Pith. sign in

REVIEW 4 minor 1 cited by

This paper proves that any social impact model on a hypergraph—where a node flips with probability set by the fraction of opposite opinions in a group—can be exactly rewritten as a pairwise imitation process on a weighted projected network,

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-03 11:46 UTC pith:SNVQG3DR

load-bearing objection The exact microscopic reduction is sound and the paper honestly reports its limits; only the abstract's unqualified macroscopic-equivalence claim overreaches.

arxiv 2601.05169 v2 pith:SNVQG3DR submitted 2026-01-08 physics.soc-ph

Reducibility of higher-order to pairwise interactions: Social impact models on hypergraphs

Jaume Llabr\'es , Ra\'ul Toral , Maxi San Miguel , Federico V\'azquez This is my paper
classification physics.soc-ph PACS 89.75.Fb89.65.-s
keywords higher-order interactionshypergraphssocial impact modelsvoter modelpairwise reductionweighted projected networksopinion dynamicspair approximation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper establishes an exact microscopic reduction: for a broad class of binary-state opinion models on arbitrary hypergraphs, the stochastic dynamics is identical to a pairwise imitation process on a projected network with suitably chosen link weights. The reduction works for any social impact function satisfying f(0)=0 and f(1)=1, and it preserves every transition probability. In the linear case, the weights are static, and a pair approximation shows that the macroscopic ordering dynamics is independent of the weight distribution—matching the standard voter model on the unweighted projected network. In the nonlinear case, the weights become state-dependent, yet the nonlinear voter model still reproduces the main macroscopic trends on well-connected hypergraphs. A sympathetic reader would care because this sharply delimits when higher-order topology genuinely changes dynamics and when a hypergraph representation is merely a modeling convenience.

Core claim

The central claim, stated as an exact mapping, is that a social impact model on a hypergraph can be reduced at the microscopic level to a pairwise model on a weighted projected network. The reduction decomposes each link weight into contributions from shared hyperedges, assigned through two conditions: probability conservation per hyperedge and flip consistency. For any configuration, the hypergraph flip probability equals the pairwise copy probability, so the two Markov chains have identical laws. For the hypergraph-linear voter model, the weights depend only on topology, not on the node states, enabling a pair approximation that yields the same macroscopic equations as the standard voter m

What carries the argument

The central object is the assignment of effective pairwise weights ω^e_i(s_i,s_j) for each hyperedge e, defined by Eqs. (16)-(17). These weights split the flip probability of the higher-order model among the neighbors of the focal node, respecting two constraints: the sum over neighbors in a hyperedge equals 1/κ_i, and the sum over opposite-state neighbors reproduces f(φ_e)/κ_i. The total weight ω_ij sums these contributions over all shared hyperedges. In the linear case, this construction yields static weights; a pair approximation for weighted, uncorrelated networks then shows that the macroscopic equations are identical to those of the standard voter model, making the weight distribution

Load-bearing premise

The pair approximation assumes that the probability a link is active depends only on the focal node's state, ignoring link weight, node degree, and network clustering, and treats the weighted projected network as uncorrelated; the paper's own simulations show this fails at low mean degree where clustering is strong.

What would settle it

On a single hyperedge of order d, pick any configuration and compare the hypergraph flip probability f(n/d)/κ_i with the sum of the assigned pairwise copy probabilities over opposite-state neighbors; if they ever differ, the exact reduction is false. Alternatively, simulate both the hypergraph dynamics and the weighted projected dynamics from identical initial conditions and compare their full trajectory distributions—the claimed mapping implies they must coincide exactly.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • For the hypergraph-linear voter model, the ordering dynamics is independent of the weight distribution and matches the voter model on the unweighted projected network, within the pair approximation and confirmed numerically across the explored parameter range.
  • For nonlinear hypergraph-voter models, the original higher-order dynamics and the reduced pairwise dynamics on the weighted projected network overlap exactly by construction, and the nonlinear voter model on the unweighted projected network provides a good effective description for well-connected hypergraphs.
  • The exact reduction implies that, for node-update dynamics, a weighted projected network fully captures the microscopic behavior of this class of hypergraph models, so any dynamical feature not captured by the projected degree distribution must arise from clustering or higher-order correlations beyond the pair approximation.
  • For the hypergraph-nonlinear voter model, fixation times are non-monotonic in the nonlinearity parameter, with an optimal value that depends on hypergraph topology, and in the well-connected limit the dynamics converges to the completed-hypergraph prediction.

