Mean-field solution of structural balance dynamics in nonzero temperature

Amir H. Shirazi; F. Rabbani; G. R. Jafari

arxiv: 1907.09773 · v1 · pith:TDCGGLLMnew · submitted 2019-07-23 · ⚛️ physics.soc-ph

Mean-field solution of structural balance dynamics in nonzero temperature

F. Rabbani , Amir H. Shirazi , G. R. Jafari This is my paper

Pith reviewed 2026-05-24 17:05 UTC · model grok-4.3

classification ⚛️ physics.soc-ph

keywords structural balancesigned networksphase transitionmean-fieldhysteresisBoltzmann-Gibbstension tolerancefirst-order transition

0 comments

The pith

Signed networks undergo a first-order phase transition to structural balance below a critical temperature, with hysteresis between balanced and imbalanced states.

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

The paper extends structural balance dynamics to nonzero temperatures by assigning energies to different triad types via Boltzmann-Gibbs statistics, where temperature acts as a measure of how much tension the network tolerates. In the mean-field treatment, the system switches from a random imbalanced configuration to a globally balanced one when temperature drops below a critical value Tc. Above Tc the balanced state cannot be reached at all. The transition is discontinuous and carries a hysteresis loop, so the final state depends on the direction of temperature change.

Core claim

The mean-field solution of the generalized balance dynamics shows a first-order phase transition at critical temperature Tc: for T > Tc the network remains in an imbalanced random state with no access to structural balance, while below Tc balance becomes accessible; the transition exhibits a hysteresis loop that crosses between the balanced and imbalanced regimes.

What carries the argument

Mean-field solution of triad-energy dynamics with Boltzmann-Gibbs weighting, where temperature parametrizes tension tolerance.

If this is right

Below Tc the network reaches structural balance from suitable initial conditions.
Above Tc only the imbalanced random state is stable.
The hysteresis loop permits the network to remain imbalanced even when temperature is lowered below Tc if started from a random configuration.
Temperature directly controls the tolerance to imbalanced triads.

Where Pith is reading between the lines

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

Real signed networks could display path-dependent outcomes, staying imbalanced despite conditions that would allow balance if history were different.
The same framework could be tested on empirical signed graphs by estimating an effective temperature from observed tension levels.
If the first-order character survives beyond mean-field, small perturbations near Tc might produce large, history-dependent shifts in global balance.

Load-bearing premise

The mean-field approximation captures the global evolution and the Boltzmann-Gibbs energy assignment to triads correctly encodes the network's tension tolerance.

What would settle it

Numerical simulation or empirical measurement showing whether the fraction of balanced triads jumps discontinuously with a hysteresis loop when a tunable tension parameter is swept across the predicted Tc.

Figures

Figures reproduced from arXiv: 1907.09773 by Amir H. Shirazi, F. Rabbani, G. R. Jafari.

**Figure 1.** Figure 1: different types of triads according to structural balance where no imbalanced triads remain (except the jammed states). However, most of the previous works were devoted to analyzing the dynamics of networks by changing the link values or structure to understand tendencies towards or away from balance, without considering temperature. Antal [14] proposed a formulation of balance theory in terms of energy wi… view at source ↗

**Figure 2.** Figure 2: (a) Graphical analysis of Eq.(10) for N=50 and in different temperatures ( [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Figure 3: The mean triad energy (−hSikSkjSjii) versus Monte Carlo steps (n) for different value of temperatures in a fully connected network with N=50. In Fig.4-(a), we can see the mean triad energy based on our analytical results, Eq.(11), which shows the firstorder phase transition and thermal hysteresis. We see that bistability exists, and as a definite temperature (T = Tc) the mean triad energy changes abruptly… view at source ↗

**Figure 4.** Figure 4: (a) Bifurcation diagram for the first-order phase transition in Eq.(11) as a function of temperature. The [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗

**Figure 5.** Figure 5: The mean triad energy (−hSikSkjSjii) versus temperature (T) for different initial states differing in the percent of balanced triads. We consider a balanced initial state with all positive links (−hSikSkjSjii = −1) and imbalanced initial state with all negative links (−hSikSkjSjii = 1). In random initial state, we have almost equal positive and negative links (−hSikSkjSjii ∼ 0). Our findings suggest that t… view at source ↗

read the original abstract

In signed networks with simultaneous friendly and hostile interactions, there is a general tendency to a global structural balance, based on the dynamical model of links status. Although the structural balance represents a state of the network with a lack of contentious situations, there are always tensions in real networks. To study such networks, we generalize the balance dynamics in nonzero temperatures. The presented model uses elements from Boltzmann-Gibbs statistical physics to assign an energy to each type of triad, and it introduces the temperature as a measure of tension tolerance of the network. Based on the mean-field solution of the model, we find out that the model undergoes a first-order phase transition from an imbalanced random state to structural balance with a critical temperature $T_{c}$, where in the case of $T > T_{c}$ there is no chance to reach the balanced state. A main feature of the first-order phase transition is the occurrence of a hysteresis loop crossing the balanced and imbalanced regimes.

