra iden tities. J. Stat. Phys. 126, 917-931,(2007)

work page 2007