Where Pith is reading between the lines

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

  • The same reduction strategy may apply to other binary-state dynamics on hypergraphs, such as threshold models, but the resulting weights will generally be state-dependent, so the simplifying static-weight regime is special to linear group influence functions.
  • The pair-approximation result suggests a practical diagnostic: if a real-world opinion process on a hypergraph shows ordering behavior different from the standard voter model on its projected network, the difference is evidence of clustering or higher-order correlations, not merely the presence of group interactions.
  • The exact microscopic equivalence implies a quantitative criterion for model selection: whenever the projected network's weighted dynamics can be simulated, the hypergraph representation is dynamically redundant for this class of models, so simpler pairwise models may be preferred for parameter estimation.
  • A testable extension would be to check whether the reduction also preserves finite-size fluctuation properties such as fixation-time distributions, not just the averaged observables reported here; if it does, the projected network could replace the hypergraph in Monte Carlo studies of rare events.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

0 major / 4 minor

Summary. The paper studies binary-state social impact models on hypergraphs, in which a node updates by selecting one of its incident hyperedges and flipping with probability f(phi_e), where phi_e is the fraction of opposite-state nodes in that hyperedge (excluding the focal node). The central claim is that, under node-update dynamics and for any social impact function with f(0)=0 and f(1)=1, this higher-order dynamics can be mapped exactly onto a pairwise imitation dynamics on a weighted projected network. The weight construction, Eqs. (16)-(17), is derived from two conditions, Eqs. (14)-(15), that enforce probability conservation and flip consistency, and the authors verify that the microscopic transition probabilities coincide for every configuration. As a particular case, hypergraph-voter models f(phi)=phi^q are analyzed: for q=1 the weights are state-independent, and a pair approximation leads to macroscopic equations identical to those of the standard voter model on the unweighted projected network; for q!=1 the weights are configuration-dependent, and the nonlinear voter model on the projected network is shown to reproduce the main trends for well-connected hypergraphs. Numerical simulations on Erdős-Rényi and z-regular random d-hypergraphs support the analytical results.

Significance. The exact microscopic reduction is a valuable theoretical contribution. It provides a constructive, parameter-free mapping from a broad class of higher-order interactions to pairwise imitation dynamics, with explicit weights that are guaranteed to preserve every transition probability. This is stronger than the existing macroscopic or approximate reducibility arguments cited in the introduction. The special case q=1 is particularly elegant: the static weights and the pair-approximation result that the ordering dynamics is independent of the weight distribution give a concrete, falsifiable prediction, and the numerical evidence confirms the equivalence for the ensembles studied. The paper is also honest in documenting the limitations of the pair approximation at low mean degree, where clustering induced by the hypergraph construction causes systematic deviations. If the central result holds, it clarifies when a hypergraph representation is necessary and when a pairwise network model is sufficient.

minor comments (4)
  1. [Abstract and Sec. VB1] The sentence 'The macroscopic dynamics is thus equivalent to that of the standard voter model on the unweighted projected network' is stated without qualification. This equivalence is derived within a pair approximation whose closure (Eq. G5) neglects weight, degree, and clustering correlations, and Fig. 2 shows quantitative deviations from the PA at low mean degree. Although the numerical HO/PW/VM overlap supports the claim for ER and RR hypergraphs, the abstract and Sec. VB1 should make explicit that the equivalence is approximate and established for the ensembles studied, to avoid overstating the generality.
  2. [Sec. IIB] The social impact function f is not explicitly restricted to take values in [0,1] for phi in [0,1]. This restriction is needed for the interpretation as a probability and for the non-negativity of the weights in Eqs. (16)-(17). Please state this assumption explicitly.
  3. [Appendix G] The pair approximation assumes that P_{-s|s}, the conditional probability that a link is active given the focal node's state, is independent of the link weight, of the degree, and of clustering. This is a strong closure. The authors state it, but it would be helpful to add a sentence noting that the weight-independence result is a consequence of this closure and is not an exact property of the full dynamics; the numerical evidence covers the ER/RR ensembles and may not hold for strongly heterogeneous hypergraphs.
  4. [Sec. VI] Typo: 'pojected network' should be 'projected network'. Also, in the caption of Fig. 2, the solid lines are labeled only as 'theoretical predictions from the Pair Approximation'; it would be clearer to explicitly state that the deviations from these lines at low mu are discussed in Sec. VB1.