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

Finite-T structural balance via mean-field yields a claimed first-order transition with hysteresis, but the closure step is the weakest link.

read the letter

The paper adds temperature to structural balance dynamics on signed networks by assigning Boltzmann-Gibbs energies to triad types and treating T as a tolerance for tension. The mean-field equations then produce a discontinuous jump from an imbalanced state to balance at a critical Tc, together with a hysteresis loop. That nonzero-T extension and the explicit first-order character are the actual new pieces relative to the zero-T literature cited in the abstract. The framing of T as a network-level tolerance parameter is a clean interpretive move and the derivation is laid out step by step in the full text. The central result therefore rests on whether the chosen mean-field factorization remains self-consistent once neighboring triads are coupled through the update rule. In signed-network models local correlations are usually strong, so the closure can easily round or remove the discontinuity; the paper does not appear to test this with direct simulations on finite graphs or small exact solutions. The four energy values for the triad types are also free parameters, which means Tc can be shifted without additional constraints from data. This is a model paper aimed at the complex-networks community that already works on balance theory. Readers who want a statistical-mechanics handle on persistent tension will find the hysteresis feature useful even if they later replace the mean-field step. The work is coherent on its own terms and the claim is falsifiable, so it should go to peer review rather than desk rejection; a referee can check the algebra and ask for numerical validation of the transition order.

Referee Report

2 major / 2 minor

Summary. The manuscript generalizes structural balance dynamics on signed networks to nonzero temperature by assigning Boltzmann-Gibbs energies to the four triad types and interpreting temperature as a measure of tension tolerance. A mean-field closure is applied to the resulting master equations, yielding a first-order phase transition at a critical temperature Tc from an imbalanced random state to a structurally balanced state, together with an associated hysteresis loop.

Significance. If the mean-field derivation is robust, the work supplies an analytic prediction for an abrupt, hysteretic transition in network balance that is not obvious from the microscopic update rules. This could be useful for modeling sudden conflict resolution in social systems and for identifying the role of a temperature-like parameter in real signed networks.

major comments (2)

[§3] §3 (Mean-field closure): The factorization ansatz used to close the equations for the triad densities is introduced without an explicit check against Monte Carlo simulations on finite graphs or against exact solutions on small complete graphs; because the central claim of a discontinuous jump and hysteresis rests entirely on this closure remaining self-consistent, the absence of such validation makes the first-order character of the transition uncertain.
[§2] §2 (Energy assignment): The four triad energies are treated as free parameters whose specific numerical values determine the location of Tc; the manuscript does not demonstrate that the existence or order of the transition is independent of these choices, which weakens the claim that the model universally exhibits a first-order transition.

minor comments (2)

The abstract states the existence of Tc and the hysteresis loop but supplies neither the mean-field equations nor the numerical value of Tc; adding one key equation or the explicit dependence of Tc on the energy parameters would improve readability.
Notation for the order parameter (fraction of balanced triads) is introduced without a clear definition in the main text; a single sentence relating it to the link signs would remove ambiguity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive report. We address each major comment below and outline the revisions we will make to strengthen the manuscript.

read point-by-point responses

Referee: §3 (Mean-field closure): The factorization ansatz used to close the equations for the triad densities is introduced without an explicit check against Monte Carlo simulations on finite graphs or against exact solutions on small complete graphs; because the central claim of a discontinuous jump and hysteresis rests entirely on this closure remaining self-consistent, the absence of such validation makes the first-order character of the transition uncertain.

Authors: We agree that direct validation would increase confidence in the closure. The factorization is the standard mean-field assumption of statistical independence at the level of triads, which closes the master equations exactly in the infinite-N limit. Nevertheless, to address the concern we will add a new section comparing the mean-field predictions against Monte Carlo simulations on finite Erdős–Rényi signed networks (N=500–2000) and against exact enumeration on K4 and K5 complete graphs. These checks will be used to quantify the range of validity of the closure and to confirm that the discontinuous jump and hysteresis survive finite-size effects. revision: yes
Referee: §2 (Energy assignment): The four triad energies are treated as free parameters whose specific numerical values determine the location of Tc; the manuscript does not demonstrate that the existence or order of the transition is independent of these choices, which weakens the claim that the model universally exhibits a first-order transition.