Circularity Check

0 steps flagged

No significant circularity: the exact HO-to-PW mapping is a parameter-free construction, and the macroscopic claims are explicitly approximation-based and tested against fresh simulations.

full rationale

The central exact result—that the hypergraph social-impact process and the weighted projected-network process have identical microscopic transition probabilities—is derived self-contained from stated update rules. Equations (13)–(17) construct weights by imposing probability conservation per hyperedge (Eq. 14) and flip consistency (Eq. 15); summing over hyperedges in Eq. (13) reproduces Eq. (7) by direct substitution, with no fitted constants and no appeal to a prior result. The only mild concern is the pair-approximation closure in Appendix G: Eq. (G5) assumes P_{−s|s} is independent of link weight and degree, and Eqs. (G6)–(G9) then yield macroscopic equations independent of weights. That is a transparent mean-field assumption, not a prediction smuggled in as its own premise; the paper explicitly labels it a pair approximation, derives the standard voter-model equations from it, and then compares them with independent simulations. The low-μ disagreement between the PA curves and the numerics is disclosed in Sec. VB1 and Fig. 2 and attributed to clustering, which further shows the theory is falsifiable rather than constructed to match data. Self-citations (e.g., Refs. [19], [30], [38]) are used as background or as external published benchmarks for the NLVM and the group-driven voter model; none of these citations carries the exact reducibility theorem, which stands on the paper's own equations. No parameter is fitted to the headline observables ξ(µ) or τ, and no claim reduces by definition to its input. Therefore the derivation chain is not circular.

Axiom & Free-Parameter Ledger

0 free parameters · 6 axioms · 0 invented entities

The paper's central claim is a self-contained derivation: given node-update dynamics, an absorbing-state f, and a normalization/consistency pair of conditions, the weights (16)-(17) reproduce the HO generator exactly with no fitted parameters. Load-bearing axiomatic content lives elsewhere: the equal-split convention (arbitrary but harmless), the N≫d binomial approximation, the PA closure that carries the macroscopic equivalence, and the restriction to connected hypergraphs. No new entities are postulated; the configuration-dependent weight field is constructed from the dynamics, not assumed.

axioms (6)
  • domain assumption Social impact functions restricted to f(0)=0 and f(1)=1 (absorbing states)
    Required by the consistency conditions (14)-(15); stated explicitly in Sec. IIIB ('the present mapping is restricted to social impact functions with absorbing states'). Excludes models without absorbing states, e.g. f(0)>0.
  • domain assumption Node-update dynamics (one node updated per time step)
    The mapping equates single-node flip rates; Sec. VI states that hyperedge-update dynamics 'cannot be mapped exactly onto a microscopically equivalent pairwise system, since pairwise models update only one node per interaction'.
  • ad hoc to paper Equal split of weights among same-state neighbors within a hyperedge
    Sec. IIIB: 'we impose the natural condition that all neighbors j∈e sharing the same state are statistically equivalent'. Not forced by (14)-(15); any split preserves the exact flip probabilities. Affects only the bookkeeping, not the central exact claim.
  • standard math Hypergeometric-to-binomial approximation in the limit N ≫ d
    App. D, Eq. (D7) with an explicit O(d/N) correction; used for the completed-hypergraph weights (Eqs. 26-29) and the drift Eq. (32). Applicable only when d is small relative to N.
  • domain assumption Pair-approximation closure: P_{−s|s} independent of link weight, degree, and clustering; uncorrelated weighted network
    App. G (text near Eq. G5) assumes a PA 'homogeneous in weights and degrees'. This carries the headline macroscopic claim (Eqs. 39-42). The paper's own Fig. 2 shows it degrades at low µ due to clustering, so the macroscopic equivalence is approximate, not exact.
  • domain assumption Simulations restricted to connected hypergraphs with conditioned low-µ ensembles
    Sec. IIA and App. A: only connected hypergraphs are used; low-µ ER generation conditions on the largest connected component with acceptance criteria, so the simulated ensembles deviate from the naive ER/RR measure. Relevant to the low-µ comparisons in Fig. 2.

pith-pipeline@v1.3.0-alltime-deepseek · 34222 in / 30027 out tokens · 311622 ms · 2026-08-03T11:46:42.286823+00:00 · methodology

0 comments
read the original abstract

We show that a general class of social impact models with higher-order interactions on hypergraphs can be exactly reduced to an equivalent model with pairwise interactions on a weighted projected network. This reduction is made by a mapping that preserves the microscopic probabilities of changing the state of the nodes. As a particular case, we introduce hypergraph-voter models, for which we compute the weights of the projected network both analytically and numerically across several hypergraph ensembles, and we characterize their ordering dynamics through simulations of both higher-order and reduced dynamics. For a linear social impact function (hypergraph-linear voter model) the weights of the projected network are static, allowing us to develop a pair approximation that describes with accuracy the time evolution of macroscopic observables, which turn out to be independent of those weights. The macroscopic dynamics is thus equivalent to that of the standard voter model on the unweighted projected network. For a power-law social impact function (hypergraph-nonlinear voter model) the weights of the projected network depend on the instantaneous system configuration. Nevertheless, the nonlinear voter model on the unweighted projected network still reproduces the main macroscopic trends for well connected hypergraphs.

Figures

Figures reproduced from arXiv: 2601.05169 by Federico V\'azquez, Jaume Llabr\'es, Maxi San Miguel, Ra\'ul Toral.

Figure 1
Figure 1. Figure 1: FIG. 1. Illustration of (a) a hypergraph [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2 [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3 [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4 [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Higher-order interactions in ecology can be hidden in plain sight

    q-bio.PE 2026-05 unverdicted novelty 6.0

    Higher-order ecological interactions can be accurately reproduced by effective pairwise models fitted to abundance time series, so interaction structure cannot be reliably inferred from time series data alone.

Reference graph

Works this paper leans on

75 extracted references · 2 linked inside Pith · cited by 1 Pith paper

  1. [1]

    Lets i be its state and κi its hyperdegree

    Select a nodeiat random. Lets i be its state and κi its hyperdegree

  2. [2]

    Letd e be its order andn e the number of nodes, excludingi, with state+1that it contains

    Select randomly one of theκi incident hyperedges e∈ Ei. Letd e be its order andn e the number of nodes, excludingi, with state+1that it contains

  3. [3]

    (6) TimeismeasuredinMonteCarlosteps(MCS).OneMCS corresponds toNupdating attempts such that, on aver- age, each node is selected once per MCS

    Nodeiflips its state, i.e.s i → −si, with a proba- bilityf(ϕ e), whereϕ e is the fraction of nodes (ex- cluding nodei) in the hyperedgeethat are in the opposite state tosi, ϕe =    ne de ifs i =−1, 1− ne de ifs i = +1. (6) TimeismeasuredinMonteCarlosteps(MCS).OneMCS corresponds toNupdating attempts such that, on aver- age, each node is selected once pe...

  4. [4]

    Select a nodeiat random

  5. [5]

    Select one of its neighborsjwith probabilityω ij

  6. [6]

    Nodeicopies the state of nodej,s i →s j. These updating rules lead to the following flipping probability for nodei: Ω(si → −si) = X j∈Vi ωij δsj ,−si .(10) Note that this model presents two absorbing states, where all nodes are in the same state, either+1or−1. It is worth noting that, for the case ofdegree- homogeneous weights, that is,ωij = 1/k i for all...

  7. [7]

    This condition ensures that a neighborjis always selected, P e∈Ei P j∈e ωe ij = 1

    Probability conservation per hyperedge: The prob- ability that a nodeirandomly selects an incident hyperedgeein the HO dynamics must be equal to the probability thaticopies one of its neighbors j∈ein the PW interaction, 1 κi = X j∈e ωe ij,(14) where we recall thatκi is the hyperdegree of nodei. This condition ensures that a neighborjis always selected, P ...