Authors: The energies are indeed free parameters that encode the relative tension of each triad type. In the present work we chose a representative set that penalizes imbalanced triads more strongly than balanced ones, which is the physically relevant regime. We will revise the manuscript to (i) state explicitly that the first-order character is expected whenever the two balanced triads have lower energy than the two imbalanced triads, and (ii) add a brief parameter scan showing that the discontinuous transition and hysteresis persist for a broad interval of energy differences (including cases where one balanced triad is only marginally favored). This will clarify the domain of universality without altering the core analytic results. revision: partial

Circularity Check

0 steps flagged

Mean-field derivation of first-order transition is independent of modeling inputs

full rationale

The paper defines a dynamical model by assigning Boltzmann-Gibbs energies to triad types and introduces temperature as a tension parameter; the mean-field equations are then solved to obtain the order-parameter jump and hysteresis loop. These outcomes are generated by the closure equations themselves rather than by fitting the transition temperature to the same data used to define the energies, and no self-citation chain or ansatz smuggling is required to reach the central claim. The derivation therefore remains self-contained against standard statistical-mechanics benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claim rests on assigning energies to the four possible triad types and on the validity of the mean-field closure; these choices are not derived from first principles within the abstract and function as modeling assumptions.

free parameters (1)

energy values for each triad type
Energies are assigned to the four possible signed triads to define the Hamiltonian; their specific numerical values are not given in the abstract and must be chosen to produce the dynamics.

axioms (2)

domain assumption Boltzmann-Gibbs statistics governs the probability of triad configurations at finite temperature
The model imports the canonical ensemble to link energy, temperature, and link-flip probabilities.
domain assumption Mean-field closure accurately represents the global link dynamics
The derivation replaces local correlations with average fields to obtain closed equations for the order parameter.

pith-pipeline@v0.9.0 · 5700 in / 1450 out tokens · 23548 ms · 2026-05-24T17:05:17.030158+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

29 extracted references · 29 canonical work pages

[1]

By considering the temperature as a randomness of the social process, regardless of the balanced state of triads, we could extend our results to society at large

Fig.5 shows that trajectories starting from a bal- anced initial state lie into the balanced ﬁxed point at −⟨SikSkjSji⟩ =−1, while trajectories starting from ran- dom and imbalanced initial states go to the random ﬁxed point at−⟨SikSkjSji⟩ = 0. By considering the temperature as a randomness of the social process, regardless of the balanced state of triads...

work page
[2]

Heider, The Journal of Psychology 21, 107 (1946)

F. Heider, The Journal of Psychology 21, 107 (1946)

work page 1946
[3]

Cartwright and F

D. Cartwright and F. Harary, Psychological review 63, 277 (1956)

work page 1956
[4]

Jeﬀrey, Journal of Peace Research 11, 229 (1974)

H. Jeﬀrey, Journal of Peace Research 11, 229 (1974)

work page 1974
[5]

N. P. Hummon and P. Doreian, Social Networks 25, 17 (2003)

work page 2003
[6]

Szell, R

M. Szell, R. Lambiotte, and S. Thurner, Proceedings of the National Academy of Sciences 107, 13636 (2010)

work page 2010
[7]

Facchetti, G

G. Facchetti, G. Iacono, and C. Altaﬁni, Proceedings of the National Academy of Sciences 108, 20953 (2011)

work page 2011
[8]

Lerner, Social Networks 45, 66 (2016)

J. Lerner, Social Networks 45, 66 (2016)

work page 2016
[9]

H. Saiz, J. G´ omez-Garde˜ nes, P. Nuche, A. Gir´ on, Y. Pueyo, and C. L. Alados, Ecography 40, 733 (2017)

work page 2017
[10]

Saeedian, N

M. Saeedian, N. Azimi-Tafreshi, G. R. Jafari, and J. Kertesz, Physical Review E 95, 022314 (2017)

work page 2017
[11]

S. A. Marvel, S. H. Strogatz, and J. M. Kleinberg, Phys. Rev. Lett. 103, 198701 (2009)

work page 2009
[12]

Abell and M

P. Abell and M. Ludwig, Journal of Mathematical Soci- ology 33, 129 (2009)

work page 2009
[13]