  8. [8]

    This leads to the condition: 1 κi f(ϕ e) = X j∈e ωe ij δsj ,−si ,(15) where we recall thatϕe is the fraction of nodes in the opposite state toiine, given by Eq

    Flip consistency: Given the selected hyperedgee, the flipping probability of nodeiin the HO dynam- ics must be equal to the probability thatiflips its state by copying a neighborj∈ewith opposite states j =−s i. This leads to the condition: 1 κi f(ϕ e) = X j∈e ωe ij δsj ,−si ,(15) where we recall thatϕe is the fraction of nodes in the opposite state toiine...

  9. [9]

    Together, these equations ensure that the PW dynamics reproduces the microscopic transition probabilities of the HO interaction model

    Therefore, the present mapping is restricted to social impact functions with absorbing states. Together, these equations ensure that the PW dynamics reproduces the microscopic transition probabilities of the HO interaction model. Equation (14) fixes the total probability assigned to the neighbors ofiwithin each hyperedgee, while Eq. (15) constrains how th...

  10. [10]

    Hypergraph-linear voter model The hypergraph-linear voter model corresponds to the linear caseq= 1of the social impact function defined in Eq. (30). For the fully connected hypergraph and the complete d-hypergraph, substituting the linear form into the re- spective expressions for the weights, Eqs. (23) and (28), yields the same state-independent weights,...

  11. [11]

    Nodeidivides a unit weight equally among itsκ i incident hyperedges, so each incident hyperedgee receivesκ −1 i

  12. [12]

    Therefore, each neighborjinegets a weight ωe ij = (κide)−1

    Within each such hyperedgee, this amount is dis- tributed equally among thed e neighbors ofiin e. Therefore, each neighborjinegets a weight ωe ij = (κide)−1. Ifiandjshare multiple hyperedges, the contributions add up. Unlike the fully connected and complete hypergraph cases, where both descriptions reduce to the VM on the complete graph, the presence of s...

  13. [13]

    (30) into 12 0 10 20 30 40 50 60 µ 0.25 0.30 0.35 0.40 0.45 0.50ξ ER (a) d 2 3 4 5 HO PW NL VM 0 10 20 30 40 50 60 µ 0.25 0.30 0.35 0.40 0.45 0.50ξ RR (b) FIG

    Hypergraph-nonlinear voter model For the fully connected hypergraph, the weights of the projected network obtained by substituting Eq. (30) into 12 0 10 20 30 40 50 60 µ 0.25 0.30 0.35 0.40 0.45 0.50ξ ER (a) d 2 3 4 5 HO PW NL VM 0 10 20 30 40 50 60 µ 0.25 0.30 0.35 0.40 0.45 0.50ξ RR (b) FIG. 3.Hypergraph-nonlinear voter model (q= 0.8).Plateau valueξvers...

  14. [14]

    Lambiotte, M

    R. Lambiotte, M. Rosvall, and I. Scholtes, From networks to optimal higher-order models of complex systems, Na- ture Physics15, 313 (2019)

    2019

  15. [15]

    G. F. de Arruda, G. Petri, and Y. Moreno, Social conta- gion models on hypergraphs, Phys. Rev. Res.2, 023032 (2020)

    2020

  16. [16]

    Battiston, G

    F. Battiston, G. Cencetti, I. Iacopini, V. Latora, M. Lu- cas, A. Patania, J.-G. Young, and G. Petri, Networks beyond pairwise interactions: Structure and dynamics, Physics Reports874, 1 (2020)

    2020

  17. [17]

    Sahasrabuddhe, L

    R. Sahasrabuddhe, L. Neuhäuser, and R. Lambiotte, Modelling non-linear consensus dynamics on hyper- graphs, Journal of Physics: Complexity2, 025006 (2021)

    2021

  18. [18]

    Battiston, V

    F. Battiston, V. Capraro, F. Karimi, S. Lehmann, A. B. Migliano, O. Sadekar, A. Sánchez, and M. Perc, Higher- order interactions shape collective human behaviour, Na- ture Human Behaviour9, 2441 (2025)

    2025

  19. [19]

    Bianconi,Higher-Order Networks, Elements in the Structure and Dynamics of Complex Networks (Cam- bridge University Press, 2021)

    G. Bianconi,Higher-Order Networks, Elements in the Structure and Dynamics of Complex Networks (Cam- bridge University Press, 2021)

    2021

  20. [20]

    Battiston, E

    F. Battiston, E. Amico, A. Barrat, G. Bianconi, G. Fer- raz de Arruda, B. Franceschiello, I. Iacopini, S. Kéfi, V. Latora, Y. Moreno, M. M. Murray, T. P. Peixoto, F. Vaccarino, and G. Petri, The physics of higher-order interactions in complex systems, Nature Physics17, 1093 (2021)

    2021

  21. [21]

    Ferraz de Arruda, A

    G. Ferraz de Arruda, A. Aleta, and Y. Moreno, Conta- gion dynamics on higher-order networks, Nature Reviews Physics6, 468 (2024)

    2024

  22. [22]

    K. F. Kee, L. Sparks, D. C. Struppa, and M. Mannucci, Socialgroups, socialmedia, andhigherdimensionalsocial structures: A simplicial model of social aggregation for computational communication research, Communication Quarterly61, 35 (2013)

    2013

  23. [23]

    Greening, Bradford R, N

    J. Greening, Bradford R, N. Pinter-Wollman, and N. H. Fefferman, Higher-order interactions: Understanding the knowledge capacity of social groups using simplicial sets, Current Zoology61, 114 (2015)

    2015

  24. [24]

    S. E. Asch, Effects of group pressure upon the modifi- cation and distortion of judgments, inGroups, leader- ship and men; research in human relations, edited by H. Guetzkow (Carnegie Press, Pittsburgh, PA, 1951) pp. 177–190

    1951

  25. [25]

    Centola, V

    D. Centola, V. M. Eguíluz, and M. W. Macy, Cascade dynamics of complex propagation, Physica A: Statistical Mechanics and its Applications374, 449 (2007)

    2007

  26. [26]

    Centola,How Behavior Spreads: The Science of Com- plex Contagions(Princeton University Press, 2018)

    D. Centola,How Behavior Spreads: The Science of Com- plex Contagions(Princeton University Press, 2018)

    2018

  27. [27]

    Granovetter, Threshold models of collective behavior, American Journal of Sociology83, 1420 (1978)

    M. Granovetter, Threshold models of collective behavior, American Journal of Sociology83, 1420 (1978)

    1978

  28. [28]

    D. J. Watts, A simple model of global cascades on ran- dom networks, Proceedings of the National Academy of Sciences99, 5766 (2002)

    2002

  29. [29]

    J. P. Gleeson, Binary-state dynamics on complex net- works: Pair approximation and beyond, Phys. Rev. X3, 021004 (2013)

    2013

  30. [30]

    Pastor-Satorras, C

    R. Pastor-Satorras, C. Castellano, P. Van Mieghem, and A. Vespignani, Epidemic processes in complex networks, Rev. Mod. Phys.87, 925 (2015)

    2015

  31. [31]

    Iacopini, G

    I. Iacopini, G. Petri, A. Barrat, and V. Latora, Simplicial models of social contagion, Nature Communications10, 2485 (2019)

    2019

  32. [32]

    L. S. Ramirez, F. Vázquez, M. San Miguel, and T. Galla, Ordering dynamics of nonlinear voter models, Phys. Rev. E109, 034307 (2024)

    2024

  33. [33]

    C. R. Kitching, L. S. Ramirez, M. San Miguel, and T. Galla, Breaking coexistence: Zealotry vs nonlinear so- cial impact, Chaos: An Interdisciplinary Journal of Non- linear Science35, 083133 (2025)

    2025

  34. [34]

    Llabrés, M

    J. Llabrés, M. San Miguel, and R. Toral, Universality of noise-induced transitions in nonlinear voter models, arXiv , 2505.21407 (2025)

    Pith/arXiv arXiv 2025

  35. [35]

    Huang,Statistical Mechanics(John wiley and Sons, 1987)

    K. Huang,Statistical Mechanics(John wiley and Sons, 1987)

    1987

  36. [36]

    Pathria and P

    R. Pathria and P. Beale,Statistical Mechanics 4th edition (Elsevier LTD, 2021)

    2021

  37. [37]

    Neuhäuser, A

    L. Neuhäuser, A. Mellor, and R. Lambiotte, Multibody interactions and nonlinear consensus dynamics on net- worked systems, Phys. Rev. E101, 032310 (2020)