Leskovec, D

J. Leskovec, D. Huttenlocher, and J. Kleinberg, in Pro- ceedings of the SIGCHI conference on human factors in computing systems (ACM, 2010) pp. 1361–1370

work page 2010
[14]

V. A. Traag, P. Van Dooren, and P. De Leenheer, PloS one 8, e60063 (2013)

work page 2013
[15]

Antal, P

T. Antal, P. L. Krapivsky, and S. Redner, Phys. Rev. E 72, 036121 (2005)

work page 2005
[16]

J. A. Davis, Human relations 20, 181 (1967)

work page 1967
[17]

S. A. Marvel, J. Kleinberg, R. D. Kleinberg, and S. H. Strogatz, Proceedings of the National Academy of Sci- ences 108, 1771 (2011)

work page 2011
[18]

Hedayatifar, F

L. Hedayatifar, F. Hassanibesheli, A. Shirazi, S. V. Fara- hani, and G. R. Jafari, Physica A: Statistical Mechanics and its Applications 483, 109 (2017)

work page 2017
[19]

Hassanibesheli, L

F. Hassanibesheli, L. Hedayatifar, H. Safdari, M. Aus- loos, and G. R. Jafari, Entropy 19, 246 (2017)

work page 2017
[20]

W. L. Langer (2nd ed., Knopf, New York, 1950)

work page 1950
[21]

Ku lakowski, P

K. Ku lakowski, P. Gawro´ nski, and P. Gronek, Interna- tional Journal of Modern Physics C 16, 707 (2005)

work page 2005
[22]

M. E. J. Newman, C. Moore, and D. J. Watts, Physical Review Letters 84, 3201 (2000)

work page 2000
[23]

Albert and A.-L

R. Albert and A.-L. Barab´ asi, Reviews of modern physics 74, 47 (2002)

work page 2002
[24]

Park and M

J. Park and M. E. J. Newman, Physical Review E 72, 026136 (2005)

work page 2005
[25]

S. N. Dorogovtsev, A. V. Goltsev, and J. F. F. Mendes, Reviews of Modern Physics 80, 1275 (2008)

work page 2008
[26]

A. M. Belaza, K. Hoefman, J. Ryckebusch, A. Bramson, M. van den Heuvel, and K. Schoors, PLoS one 12, e0183696 (2017)

work page 2017
[27]

H. Du, X. He, J. Wang, and M. W. Feldman, Physica A: Statistical Mechanics and its Applications 503, 780 (2018)

work page 2018
[28]

A. M. Belaza, J. Ryckebusch, A. Bramson, C. Casert, K. Hoefman, K. Schoors, M. van den Heuvel, and B. Vandermarliere, Physica A: Statistical Mechanics and its Applications 518, 270 (2019)

work page 2019
[29]

Kirkley, G

A. Kirkley, G. T. Cantwell, and M. E. J. Newman, Phys- ical Review E 99, 012320 (2019). 8 Appendix A: Partition F unction Calculations In this appendix, we have given the partition function calculations for the Hamiltonian Eq. (1) and introduce hij as an external ﬁeld on Sij, Hij =−Sij ∑ Sij =±1 SjkSki−Sijhij. (A1) Therefore, the partition function can be...

work page 2019

[1] [1]

By considering the temperature as a randomness of the social process, regardless of the balanced state of triads, we could extend our results to society at large

Fig.5 shows that trajectories starting from a bal- anced initial state lie into the balanced ﬁxed point at −⟨SikSkjSji⟩ =−1, while trajectories starting from ran- dom and imbalanced initial states go to the random ﬁxed point at−⟨SikSkjSji⟩ = 0. By considering the temperature as a randomness of the social process, regardless of the balanced state of triads...

work page

[2] [2]

Heider, The Journal of Psychology 21, 107 (1946)

F. Heider, The Journal of Psychology 21, 107 (1946)

work page 1946

[3] [3]

Cartwright and F

D. Cartwright and F. Harary, Psychological review 63, 277 (1956)

work page 1956

[4] [4]

Jeﬀrey, Journal of Peace Research 11, 229 (1974)

H. Jeﬀrey, Journal of Peace Research 11, 229 (1974)

work page 1974

[5] [5]

N. P. Hummon and P. Doreian, Social Networks 25, 17 (2003)

work page 2003

[6] [6]

Szell, R

M. Szell, R. Lambiotte, and S. Thurner, Proceedings of the National Academy of Sciences 107, 13636 (2010)

work page 2010

[7] [7]