    2020

  38. [38]

    Meloni, A

    S. Meloni, A. Gabrielli, and P. Villegas, Higher- order contagion processes in 1.99 dimensions (2025), arXiv:2502.18004 [cond-mat.stat-mech]

    Pith/arXiv arXiv 2025

  39. [39]

    Young, G

    J.-G. Young, G. Petri, and T. P. Peixoto, Hyper- graph reconstruction from network data, Communica- tions Physics4, 135 (2021)

    2021

  40. [40]

    Malizia, A

    F. Malizia, A. Corso, L. V. Gambuzza, G. Russo, V. La- tora, and M. Frasca, Reconstructing higher-order inter- actions in coupled dynamical systems, Nature Commu- nications15, 5184 (2024)

    2024

  41. [41]

    Lucas, L

    M. Lucas, L. Gallo, A. Ghavasieh, F. Battiston, and M. D. Domenico, Reducibility of higher-order net- works via information flows (2025), arXiv:2404.08547 [physics.soc-ph]

    arXiv 2025

  42. [42]

    Latané, The psychology of social impact., American psychologist36, 343 (1981)

    B. Latané, The psychology of social impact., American psychologist36, 343 (1981)

    1981

  43. [43]

    Kim, D.-S

    J. Kim, D.-S. Lee, B. Min, M. A. Porter, M. San Miguel, and K. I. Goh, Competition between group interac- tions and nonlinearity in voter dynamics on hypergraphs, 16 Phys. Rev. E111, L052301 (2025)

    2025

  44. [44]

    A model for spatial conflict, Biometrika60, 581 (1973)

    1973

  45. [45]

    R. A. Holley and T. M. Liggett, Ergodic theorems for weakly interacting infinite systems and the voter model, The Annals of Probability3, 643 (1975)

    1975

  46. [46]

    T. M. Liggett,Interacting Particle Systems(Springer, New York, 1985)

    1985

  47. [47]

    T. M. Liggett,Stochastic Interacting Systems: Con- tact, Voter and Exclusion Processes(Springer,NewYork, 1999)

    1999

  48. [48]

    Suchecki, V

    K. Suchecki, V. M. Eguíluz, and M. San Miguel, Con- servation laws for the voter model in complex networks, Europhysics Letters69, 228 (2004)

    2004

  49. [49]

    Suchecki, V

    K. Suchecki, V. M. Eguíluz, and M. San Miguel, Voter model dynamics in complex networks: Role of dimen- sionality, disorder, and degree distribution, Phys. Rev. E 72, 036132 (2005)

    2005

  50. [50]

    Sood and S

    V. Sood and S. Redner, Voter model on heterogeneous graphs, Phys. Rev. Lett.94, 178701 (2005)

    2005

  51. [51]

    Vazquez and V

    F. Vazquez and V. M. Eguíluz, Analytical solution of the voter model on uncorrelated networks, New Journal of Physics10, 063011 (2008)

    2008

  52. [52]

    Castellano, S

    C. Castellano, S. Fortunato, and V. Loreto, Statistical physics of social dynamics, Reviews of Modern Physics 81, 591 (2009)

    2009

  53. [53]

    Erdős and A

    P. Erdős and A. Rényi, On random graphs I, Publ. Math. Debrecen6, 18 (1959)

    1959

  54. [54]

    E. N. Gilbert, Random graphs, The Annals of Mathe- matical Statistics30, 1141 (1959)

    1959

  55. [55]

    Erdős and A

    P. Erdős and A. Rényi, On the evolution of random graphs, Bull. Inst. Internat. Statist38, 343 (1960)

    1960

  56. [56]

    Bollobas and P

    B. Bollobas and P. Erdős, Cliques in random graphs, Mathematical Proceedings of the Cambridge Philosoph- ical Society80, 419–427 (1976)

    1976

  57. [57]

    Frieze and M

    A. Frieze and M. Karoński,Introduction to random graphs(Cambridge University Press, 2015)

    2015

  58. [58]

    B.Bollobás,Random Graphs,2nded.,CambridgeStudies in Advanced Mathematics (Cambridge University Press, 2001)

    2001

  59. [59]