Facchetti, G

G. Facchetti, G. Iacono, and C. Altaﬁni, Proceedings of the National Academy of Sciences 108, 20953 (2011)

work page 2011

[8] [8]

Lerner, Social Networks 45, 66 (2016)

J. Lerner, Social Networks 45, 66 (2016)

work page 2016

[9] [9]

H. Saiz, J. G´ omez-Garde˜ nes, P. Nuche, A. Gir´ on, Y. Pueyo, and C. L. Alados, Ecography 40, 733 (2017)

work page 2017

[10] [10]

Saeedian, N

M. Saeedian, N. Azimi-Tafreshi, G. R. Jafari, and J. Kertesz, Physical Review E 95, 022314 (2017)

work page 2017

[11] [11]

S. A. Marvel, S. H. Strogatz, and J. M. Kleinberg, Phys. Rev. Lett. 103, 198701 (2009)

work page 2009

[12] [12]

Abell and M

P. Abell and M. Ludwig, Journal of Mathematical Soci- ology 33, 129 (2009)

work page 2009

[13] [13]

Leskovec, D

J. Leskovec, D. Huttenlocher, and J. Kleinberg, in Pro- ceedings of the SIGCHI conference on human factors in computing systems (ACM, 2010) pp. 1361–1370

work page 2010

[14] [14]

V. A. Traag, P. Van Dooren, and P. De Leenheer, PloS one 8, e60063 (2013)

work page 2013

[15] [15]

Antal, P

T. Antal, P. L. Krapivsky, and S. Redner, Phys. Rev. E 72, 036121 (2005)

work page 2005

[16] [16]

J. A. Davis, Human relations 20, 181 (1967)

work page 1967

[17] [17]

S. A. Marvel, J. Kleinberg, R. D. Kleinberg, and S. H. Strogatz, Proceedings of the National Academy of Sci- ences 108, 1771 (2011)

work page 2011

[18] [18]

Hedayatifar, F

L. Hedayatifar, F. Hassanibesheli, A. Shirazi, S. V. Fara- hani, and G. R. Jafari, Physica A: Statistical Mechanics and its Applications 483, 109 (2017)

work page 2017

[19] [19]

Hassanibesheli, L

F. Hassanibesheli, L. Hedayatifar, H. Safdari, M. Aus- loos, and G. R. Jafari, Entropy 19, 246 (2017)

work page 2017

[20] [20]

W. L. Langer (2nd ed., Knopf, New York, 1950)

work page 1950

[21] [21]

Ku lakowski, P

K. Ku lakowski, P. Gawro´ nski, and P. Gronek, Interna- tional Journal of Modern Physics C 16, 707 (2005)

work page 2005

[22] [22]

M. E. J. Newman, C. Moore, and D. J. Watts, Physical Review Letters 84, 3201 (2000)

work page 2000

[23] [23]

Albert and A.-L

R. Albert and A.-L. Barab´ asi, Reviews of modern physics 74, 47 (2002)

work page 2002

[24] [24]

Park and M

J. Park and M. E. J. Newman, Physical Review E 72, 026136 (2005)

work page 2005

[25] [25]

S. N. Dorogovtsev, A. V. Goltsev, and J. F. F. Mendes, Reviews of Modern Physics 80, 1275 (2008)

work page 2008

[26] [26]

A. M. Belaza, K. Hoefman, J. Ryckebusch, A. Bramson, M. van den Heuvel, and K. Schoors, PLoS one 12, e0183696 (2017)

work page 2017

[27] [27]

H. Du, X. He, J. Wang, and M. W. Feldman, Physica A: Statistical Mechanics and its Applications 503, 780 (2018)

work page 2018

[28] [28]

A. M. Belaza, J. Ryckebusch, A. Bramson, C. Casert, K. Hoefman, K. Schoors, M. van den Heuvel, and B. Vandermarliere, Physica A: Statistical Mechanics and its Applications 518, 270 (2019)

work page 2019

[29] [29]

Kirkley, G

A. Kirkley, G. T. Cantwell, and M. E. J. Newman, Phys- ical Review E 99, 012320 (2019). 8 Appendix A: Partition F unction Calculations In this appendix, we have given the partition function calculations for the Hamiltonian Eq. (1) and introduce hij as an external ﬁeld on Sij, Hij =−Sij ∑ Sij =±1 SjkSki−Sijhij. (A1) Therefore, the partition function can be...

work page 2019