    López, Weighted projected networks: Mapping hyper- graphs to networks, Phys

    E. López, Weighted projected networks: Mapping hyper- graphs to networks, Phys. Rev. E87, 052813 (2013)

    2013

  60. [60]

    Carletti, F

    T. Carletti, F. Battiston, G. Cencetti, and D. Fanelli, Random walks on hypergraphs, Phys. Rev. E101, 022308 (2020)

    2020

  61. [61]

    Slanina and H

    F. Slanina and H. Lavička, Analytical results for the sz- najd model of opinion formation, Eur. Phys. J. B35, 279 (2003)

    2003

  62. [62]

    Pugliese and C

    E. Pugliese and C. Castellano, Heterogeneous pair ap- proximation for voter models on networks, Europhysics Letters88, 58004 (2009)

    2009

  63. [63]

    J. P. Gleeson, S. Melnik, J. A. Ward, M. A. Porter, and P. J. Mucha, Accuracy of mean-field theory for dynamics on real-world networks, Phys. Rev. E85, 026106 (2012)

    2012

  64. [64]

    Min and M

    B. Min and M. San Miguel, Fragmentation transitions in a coevolving nonlinear voter model, Scientific Reports7, 12864 (2017)

    2017

  65. [65]

    Jędrzejewski, Pair approximation for theq-voter model with independence on complex networks, Phys

    A. Jędrzejewski, Pair approximation for theq-voter model with independence on complex networks, Phys. Rev. E95, 012307 (2017)

    2017

  66. [66]

    Castellano, D

    C. Castellano, D. Vilone, and A. Vespignani, Incomplete ordering of the voter model on small-world networks, Eu- rophysics Letters63, 153 (2003)

    2003

  67. [67]

    R. M. Corless, G. H. Gonnet, D. E. G. Hare, D. J. Jeffrey, and D. E. Knuth, On the Lambert function, Advances in Computational Mathematics5, 329 (1996)

    1996

  68. [68]

    López, The distribution of the number of node neigh- borsinrandomhypergraphs,JournalofPhysicsA:Math- ematical and Theoretical46, 305003 (2013)

    E. López, The distribution of the number of node neigh- borsinrandomhypergraphs,JournalofPhysicsA:Math- ematical and Theoretical46, 305003 (2013)

    2013

  69. [69]

    Abramowitz and I

    M. Abramowitz and I. A. Stegun, eds.,Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables(Dover, New York, 1972). 17 Appendix A: Generation of randomd-hypergraphs

    1972

  70. [70]

    GivenNnodes and a target average hyperdegreeµd, we draw a total ofµdN/(d+ 1)distinctd-hyperedges uniformly at random from the set of all possibled-hyperedges

    Erdős–Rényid-hypergraphs We consider the generalization of the Erdős–Rényi (ER) graphs [40–42] tod-hypergraphs [44]. GivenNnodes and a target average hyperdegreeµd, we draw a total ofµdN/(d+ 1)distinctd-hyperedges uniformly at random from the set of all possibled-hyperedges. This algorithm is used for most values ofµd. However, for small values ofµd it do...

  71. [71]

    Randomly selectd+ 1nodes with at least one remaining stub

  72. [72]

    Accepthonly if alld+ 1nodes are distinct andhhas not appeared before

    Form a candidate hyperedgehfrom these nodes. Accepthonly if alld+ 1nodes are distinct andhhas not appeared before

  73. [73]

    Ifhis accepted, we decrease each involved node’s stub count by one and store the new hyperedge

  74. [74]

    Ifhis rejected, we simply discard it and continue sampling. We monitor the number of iterations without progress and impose a maximum stall threshold in order to avoid getting trapped in pathological configurations where the remaining stubs cannot be matched without creating compatible or repeated hyperedges. If this threshold is exceeded, the entire cons...

  75. [75]

    Note that this quantity is symmetricR ij =R ji

    Erdős–Rényid-hypergraphs To quantify deviations from the sparse limit, we introduce themultiplicityRij as the number ofd-hyperedges that simultaneously contain nodesiandj. Note that this quantity is symmetricR ij =R ji. Given that there areM (2) d = N−2 d−1 possible hyperedges containing bothi, j, and each is created independently with probabilityp=µ d/ N...